Selective inhibition of vagal afferent nerve pathways regulating cough using Na_v 1.7 shRNA silencing in guinea pig nodose ganglia

Abstract

Adeno-associated virus delivery systems and short hairpin RNA (shRNA) were used to selectively silence the voltage-gated sodium channel Na_V 1.7 in the nodose ganglia of guinea pigs. The cough reflex in these animals was subsequently assessed. Na_V 1.7 shRNA was delivered to the majority of nodose ganglia neurons [50–60% transfection rate determined by green fluorescent protein (GFP) gene cotransfection] and action potential conduction in the nodose vagal nerve fibers, as evaluated using an extracellular recording technique, was markedly and significantly reduced. By contrast, <5% of neurons in the jugular vagal ganglia neurons were transfected, and action potential conduction in the jugular vagal nerve fibers was unchanged. The control virus (with GFP expression) was without effect on action potential discharge and conduction in either ganglia. In vivo, Na_V 1.7 silencing in the nodose ganglia nearly abolished cough evoked by mechanically probing the tracheal mucosa in anesthetized guinea pigs. Stimuli such as capsaicin and bradykinin that are known to stimulate both nodose and jugular C-fibers evoked coughing in conscious animals was unaffected by Na_V 1.7 silencing in the nodose ganglia. Nodose C-fiber selective stimuli including adenosine, 2-methyl-5-HT, and ATP all failed to evoke coughing upon aerosol challenge. These results indicate that cough is independently regulated by two vagal afferent nerve subtypes in guinea pigs, with nodose Aδ fibers regulating cough evoked mechanically from the trachea and bradykinin- and capsaicin-evoked cough regulated by C-fibers arising from the jugular ganglia.

studies in guinea pigs indicate that at least two vagal afferent nerve subtypes can induce coughing upon activation (7, 9). In addition to C-fibers, which conduct action potentials at <2 m/s and are activated by bradykinin and agonists of transient receptor potential vanilloid-1 (TRPV1) and transient receptor potential ankyrin-1 (TRPA1) ion channels, cough is also regulated by neurons that conduct action potentials in the Aδ range (∼5 m/s). These small, myelinated afferent nerves are exquisitely sensitive to punctuate mechanical stimulation and rapid decreases in mucosal pH (15, 23) but are insensitive to activators of TRPV1 or TRPA1.

The Aδ vagal afferents that regulate cough arise from the nodose ganglia and terminate in the subepithelial extracellular matrix of the extrapulmonary airways (21). C-fibers innervating the guinea pig respiratory tract project from both the nodose and jugular ganglia. The unmyelinated afferents derived from these embryologically distinct ganglia have unique activation profiles and sites of termination in the airways and lungs and thus represent physiologically distinct subtypes (17, 26). Both C-fiber subtypes can be stimulated by agents commonly used to evoke cough in conscious guinea pigs and humans including bradykinin, capsaicin, citric acid, and allyl isothiocyanate (AITC) (19).

While it is generally accepted that C-fibers regulate cough, which C-fiber subtype(s) initiate coughing upon activation is unknown. As noted above, the C-fiber-selective stimuli used to evoke coughing activate both nodose and jugular C-fibers. Jugular C-fibers terminating in the central airways are thought to initiate cough upon activation, but agents that stimulate jugular C-fibers and not nodose C-fibers have not been identified. Conceivably, delivering bradykinin, capsaicin, or TRPA1 activators only to the extrapulmonary airways could be used to selectively stimulate jugular C-fibers, where few, if any, nodose C-fibers terminate. But such experiments are only possible in anesthetized animals, and unfortunately, coughing evoked by C-fiber-selective stimuli (e.g., capsaicin, bradykinin) is prevented by general anesthesia (9). In fact, under certain experimental conditions bronchopulmonary C-fiber activation can be acutely inhibitory of cough in anesthetized animals (24, 25). These and other observations have led some to question the role of C-fibers in cough or propose indirect mechanisms by which they might initiate coughing (27).

Determining which bronchopulmonary vagal afferent nerve subtypes regulate cough requires interventions that selectively stimulate or inhibit the activation of the subtypes implicated in cough. For example, our previous studies indicate that nodose but not jugular C-fibers are activated by adenosine A1 or A2 receptor agonists, 5-HT3 receptor agonists, and ATP/purinergic P2X2/3 receptor agonists (10, 11, 26). As an alternative approach we have used short hairpin RNA (shRNA) technology to modify gene expression in subsets of vagal sensory ganglia neurons (14). We recently described an in vivo gene silencing approach targeting the expression of voltage-gated sodium channels (22). Action potential conductance in all vagal nerves is dependent on tetrodotoxin (TTX)-sensitive sodium channels. Among the nine voltage-gated sodium channel (Na_V) channel subunits, Na_V 1.7 is the predominant TTX-sensitive subunit expressed in guinea pig nodose and jugular vagal afferent neurons (16, 22). Selectively silencing Na_V 1.7 using shRNA greatly decreased the excitability of all vagal afferent nerve subtypes as assessed in vitro, whereas silencing Na_V 1.7 in both the nodose and jugular ganglia essentially abolished the cough reflex measured in vivo. We therefore hypothesized that selectively silencing Na_V 1.7 only in vagal afferent nerves arising from the nodose ganglia could be used to determine which afferent nerve subtypes regulate cough. We describe the cough reflex in conscious and anesthetized animals following selective silencing of Na_V 1.7 in the nodose ganglia. We also describe the relative tussive effects of C-fiber-selective stimuli in conscious guinea pigs.

MATERIALS AND METHODS

AAV8-Based shRNA Injection In Vivo

The design of enhanced green fluorescent protein (eGFP) encoding shRNA targeting the guinea pig Na_V 1.7, subcloning the shRNA into the adeno-associated virus (AAV8)-based vectors, and the evaluation of the AAV-based shRNA were shown previously (22). Male Hartley guinea pigs weighing 190–250 g (Hilltop Laboratory Animals, Scottsdale, PA) were used for the AAV-8-based shRNA (Gene Therapy Program, University of Pennsylvania) injection and the method has been previously described (14). In brief, animals were anesthetized with a mixture of ketamine (50 mg/kg)-xylazine (10 mg/kg), and an incision was made over a superficial portion of the masseter muscle area. Micropipettes (∼20 μm tip diameter) were filled with solution of virus using capillary force [1–1.5 μl, (1–3) × 10¹² PFU] and assembled to a 1-ml syringe via plastic tubing. The tip of the micropipette was gently inserted and virus was then slowly injected into the nodose ganglia by depressing the plunger. Control animals received the virus encoding eGFP-control (scrambled) shRNA on both nodose ganglia, while the other animal group received the one encoding eGFP-Na_V1.7 shRNA. After recovery from surgery the animals were cared for in the vivarium of the Johns Hopkins Asthma and Allergy Center. All procedures are approved by the Johns Hopkins Animal Care and Use Committee.

Cough Experiments on Conscious Animals

Capsaicin challenge on virus-injected animals.

Four weeks after the virus injection, coughing experiments in conscious guinea pigs were performed as described previously (9, 22). Animals were placed in a chamber continuously filled with room air using an air pump, and their respiratory activity was monitored by sound and pressure changes within the chamber. Capsaicin (1, 3, and 10 μM; Sigma, St. Louis, MO) dissolved in saline was delivered to the chamber by an ultrasonic nebulizer (generates ∼5 μm particle size) via plastic tubing for 5 min at each dose, followed by an additional 5 min with the nebulizer turned off. Respiratory activity was recorded using an EMKA data acquisition system (Falls Church, VA). Visual monitoring and the sound and pressure recordings were used to quantify coughing.

Selective agonist challenges in naïve animals.

Male Hartley guinea pigs weighing 180–250 g were used. The animals were placed in a chamber as described above and then the selective nodose C-fiber stimulants adenosine (10 mM), 2-methyl-5-HT (5 mM), or ATP (10 μM; Sigma) were delivered for 5 min with nebulizer on followed by another 5 min with the nebulizer turned off. Immediately after challenges with the nodose selective stimulants, the animals were subsequently challenged with capsaicin (3 μM) or AITC (10 mM; Sigma), C-fiber stimulants that activate both nodose and jugular C-fibers. In separate animals we also studied the effects of 1 mg/ml bradykinin aerosols on cough. The aerosol concentrations of each agent used were chosen based on published evidence for their effects on cough and were typically 30–1,000× that required to evoke C-fiber discharge in vitro. For the control saline experiments, animals were rested for 10 days then the cough experiment was performed with saline.

Mechanically Induced Cough in Anesthetized Animals

Ten days after the capsaicin challenge, the virus-injected guinea pigs were anesthetized with urethane (1.5 g/kg ip). An approximately 6-cm incision was made over a shaved area over the extrathoracic trachea and larynx. The skin was retracted from the exposed muscle area using clamps. The upper to middle trachea was exposed by blunt dissection using blunt-tipped instruments. Sterile cotton pads and weak vacuum suction were used to control bleeding throughout the surgery. A midline incision of the trachea was performed (∼4 cm), and the tracheal mucosa was mechanically probed with blunt-end forceps. Any cough that occurred during the mechanical manipulation of the trachea was counted.

Vagus Nerve Extracellular Recording

After the mechanical cough experiments, the animals were euthanized with CO₂, and the vagus nerve with nodose and jugular ganglia were dissected out with continuous wash with Krebs solution (in mM: 118 NaCl, 5.4 KCl, 1 NaHPO₄, 1.2 MgSO₄, 1.9 CaCl₂, 25 NaHCO₃, and 11.1 dextrose, pH 7.4). GFP expression levels in the vagus were checked under an upright microscope illuminated with mercury arc to confirm the virus transduction. Adhering connective tissue of the nodose ganglia was first pinned to the bottom of a recording chamber that contained silicone elastomer. The other end of the vagus was suctioned tightly into a Krebs solution containing stimulus electrode. The chamber was perfused with Krebs solution (37 ± 2°C), and the recording electrodes were filled with 3 mM KCl. The vagus nerve was stimulated with a supramaximal square pulse (60 V, 0.8 ms pulse duration), and compound action potentials were measured at 9 to 13 positions within the nodose ganglia. The positions were approximately equally spaced to provide an unbiased sampling of the whole ganglion, as we described previously (22). At each position, the electrode as stepped down along the Y-axis into the ganglion only the maximum obtainable response was recorded. The same recordings were performed on jugular ganglia. The measured action potentials were amplified (Microelectrode AC amplifier 1800; A-M Systems, Everett, WA), filtered (0.3 kHz of low cut-off and 1 kHz of high cut-off), monitored on an oscilloscope (TDS340; Tektronix, Beaverton, OR), recorded and analyzed by NerveOflt software (Phocic, Baltimore, MD). The glass electrode has a resistance about 1.5 MΩ. The multiunit compound potentials are a reflection of action potentials arriving from individual nerve fibers being conducted to cell bodies in the sensory ganglia. This technique records almost exlusively somal spikes; when the electrode is moved away from the cell bodies and inserted into the vagus nerve per se, virtually no activity is recorded. This allows for selective assessment of nodose or jugular sensory activity without contamination by spikes in preganglionic parasympathetic fibers. To get some quantitative estimate of the number of action potentials being conducted, we evaluated the area under the curve of the compound potential from our digital recordings. These data are represented in arbitrary units of voltage × time. An individual action potential will provide a large voltage signature in neurons situated very close to the electrode, whereas a smaller signature of the cell body is further from the electrode. A 50% reduction in the area under the voltage-time curve does not mean precisely 50% fewer action potentials. The activity often arises in two generalized waves, the initial wave reflecting action potentials in A-fibers (∼15–3 m/s), and the second wave reflecting the C-fibers (∼1.5–0.5 m/s). Most of the time the vagus was too short to study the A-fibers, because they were contaminated with an indefinable component of the shock artifact, therefore, the A-wave component was not included in the analysis. The problem with using a longer segment of vagus to further separate the A-wave from the artifact is that fewer and fewer A-fibers are present in the vagus as you move caudally (e.g., nearly all subdiaphragmatic nerve fibers are unmyelinated). In the present study we noted A-fiber activity uncontaminated by the artifact in 15% of the control ganglia and 9% of the treated ganglia.

Cell Dissociation and GFP Expression Check

After the extracellular experiments, the nodose and jugular ganglia were separately cleared of connective tissue with continuous wash in Krebs solution. Each ganglion obtained from an animal (two nodose and two jugular ganglia) was separately dissociated. Ganglia were then incubated in the solution containing dispase and collagenase (2 mg/ml each) dissolved in Hanks' balance salt solution without calcium or magnesium (Sigma) at 37°C. Cells were washed three times with L-15 medium (Invitrogen) supplemented with 10% fetal bovine serum (Invitrogen) and then plated on laminin-treated glass coverslips. Within 12 h, the coverslips were mounted on a chamber continuously perfused with Krebs or PBS solutions, and the GFP-positive and the total number of cells were counted under an upright microscope illuminated with mercury arc using a ×20 objective.

Statistics

All data are means ± SE. Statistical analysis was assessed with Prism software (GraphPad). Each analysis method is noted in the figure legends. Statistical significance of differences was accepted at P < 0.05.

RESULTS

Na_V 1.7 shRNA Virus Injection in Nodose Ganglia

We have previously shown that injecting a saturating amount (2–3 μl) of AAV encoding Na_V 1.7 shRNA and eGFP gene directly into guinea pig nodose ganglia in vivo led to GFP expression in ∼80% of the nodose ganglia neurons and in many jugular ganglia neurons (22). The transfected neurons, identified for analysis based on visual evidence of eGFP expression, had a 90% reduction in Na_V 1.7 mRNA, with no change in mRNA levels for other Na_V subunits. This effect on Na_V 1.7 mRNA levels was associated with a large reduction in TTX-sensitive sodium currents, reduced action potential conduction, and a decrease in the peak frequency at which action potentials could be elicited in these neurons (22).

To differentiate the roles of neurons arising from the nodose and jugular ganglia in cough, the same AAV shRNA method described above was employed here, except a smaller amount of Na_V 1.7 shRNA-containing virus (0.8–1 μl) was injected into the nodose ganglia, aiming to selectively transfect nodose neurons, while sparing jugular neurons. Based on GFP fluorescence, this treatment led to the transfection of a majority of nodose neurons, while <5% of jugular ganglia neurons were transfected (Fig. 1, A and B).

Fig. 1.

A: tissue section of the nodose ganglion isolated from a guinea pig previously treated with adeno-associated virus (AAV)-enhanced green fluorescence protein (eGFP) showing the successful tansfection of nodose neurons (top). The actual percentage of GFP-containing neurons was assessed after the neurons were enzymatically dissociated (e.g., bottom of A) B: histograms of the percentage of GFP-positive neurons in the dissociated nodose and jugular ganglia. The ganglia were dissected from guinea pigs in which the nodose ganglia was microinjected with AAV-eGFP (control) or with AAV-eGFP-Na_V 1.7 short hairpin RNA (shRNA). The data represent the means ± SE of ganglia (n) evaluated. Statistical comparisons were made using unpaired t-tests between nodose and jugular ganglia in each control and AAV-eGFP-treated groups. *Significant difference (P < 0.01) between the values in the nodose versus jugular ganglia. In two animals, fewer than 20% of the nodose ganglion neurons were expressing visible eGFP; these animals were excluded from subsequent functional studies.

To evaluate the effects of Na_V 1.7 silencing on nodose and jugular afferent nerve activity, we used extracellular recordings to monitor compound action potentials (22). This technique measures the amount of action potential activity evoked and conducted to the cell bodies in the sensory ganglia. A recording electrode was placed in at least nine positions in each nodose, and jugular ganglion and mulitunit action potentials were recorded (see examples in Fig. 2). Electrode placements targeted neuronal cell bodies in the ganglia, thus excluding vagal axonal fibers passing adjacent to the ganglia from the extracellular recordings. Results were expressed as an area in voltage × time units (v.t; see materials and methods). In keeping with the limited transfection of jugular ganglia neurons using our modified AAV microinjection strategy (<5%) (Fig. 1), injecting Na_V 1.7 shRNA into the nodose ganglia was without effect on the compound potentials recorded in the jugular ganglia, which averaged 32.4 ± 4.8 v.t (n = 42) and 28.7 ± 3.4 (n = 54) in control and shRNA-transfected animals, respectively (P > 0.1). By contrast, compound action potential area in the nodose ganglia was reduced by 35% following Na_V 1.7 shRNA transfection in the ganglia (P < 0.01). We then separated the data into those recording positions where limited action potential activity (<20 v.t), a moderate amount (20–60 v.t), and where robust discharge (>60 v.t) was recorded (Fig. 2). In the Na_V 1.7 shRNA-treated ganglia, over 25% of the positions recorded displayed little, if any, activity, while this occurred in only ∼5% of recordings made in the control ganglia (n = 158 recordings). Conversely, more than twice as many positions in the control nodose ganglia displayed robust action potential discharge compared with the Na_V 1.7 shRNA-treated ganglia. These data and our previous study indicate that Na_V 1.7 is important for the initiation and/or conduction of action potentials in vagal sensory nerve fibers. The data also indicate that the protocol used in the present study had selectively targeted and silenced neurons in the nodose ganglia.

Fig. 2.

A: histogram graph displaying proportion of recorded sites with an action potential compound area recorded from nodose ganglia corresponding to the three levels of activity [Control: recorded points = 158, nodose = 14, animal = 7, voltage-gated sodium channel (Na_V) 1.7 shRNA: recorded points = 184, nodose = 16, animal = 8]. The action potential activities are categorized into three groups based on calculating the action potential compound area. Those with little or no activity included all position of <20 voltage × time (v.t) units, those with moderate activity (20–60 v.t), and those positions with robust activity (>60 v.t) Top, representative action potential compound traces for each group, with the corresponding conduction velocities. B: compound action potential area recorded from all positions in nodose ganglia treated with control or Na_V 1.7 shRNA. The bars represent the means ± SE of 158 and 184 determinations from control and Na_V 1.7 shRNA-treated ganglia, respectively. *Significant difference (P < 0.0001) between the values.

Mechanical Force-Induced Cough in Anesthetized Animals

After the animals were anesthetized with urethane, we evaluated cough evoked by mechanical stimulation of the airways in control and Na_V 1.7 shRNA-treated guinea pigs. As we have previously described, this cough is thought to occur as a result of stimulating nodose Aδ fibers that terminate in the tracheal and bronchial submucosa (9). In control animals (nodose ganglia injected with AAV not encoding Na_V 1.7 shRNA), mechanical stimulation of the tracheal mucosa consistently evoked coughs, with totals ranging from 5 to 23 coughs/animal and averaging 11 ± 3 coughs overall (n = 7; Fig. 3). By contrast, the same mechanical stimuli evoked few, if any, coughs in animals receiving Na_V 1.7 shRNA injections of the nodose ganglia (2 ± 1 coughs; n = 8; Fig. 3). This is consistent with our hypothesis that cough induced by mechanically stimulating the airways mucosa in anesthetized guinea pig occurs secondary to action potential discharge and conduction in nodose ganglia neurons.

Fig. 3.

Average number of coughs evoked mechanically in an anesthetized guinea pig previously treated with control (open bar) or Na_V 1.7 shRNA (hatched bar). The upper to middle trachea was exposed by blunt dissection using blunt-tipped instruments. Sterile cotton pads and weak vacuum suction were used to control bleeding throughout the surgery. A midline incision of the trachea was performed (∼4 cm) and the tracheal mucosa was mechanically probed with blunt-end forceps repeatedly. Any cough that occurred during the approximate 10 min of mechanical manipulation of the trachea was counted. Statistical comparisons were made using unpaired t-tests between control and Na_V 1.7 shRNA treated groups. *Significant difference (P < 0.01) between the values.

Capsaicin-Mediated Cough in Conscious Animals

Capsaicin was equally effective at evoking cough in conscious guinea pigs in which the nodose ganglia were either treated with AAV carrying Na_V 1.7 shRNA or the control AAV (Fig. 4, A and B). This suggests that capsaicin-evoked cough in conscious guinea pigs results from jugular C-fiber activation and occurs independent of the Aδ and C-fibers arising from the nodose ganglia.

Fig. 4.

A: representative traces of coughs initiated by introducing capsaicin in an experimental chamber. B: coughs evoked by introducing capsaicin (1, 3, and 10 μM) were counted from control shRNA and Na_V 1.7 shRNA-treated animals. For each dose, capsaicin was applied by a nebulizer for 5 min and the total number of coughs was counted over a 10-min period. The data represent the means ± SE of n animals. Statistical comparisons were made using unpaired t-tests between control and Na_V 1.7 shRNA-treated groups and those values are not significantly different.

To further address the hypothesis that C-fibers arising from the jugular and not the nodose ganglia are largely responsible for chemical irritant-induced coughing observed in conscious guinea pigs, naïve conscious animals were challenged with a variety of nodose selective and nonselective C-fiber stimulants. Capsaicin and AITC, agents that indiscriminately stimulate both jugular and nodose C-fibers, consistently evoked coughing in these animals (Fig. 5). Bradykinin, which also activates both nodose and jugular C-fibers, also evoked coughing (18 ± 4 coughs evoked by 1 mg/ml aerosols of bradykinin; n = 8, data not shown). By contrast, ATP, adenosine, and 2-methyl-5-HT, which selectively activate nodose C-fibers innervating the lungs (10, 16), were no different from vehicle in their ability to evoke coughing (Fig. 5).

Fig. 5.

Coughs evoked by nebulizing vehicle (saline; n = 8), adenosine (10 mM; n = 6), 2-methyl-5-HT (5 mM; n = 5, ATP; 10 μM; n = 5), capsaicin (3 μM; n = 8), and allyl isothiocyanate (AITC, 10 mM; n = 5) were counted from naive animals. The animals were treated with the drug for 5 min, and the total number of coughs was recorded over a 10-min period. The data shows the means ± SE. Statistical comparisons were made using one-way ANOVA followed by Bonferroni's Multiple Comparison Test. *Significant difference (P < 0.01) from vehicle, adenosine, 2-methyl-5-HT, and ATP.

DISCUSSION

Vagal afferent nerves innervating the respiratory tract are derived from neurons situated in the nodose and jugular ganglia. The neurons in these two disparate ganglia have different embryologic origins, distinct developmental and neurochemical features, and unique functional phenotypes (2, 3, 19). The results presented here provide further evidence that readily distinguishable bronchopulmonary vagal afferent nerve subtypes arising from both the nodose and jugular ganglia independently regulate the cough reflex.

To better define the role of vagal afferent nerve subtypes in cough we have identified interventions that selectively target the various subtypes, either by inhibiting their activation or activating them independent of the other afferent nerves innervating the airways and lungs. In all vagal afferent neurons, action potential conduction is absolutely dependent on TTX-sensitive Na⁺ channels. We previously reported that Na_V 1.7 is quantitatively the major, though not only, TTX-sensitive Na⁺ channel subunit expressed in both nodose and jugular ganglia neurons (16, 22). Transfecting shRNA targeting Na_V 1.7 in vagal sensory neurons specifically reduced Na_V 1.7 mRNA expression by >90% and substantially decreased the TTX-sensitive sodium current in these same neurons. Under conditions where Na_V 1.7 was silenced in both the nodose and jugular ganglia, capsaicin-evoked cough in conscious guinea pigs was markedly inhibited.

Our previous studies revealed that while general anesthesia prevents coughing evoked by C-fiber-selective stimuli including capsaicin, cough can still be reliably evoked in anesthetized guinea pigs by mechanically stimulating the tracheal mucosa or by acid applied topically to the tracheal mucosa. These cough responses likely result from the activation of capsaicin-insensitive Aδ fibers arising from the nodose ganglia (9, 21). Consistent with our previous studies implicating nodose Aδ fibers in cough in anesthetized animals, selectively silencing Na_V 1.7 in the nodose ganglia nearly abolished mechanically evoked cough in anesthetized guinea pigs.

Based on visualized expression of eGFP (which was included in our AAV construct), nearly 60% of the nodose neurons were transfected using the microinjection strategy employed in the present study. This may underestimate the true value, as we previously found that when detection of eGFP is enhanced by immunohistochemistry, substantially more transfected neurons are apparent (22). By contrast, <5% of the jugular ganglia neurons displayed visible evidence of eGFP expression. Compound potential recordings indicated that action potential conduction was unaltered in the jugular ganglia, confirming that transfection of jugular neurons was limited or absent. Consistent with our previous report, however, there was a large reduction in the ability of the nodose afferents to conduct action potentials evoked by supramaximal current pulses (22). We have thus selectively silenced Na_V 1.7 in the nodose ganglia.

The Na_V 1.7 shRNA injection into the nodose ganglia reduced but did not abolish the nodose compound potential evoked by a single supramaximal stimulating impulse. The residual activity is likely attributable to those nodose neurons that were not transfected and to the transfected neurons where there was sufficient TTX-sensitive current remaining (perhaps attributable to other Na_V channel subunits) to support conduction. Despite this residual activity, cough was still nearly abolished in the anesthetized animals receiving Na_V 1.7 shRNA. This is not a surprising result, inasmuch as our previous work indicates that cough in anesthetized guinea pigs is very sensitive to the number of units activated in response to a stimulus and the peak action potential frequency in these active units (8). Functionally reducing the number of vagal afferents activated by ∼50% (through unilateral vagotomy) or reducing the peak action potential frequencies (evoked electrically) to <10 Hz markedly inhibited or abolished coughing in most animals. In our previous study we noted that silencing Na_V 1.7 shRNA not only increased action potential threshold but also decreased the frequency at which action potentials could be evoked. The response to a prolonged suprathreshold current step peaked at <10 Hz following Na_V 1.7 silencing but reached 30–40 Hz in control neurons (22). The combined effect of silencing entirely many Aδ fibers and diminishing peak action potential frequencies in many more of these fibers were obviously sufficient to prevent cough.

Bronchopulmonary C-fiber subtypes arise from the nodose and jugular ganglia and can be differentiated in part by their ganglionic origin but also by their sites of termination in the airways and lungs. Jugular C-fibers terminate throughout the extrapulmonary and intrapulmonary airways while nodose C-fibers sparsely innervate the extrapulmonary airways (26). Stimuli that activate both nodose and jugular ganglia C-fibers in guinea pigs readily evoke coughing when delivered as aerosols to conscious guinea pigs (1, 4, 6, 9, 12, 18). Which of these C-fiber subtypes drive the cough reflex in conscious guinea pigs has not previously been determined. AAV transfection indiscriminately transfects both large and small neurons (14) so we speculated that selectively silencing Na_V 1.7 in the nodose ganglia could be used as a tool to evaluate the relative role of nodose and jugular C-fibers in cough. Indeed, our compound action potential analyses revealed that conduction in C-fibers (conduction velocities of 0.4–1.5 m/s) was significantly inhibited. Despite Na_V 1.7 silencing in the majority of nodose ganglia neurons and the marked inhibition of Aδ fiber-dependent cough in anesthetized guinea pigs, capsaicin-induced cough in conscious guinea pigs was unaffected.

The profound effects of Na_V 1.7 silencing in the nodose ganglia on mechanically evoked cough in anesthetized guinea pigs, but its lack of effect on capsaicin-evoked coughing in conscious guinea pigs is consistent with the hypothesis that stimulation of jugular C-fibers, more than nodose C-fibers, accounts for the coughing evoked by capsaicin and other C-fiber selective stimuli in conscious guinea pigs, while capsaicin-insensitive nodose Aδ fibers regulate mechanically induced cough in anesthetized guinea pigs. An obvious prediction of this hypothesis is that selectively silencing jugular C-fibers would not inhibit mechanically evoked cough in anesthetized animals but would block the capsaicin-induced cough in conscious animals. We have been unable to reliably microinject AAV selectively into the jugular ganglia, because these ganglia are situated in the jugular foramen and are poorly accessible even with surgery. But the inability of nodose C-fiber-selective stimuli (ATP, adenosine, 2-methyl-5-HT) to evoke coughing in conscious animals that reliably cough in response to C-fiber stimuli that activate both jugular and nodose C-fibers (capsaicin, AITC) further implicates jugular C-fibers in cough and argues against an excitatory role for nodose C-fibers in cough. We speculate that nodose C-fibers may regulate the sensation of dyspnea (7). Nodose C-fibers may also be responsible for the inhibition of cough observed in anesthetized animals (24, 25) (B. J. Canning and Y. -L.Chou, unpublished observations).

It has been suggested that C-fiber activation evokes cough indirectly, secondary to autonomic reflexes or axonal reflexes resulting from peripheral neurokinin release (27). The evidence in favor of this assertion is that exogenously administered substance P can activate intrapulmonary rapidly adapting receptors and that stimuli that activate C-fibers may also activate RARs through tachykinin release (5, 13). Substance P aerosols also evoked cough in some but not other studies (27). A prediction of this hypothesis is that selectively inhibiting Aδ activation would inhibit C-fiber-evoked coughing. This was not apparent in the present study, however, as Na_V 1.7 silencing prevented Aδ-dependent coughing in anesthetized guinea pigs but was without effect on C-fiber-dependent coughing evoked in conscious guinea pigs. As we have argued extensively elsewhere (7), these and other data further differentiate the Aδ fibers regulating cough from the intrapulmonary RARs that do not initiate coughing upon activation.

Perspectives and Significance

The data provide additional independent support for the hypothesis that two afferent nerve subtypes can upon activation lead to coughing in guinea pigs. One type comprises neural crest-derived jugular C-fibers and the other placodally derived nodose Aδ fibers. It is likely that evolutionarily these two disparate pathways have been selected to serve different purposes. This raises the possibility that one type of cough-evoking fiber may be more involved in certain pathological coughs than the other. It also reveals the potential flexibility in developing anti-tussive agents that are selective in their inhibition of the two subtypes of cough-evoking afferent nerves. The data should not be taken as evidence that these pathways are entirely parallel and noninteracting. In fact, evidence supports the view that these pathways may converge in the central nervous system such that activation of C-fibers “central sensitizes” the A-fiber cough pathway. After vagal C-fiber activation, cough can be evoked in anesthetized animals by stimulation of the trachea nodose Aδ-fibers with intensities normally below the cough threshold (20). The data present indicates, however, that this type of central interaction is not a necessary requirement for C-fiber-evoked cough.

GRANTS

This work was supported in part by funding from the National Institutes of Health, Bethesda, MD, and from GlaxoSmithKline Pharmaceuticals.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

Author contributions: Y.M., M.J.C., B.J.U., and B.J.C. conception and design of research; Y.M., F.R., Y.-L.C., and B.J.C. performed experiments; Y.M. and B.J.C. analyzed data; Y.M., M.J.C., B.J.U., and B.J.C. interpreted results of experiments; Y.M. prepared figures; Y.M. and B.J.U. drafted manuscript; Y.M., M.J.C., B.J.U., and B.J.C. edited and revised manuscript; Y.M., M.J.C., B.J.U., and B.J.C. approved final version of manuscript.