Molecular/Ionic Basis of Vagal Bronchopulmonary C-Fiber Activation by Inflammatory Mediators

Abstract

Stimulation of bronchopulmonary vagal afferent C fibers by inflammatory mediators can lead to coughing, chest tightness, and changes in breathing pattern, as well as reflex bronchoconstriction and secretions. These responses serve a defensive function in healthy lungs but likely contribute to many of the signs and symptoms of inflammatory airway diseases. A better understanding of the mechanisms underlying the activation of bronchopulmonary C-fiber terminals may lead to novel therapeutics that would work in an additive or synergic manner with existing anti-inflammatory strategies.

Introduction

In his classic treatise “The Integrative Action of the Nervous System” (77), Sherrington discusses a subset of sensory nerves that appear to be “…adapted to a whole group of excitants, a group of excitants which has in relations to the organism one feature common to all its components, namely a nocuous character.” He termed these sensory nerves “noci-ceptive” nerves, now referred to as “nociceptors,” and reasoned that they “attach to the skin a specific sense of its own injuries” (77).

By informing the organism of potential injury and causing nocifensive sensations and reflexes, nociceptors provide an important host defense function. As part of its vital function in gas exchange, the human respiratory tract samples over 6,000 liters of air each day. The respiratory tract is at the mercy of the environment with respect to how clean such inhaled air is. The air we breathe ranges on a spectrum from being relatively pure to being filled with particulate matters, irritants, and potentially serious microorganisms and toxicants. It is therefore no wonder we find the airways richly innervated with the same type of sensory nociceptors first described in the skin by Sherrington. As in the somatosensory system, the respiratory nociceptors are mainly unmyelinated nerves that conduct action potentials very slowly (~1 m/s), i.e., C fibers. Unlike the somatosensory system where acute stimulation of nociceptors often leads to pain, activation of respiratory nociceptors seldom leads to pain but instead can lead to disturbances in breathing, the urge to cough and dyspneic sensations, as well as strong parasympathetic reflex bronchospasm and secretions (19, 58). In addition, activation of neurokinin-containing C fibers in the lungs can evoke local axon reflexes that lead to neurogenic inflammatory reactions comprising plasma extravasation and inflammatory cell infiltration (60).

Virtually all afferent C fibers in both the somatosensory and visceral nervous systems can be stimulated by chemical mediators associated with inflammation. A better understanding of how C-fiber terminals in the respiratory tract are activated at sites of inflammation may reveal novel therapeutic targets that would work synergistically with the anti-inflammatory strategies aimed at relieving symptoms and suffering of those with inflammatory airway diseases. This review is meant to provide a relatively brief overview of this concept, focusing specifically on the mechanisms by which an inflammatory mediator directly activates bronchopulmonary C-fiber terminals.

Subtypes of C Fibers

The vast majority of bronchopulmonary afferent nerves are carried by the left and right vagus nerves. In the cat, among the 5,000 or so vagal fibers reaching the airways, ~4,000 are unmyelinated C fibers (2). Among the remaining 1,000 fibers are preganglionic parasympathetic nerves and myelinated afferent A fibers. In addition to vagal C fibers, the lungs and trachea are also innervated by a minority of spinal C fibers, with their cell bodies situated in the thoracic dorsal root ganglia (43). In the 1970s, work largely with dogs and cats by the Coleridges and their colleagues led to the recognition that the vagal C fibers could be subdivided based on their vascular sensitivity to chemical activators; i.e., whether the terminals were more immediately accessible via the pulmonary circulation or were more immediately accessible via the systemic or bronchial circulation. The former were termed pulmonary C fibers, and the latter were termed bronchial C fibers (17, 19). The pulmonary C fibers are included in the afferent fibers evaluated in the pioneering work of Paintal, who originally termed them “deflation” receptors and later “J-receptors,” since he intuited that they terminate juxtaposed to the pulmonary capillaries (69). Although some clear distinctions were noted in the activation profile of bronchial versus pulmonary C fibers, there has been little published in the way of thorough descriptions of their respective phenotypes.

Rather than being based on the location of the terminals, vagal C fibers in the respiratory tract have also been subdivided based on the location of their cell bodies. This has been most thoroughly evaluated in guinea pigs and mice (58, 65, 87). Retrograde tracing studies in guinea pigs have shown that the extrapulmonary and intrapulmonary structures receive similar amounts of innervation from neurons situated in both vagal sensory ganglia, namely the nodose ganglia and the jugular (or supranodose) ganglia (43). Neurons situated in the nodose ganglia are derived embryologically from the epibranchial placodes, whereas those in the jugular ganglia are derived, like DRG neurons, from the neural crest (3). In guinea pigs, the vagal C fibers in the extrapulmonary airways are derived predominately from jugular neurons, whereas nodose C fibers terminate mainly within the lungs (7, 87) (FIGURE 1). By putting the two nomenclatures together, one might surmise that bronchial C fibers are akin to jugular C fibers, whereas pulmonary C fibers and J-receptors are more akin to nodose C fibers. However, it should be kept in mind that there are also a minority of nodose C fibers in extrapulmonary airways and that there are many jugular C fibers that terminate deep in the lungs. What is clear is that nodose and jugular C fibers have very distinct activation profiles and that activation profile is relatively constant, irrespective of where in the respiratory tract the C fiber terminates. In fact, the distinction in the activation profile of bronchopulmonary nodose versus jugular C fibers even holds when nodose is compared with jugular C fibers in a different tissue, such as the esophagus (93). Importantly, elegant tracing studies are proving that jugular and nodose C fibers not only have distinct chemical activation profiles, but they also project to different nuclei in the CNS, with jugular fibers terminating in the para-trigeminal nucleus and nodose fibers terminating mainly in the nucleus of the solitary tract (24, 61) (FIGURE 1). This motivates investigations into the important questions as to the nature of the sensations and reflexes that are evoked by selective activation of neural crest C fibers (i.e., those from the jugular and dorsal root ganglia) versus placodal C fibers (derived from nodose ganglia) (10, 11).

FIGURE 1.

Peripheral terminations of nodose and jugular C fibers in the respiratory tract

Left: schematic illustration of the peripheral terminations of nodose (green) and jugular (red) C fibers in the respiratory tract, along with central terminations in the brain stem. The jugular C fibers terminate in extrapulmonary (larynx, trachea, main bronchi) and intrapulmonary tissue, whereas nodose C fibers mainly terminate within the lungs. Centrally, the jugular fibers terminate in the paratrigeminal nucleus (Pa5), whereas the nodose fibers terminate primarily in the nucleus of the solitary tract (nTS). Right: A, A’: photomicrograph and corresponding Nissl staining of a caudal guinea pig brainstem section showing the Pa5 and nTS locations for microinjection of cholera toxin subunit-b (CT-B) tracers conjugated with 488 (green) or 594 (red) fluorophores, respectively. B: the tracer placed in the nTS retrogradely labeled neurons in the nodose ganglion (NG), whereas tracer in the Pa5 retrogradely labeled neurons in the jugular neurons. C: data quantified in B as histograms. This figure from Ref. 24 and used with permission from Frontiers in Physiology.

Chemical activation of vagal C fibers has generally been studied using teased fiber single-unit recording techniques (19, 50, 68). With this technique, one can obtain the precise conduction velocity of the fiber under interrogation, but it is not possible to discern whether it is a nodose or a jugular C fiber. To study specifically nodose versus jugular C fibers, we have used extracellular recording electrodes placed near individual bronchopulmonary neurons within the nodose or jugular ganglion in ex vivo trachea-lung-vagus nerve preparations (see FIGURE 3). An ex vivo preparation is optimal for the analysis of a chemical mediator applied directly to a nerve’s receptive field with minimal indirect effects. However, it is less useful for the study of activation by physiologically relevant mechanical distortions of the receptive field that may occur during respiration, tissue edema, etc. More recently, we have adapted a two-photon Ca²⁺ imaging method to study activation of nodose and jugular bronchopulmonary C fibers. In this case, the tissues are isolated from transgenic mice expressing the genetically encoded Ca²⁺ indicator GCaMP6s under the control of sensory neuron-selective phosphoinositide-interacting regulator of transient receptor potential channels (Pirt) promoter [Pirt-Cre;R26-GCaMP6 mice (34)] or from mice having received intraganglionic injection of AAV-GCamP6s ~4 wk before experimentation. When action potentials generated at the peripheral terminals are conducted to the CNS, they also invade the cell soma in the sensory ganglion. GCaMP6s is highly sensitive to changes in calcium concentrations. We have noted that the calcium influx during a single action potential in a GCaMP6s-expressing vagal ganglion neuron is sufficient to generate a fluorescent signal detectable by two-photon microscope. Furthermore, we have shown that the intensity of the signal is a linear function of the action potential number elicited by electrical stimulation between 1 and 10 Hz (in 5-s trains). The enormous advantage of this imaging technique over single-fiber electrophysiolgical recordings, where often a single fiber is studied per animal per day, is that the activity of 100 s of bronchopulmonary fibers can be evaluated in a single experiment. The disadvantages are that one does not obtain conduction velocity information of the fiber, and less precise information is provided with respect to the pattern and frequency of action-potential discharge. The relative utility of these two techniques is described in our study on shingosine-1 phosphate activation of bronchopulmonary C fibers (70). An example of interferon gamma-induced activation of bronchopulmonary vagal afferent nerves detected by GCamp6s imaging is shown in FIGURE 3.

FIGURE 3.

Photograph and schematic of an ex vivo vagus-innervated mouse trachea-lungs preparation used to evaluate the activity of single C fibers in the airways by two-photon Ca²⁺ imaging or extracellular electrophysiological recording techniques

The inflammatory mediators are applied to the receptive fields in the airways via trachea perfusion. The activation of a C fiber in response to an inflammatory mediator can be recorded as Ca²⁺ increases or action potential discharges in the cell soma located in the nodose/jugular vagal ganglion complex, which is placed in the small chamber. The latter is separated from the bigger chamber housing the trachea and lungs. Activation of bronchopulmonary afferent nerves ~2 min after IFNγ infusion into the trachea is given as an example of two-photon Ca²⁺ imaging from nodose/jugular complexes of Pirt-Cre;R26-GCaMP6 mice. Action-potential discharges in response to bradykinin recorded from single nodose C fibers of a wild-type and a Trpa1/Trpv1 double-knockout mice are given as an example of extracellular recordings.

A summary of some of the properties of nodose versus jugular bronchopulmonary C fibers is found in Table 1. It should be kept in mind that information in this table is derived from studies that have been largely limited to mice, rats, and/or guinea pigs; caution is therefore warranted when generalizations to other species are made. Little is known about jugular versus nodose bronchopulmonary C fibers in larger mammals, including humans; although some of the key distinctions found in the mouse and guinea pig have held up, at least at the level of gene expression, in the monkey (36).

Table 1.

Characteristics of nodose and jugular bronchopulmonary C fibers

Feature	Nodose C Fibers	Jugular C Fibers
Embryonic origin	Epibranchial placode	Neural crest
Main neurotrophin receptor	TRKB	TRKA
	GFRα/RET	GFRα/RET
Substance P content	+/–	++
Central terminations	nTS	Para-trigeminal nucleus
Peripheral terminations	Mainly intrapulmonary	Extra- and intrapulmonary
Conduction velocity	≤1 m/s	≤1 m/s
Chemical activators1	TRPV1, TRPA1 agonists	TRPV1, TRPA1 agonists
	Bradykinin (B2)	Bradykinin (B2)
	Adenosine (A1, A2a)	Serotonin (5HT1, 5HT4)
	ATP (P2X2,3)	Nicotine
	Thrombin (PAR1)	Sphingosine 1-P (S1PR3)
	5HT (5HT3)	Interferons
	Nictoine
	Sphingosine 1-P (S1PR3)
	Interferons

¹ This is based on data largely from guinea pigs and mice and should not be imprudently generalized to other species.

Mechanisms of C-Fiber Activation by Inflammatory Mediators

Bronchopulmonary nociceptive C fibers can be activated by mechanical forces, e.g., those associated with mucus accumulation, anosmotic conditions, embolisms, or tissue engorgement, but the best-studied mechanism for their activation is via chemical mediators. What all activating stimuli have in common is their ability to set in motion ionic events at the terminal leading to membrane depolarization. This depolarization is referred to as the “generator potential.” If, and only if, the generator potential is of sufficient magnitude to activate voltage-gated sodium channels (NaVs) will the C fiber be activated. Otherwise, the generator potential will electronically fade back to the resting membrane potential without informing the CNS. For an inflammatory mediator to directly stimulate a C fiber, three discrete steps must be completed. First, with few exceptions, an inflammatory mediator must interact with its cognate receptor in the nerve membrane to form a mediator-receptor interaction. Second, the stimulated receptor must lead to opening (or closing) of ion channels that results in a generator potential. Finally, the generator potential must reach the voltage threshold for NaV activation, which results in a rapid Na⁺ influx across the membrane leading to the formation of action potentials and their conduction to the central terminals (FIGURE 2B). Below, we will briefly unpack each of these three events.

FIGURE 2.

Whole mount tracheal preparation immunostained for GFP

A: whole mount tracheal preparation immunostained for GFP showing an image of the terminations of a single (presumed) jugular C fiber in the tracheal mucosa in the guinea pig in which the vagal afferent neurons selectively expressed GFP following in vivo transfection with AAV-GFP (35). This structure was in the upper third of the trachea. Horizontal bar is 50 µm. The parental axon (indicated by asterisk) gives off multiple branches that ramify repeatedly (examples indicated by arrowheads) just beneath the epithelium. Terminal segments (examples indicated by arrows) are located between the epithelial cells. B: a description of the three major processes that are required for an inflammatory mediator to overtly activate a C fiber (the tracing is a patch-clamp recording from an isolated neuron). C: left is a dissection showing the mouse nodose/jugular vagal ganglion complex; right is an image of the jugular/nodose complex dissected from a Wnt1Cre/R26R mouse. The dark neurons reflect Wnt1-expressing neural crest-derived jugular neurons; the unstained neurons are placodal-derived nodose neurons. Note that the rostral section of the complex (J) comprises mainly jugular neurons, whereas the caudal section N comprises nodose neurons. In between is a section of the complex (X) that contains a mixture of nodose and jugular neurons. D: depiction of two methods used to obtain TRPV1-expressing neurons for RNAseq analysis; one based on capsaicin responsiveness (left), and the other based on red fluorescent neurons from TRPV1-tdTomato mice (right), modified from Ref. 89. E: example from RNAseq analysis of mRNA expression of receptors for inflammatory mediators that were essentially not expressed by TRPV1-tdTomato neurons (red) or that were expressed in nodose and/or jugular TRPV1-tdTomato neurons (black). Data were obtained from Ref. 89.

Mediator-Receptor Interactions in Bronchopulmonary C Fibers

Historically, the most common chemical mediators used to stimulate bronchopulmonary C fibers were phenyldiguanide (now recognized as a 5HT₃ receptor agonist) and capsaicin (now known to be TRPV1 channel agonist) (19). Over the decades, the list of inflammatory mediators for C-fiber activation has grown to include other biogenic amines (13, 14, 22, 71, 74), certain eicosanoids and other lipid mediators (16, 18, 70, 76), TRPV1 and TRPA1 activators (64), and cytokines (52, 89). There has been little in the way of a systematic evaluation of which chemical mediators can directly interact with receptors at the terminals leading to airway C-fiber activation. A few years ago, we begun such an evaluation in the mouse (89). We dissociated neurons from the nodose and jugular ganglia, and used a low-input, deep RNA sequence (transcriptome) analysis to determine which, among the myriad inflammatory mediator receptors, are expressed by the neurons (FIGURE 2, C–E). To focus on C fibers, we took advantage of the fact that capsaicin, via TRPV1, is a strong activator of both nodose and jugular C fibers. In one study, we “picked” for analysis only those neurons expressing td-tomato from genetically engineered TRPV1-td-Tomato mice. In a parallel study, we used wild-type mice and picked only those neurons that responded to capsaicin with an elevation in intracellular calcium (FIGURE 2D). Each method has advantages and disadvantages; fortunately, each method yielded relatively similar results. The extensive data from the nodose and jugular C-fiber neuron transcriptome analyses are available as a supplement in our paper (89).

As expected, there were receptors expressed by nodose and jugular C fibers for many inflammatory mediators, but this represented a modest group of receptors. The vast majority of inflammatory mediator receptors were not expressed by either nodose or jugular C-fiber neurons (FIGURE 2E). It is worth keeping in mind that, even if the receptor mRNA is expressed, it does not mean that its stimulation will lead to nerve terminal activation. This limited number of expressed receptors supports a strategy whereby blocking a single-mediator receptor may lead to profound reduction in C-fiber activation in certain disease states; which receptor to block will of course depend on the nature of the pathology. In recent years, for example, blocking the receptor for ATP-induced activation of C fibers (P2X₃ or P2X_2/3) has been found to substantially reduce coughing in those suffering from chronic idiopathic cough (1).

Induction of the Generator Potential

Ionotropic receptor-mediated generator potentials.

Chemical mediators that directly activate C fibers can be mechanistically subdivided into two categories: those that stimulate ionotropic receptors (ligand-gated ion channels) and those that activate metabotropic receptors. The mechanism by which a mediator leads to a generator potential is rather self evident for ionotropic receptors, since the receptors themselves are generally non-selective cation channels that are opened on interaction with its ligand. Examples of ionotropic receptors relevant to bronchopulmonary C-fiber activation include 5-HT₃ receptors, cholinergic nicotinic receptors, purinergic P2X receptors, and TRP channels.

Serotonin (5-HT3 receptors). Phenydiguanide has long been used to activate bronchopulmonary C fibers. This is now known to selectively activate the terminals via interacting with 5-HT3 receptors, although curiously not guinea pig 5-HT3 receptors (47). Among the numerous 5-HT receptors, 5-HT3 is the only ionotropic receptor. When 5-HT or other 5-HT3 agonists interact with the receptor/channel, it leads to a conformational change of the receptor that opens the non-selective cation pore. In guinea pigs and mice, the 5-HT3 receptor gene is selectively expressed in nodose C fibers; accordingly, 5-HT3 selective agonists strongly activate nodose but not jugular bronchopulmonary C fibers in these species (14, 74).

Acetylcholine (nicotinic receptors). Activation of cholinergic nicotinic ionotropic receptors can lead to strong activation of bronchopulmonary C fibers. The extent to which endogenous acetylcholine has access to nicotinic receptors on C-fiber terminals is unknown, but exogenously applied nicotine can directly depolarize vagal sensory neurons and lead to action-potential discharge at the terminals (40, 49). Indeed, when nonsmokers volunteer to inhale a puff of smoke from a nicotine-containing cigarette, the intense irritations and urge-to-cough sensations are largely inhibited by treatment with a nicotinic receptor antagonist (48).

Adenosine triphosphate (P2X receptors). ATP can stimulate both metabotropic GPCRs, referred to as P2Y receptors, and ionotropic receptors, referred to as P2X receptors. Virtually all nodose and jugular neurons innervating the airways express P2X3 receptors, whereas P2X2 is expressed selectively by nodose neurons (45, 65). Pelleg and Hurt were first to show that ATP strongly activates pulmonary C fibers in the dog via P2x receptor stimulation (71). In guinea pigs and mice, ATP activates nodose but not jugular C fibers (65, 87). This can be explained by the fact that activation of the heteromeric P2X2,3 receptor leads to larger and more slowly inactivating depolarizing currents in nodose neurons compared with the much smaller and rapidly inactivating currents evoked via the homomeric P2X3 receptor in jugular neurons (45). The explanation for a lack of sufficient current via the P2X3 homomeric receptors to cause action-potential discharge at jugular C-fiber terminals could include rapid inactivation, insufficient density, or inhibitory regulation. An example of an inhibitory regulator known to be expressed by vagal sensory neurons is Pirt, a small transmembrane protein that has been shown to inhibit the amount of ionic conductance through P2X3 receptors (26).

Transient receptor potential channels. Many transient receptor potential (TRP) channels can act as ligand-gated ion channels. The two most relevant ligand-gated TRP channels expressed by bronchopulmonary C fibers are TRPV1 and TRPA1 (89); their activation leads to coughing in human volunteers (6, 20). TRPV1 is the receptor for capsaicin, a stimulant that has long been known to activate airway C fibers (19). Potential endogenous ligands for TRPV1 on bronchopulmonary C fibers include certain lipoxygenase products of arachidonic acid, endocannabinoids, and acidic solutions (39, 79).

TRPA1 is more promiscuously activated by chemicals than TRPV1. TRPA1 can be activated by several environmental irritants, including ozone, isocyanates, and saturated aldehydes (27, 85). Endogenously, many electrophilic compounds found in inflamed airways have been noted to activate bronchopulmonary C-fiber terminals exclusively via TRPA1 activation; these include alkenals such as 4-oxononenal (84), oxidative metabolites of several prostaglandins (86), mitochondria-derived reactive oxygen species (66), and nitrated fatty acids (83).

GPCRs-mediated generator potentials.

The majority of inflammatory autacoids that directly activate C fibers are agonists for GPCRs. Examples of such autacoids that can lead to strong activation of bronchopulmonary C fibers include bradykinin (B2 receptors) (25, 31, 33), adenosine (A1 and A2a receptors) (13), prostanoids (32, 76), protease activated receptors (46), histamine (22), and serotonin (5-HT1 and 5HT2 receptors) (74). Unlike the ionotropic receptors, it is less obvious how stimulation of a GPCR leads to a generator potential at the level of the C-fiber terminal. Knowledge of GPCR signal transduction in sensory nerves has accrued from studies carried out in heterologous cell systems and at the level of sensory cell bodies usually dissociated from the dorsal root ganglia. For technical reasons, there have been no direct studies of the signaling mechanisms at the terminal receptive fields of C fibers. This is more than a mere passing concern, given the importance of organelles and the microenvironment in the GPCR signaling processes; both the microenvironment and the inventory of organelles are likely to be different between the cell soma and nerve terminals (41, 42).

TRPV1 and TRPA1 are often implicated in GCPR-mediated generator potentials, but how this comes about is both complex and multifaceted (88). One idea is that TRPV1 and TRPA1 channels in the membrane are inhibited by the membrane phospholipid, phosphatidylinositol 4,5 bisphophate [PtdIns (4,5P2)]. A mediator interacting with its GPCR leads to the activation of phospholipase C (PLC), which cleaves the inositol moiety of PtdIns (4,5P2), thereby removing its inhibitory influence on the channel (12). Other studies indicate more complex roles of phoshotidylinositol in TRPV1 regulation (54). Another mechanism by which GPCR can lead to TRPV1 activation is via the liberation of arachidonic acid from the sn-2 positon of membrane phospholipids and the subsequent formation of lipoxygenase products that directly interact with the channel as intracellular “capsaicinoids” (78). Third, GPCR-dependent activation of kinases (e.g., protein kinase C and protein kinase A) can phosphorylate TRP channels in a manner that leads to an enhancement in their activation (5, 75).

With respect to bronchopulmonary C-fiber terminals, pharmacologically inhibiting TRPV1 has been found to reduce the total number of bradykinin-induced action potentials in jugular C fibers innervating the guinea pig trachea (8) and to reduce bradykinin-induced coughing in conscious guinea pigs (28). The reduction of bradykinin-induced action-potential discharge by TRPV1 antagonism was mimicked by a lipoxygenase inhibitor and was absent when lipoxygenase was blocked (8). By contrast, the peak action-potential discharge frequency in nodose C fibers evoked by bradykinin B2-receptor activation in mouse lungs was not different between wild-type and TRPV1 knockout mice (38). Moreover, in unpublished observations, we have noted that bronchopulmonary nodose C fibers in mice lacking both TRPV1 and TRPA1 respond normally with action-potential discharge to both bradykinin (n = 6, P > 0.1; see example in FIGURE 3) and PAR1 agonists; nonselectively blocking TRPA1, TRPV1, and other TRP channels also had no influence on GPCR-mediated action potential discharge (not shown).

That neither TRPV1 nor TRPA1 appear to contribute substantively to the GPCR activation of nodose C-fiber terminals in the mouse airways begs the question as to what types of ion channels other than, or in addition to, TRP channels are opened (or closed) to cause the generator potentials. Chloride channels represent one possibility. Adult neurons have an active Na⁺-K⁺-Cl⁻ cotransporter (NKCC1) that increases the intracellular chloride concentration to a point where its reversal potential is significantly more positive than the resting potential; when chloride channels open a depolarizing current due to chloride, efflux therefore ensues (56, 82). When guinea pig airway-specific neurons dissociated from the nodose ganglia were exposed to bradykinin, the depolarizing current was comprised mainly of a calcium-activated chloride conductance along with an inhibition of a potassium conductance (67). More recent findings have been made with nociceptive neurons isolated from DRGs, where the GPCR-mediated chloride conductance was attributed to the opening of TMEM16A/Ano1 chloride channels secondary to the release of intracellular calcium stored in the endoplasmic reticulum juxtaposed to the channel (30, 53). In nodose neurons innervating the stomach, similar findings have been made with cholecystokinin-induced stimulation, except the chloride channel was found to be TMEM16B/Ano2 (90). Bradykinin was also found to activate a chloride conductance in airway jugular neurons; moreover, non-selective chloride channel blockers partially inhibited bradykinin-induced action-potential discharge in jugular C-fiber terminals innervating the trachea (51) and can lead to the inhibition of cough (56).

The nature of the potassium conductance inhibited by bradykinin in nodose neurons has not been clearly delineated. The so-called M-channels are a candidate (53). The M-current, so named because it was inhibited by activation of muscarinic cholinergic receptors, is now known to be mediated by Kv7 channels, which in turn are products of KCNQ genes. The expression of KCNQ mRNA and Kv7 subunits has been demonstrated in mouse and rat nodose neurons (81, 91). The functional M-channels expressed in nodose neurons specifically innervating mouse lungs, most likely composed of KCNQ2 and/or KCNQ3 (Kv7.2 and Kv7.3) subunits, can be inhibited by GPCR activation, but blocking this channel only modestly depolarizes the nerve and does not overtly lead to action potentials at the nerve terminals (81). Opening these channels with retigabine, a Kv7-channel opener developed for the treatment of epilepsy, causes a large hyperpolarization of the membrane and strongly inhibits the excitability of the C-fiber nerve terminals in the respiratory tract. Consistent with this, retagabine inhibits cough in conscious mice evoked by nebulized irritant gases (81).

Members of Kv1 voltage-gated potassium channel family may also contribute to the regulation of generator potential. These channels are encoded by Kcna genes, rapidly activate at negative (or subthreshold) voltages, and exhibit very slow or little inactivation (73). The selective potassium channel blocker α-dendrotoxin (α-DTX) causes action-potential discharge in nodose nociceptors terminating in the guinea pig trachea (59) and capsaicin-sensitive and -insensitive C fiber in mouse lungs (Sun H, Patil MJ, Undem BJ, unpublished observations). This is most likely the result from inhibition of Kv1.1-, Kv1.2-, and/or Kv1.6-containing heterotetrameric Kv1 channels (72). Another family of ion channels whose inhibition by GPCR-signaling mechanisms could potentially lead to GPCR-mediated generator potentials is the two-pore domain potassium channel (K2P) family encoded by KCNK genes. They are constitutively open at the resting membrane potential, albeit to variable degrees, and responsive to a variety of physical and chemical stimuli. The 15 members of the mammalian K2P family are subdivided into six subfamilies according to sequence similarities. Three of these subfamilies (TASK, TREK, and TRESK) are known to be regulated by GPCR signaling pathways. Particularly, activation of G_q-coupled receptors inhibits TASK and TREK channels and increases neuronal excitability (29, 55).

Cytokine receptor-mediated generator potentials.

As orchestrators of the afferent and efferent limbs of immune responses, cytokines can indirectly have profound effects on nociceptor activity, but the evidence for direct cytokine-mediated activation of C-fiber terminals is relatively scant. As shown in FIGURE 2E, mouse vagal sensory neurons express little in the way of cytokine receptor mRNA, but they do express receptors for both type 1 and type 2 interferons. Consistent with this, IFN-α, IFN-β, and IFN-γ can directly activate mouse neurons isolated from the nodose ganglia (89) and also evoke action-potential discharge at C-fiber terminals in the lungs (see example of IFN-γ activating bronchopulmonary afferent terminals in FIGURE 3). IFN-γ has also been found to acutely activate the cell bodies of neurons isolated from rat nodose ganglia (23). TNF-α can directly interact with neurons isolated from rat nodose neurons, but its physiological consequence appears to be more of a sensitization of the terminal to other stimuli than an overt activation (action-potential discharge) (52). IL-1β has been found to acutely stimulate a subset of neurons in mouse nodose ganglia, but whether these are nociceptors was not determined (80).

How activation of a cytokine receptor signals to ion channels involved in generator potentials in airway nociceptors is a relatively unexplored area of research. Janus kinase (JAK) is typically involved in interferon signaling, and JAK 1 is richly expressed in nodose and jugular C-fiber neurons (89). Moreover, a JAK antagonist inhibited IFN-γ-induced stimulation of rat nodose cell bodies (23) and IFN-β-induced activation of vagal afferent nerve terminals in the mouse lung (Patil MJ, Undem BJ, unpublished observations). The nature of the ion channels downstream from JAK that are involved in interferon-induced vagal afferent nerve activation remains unknown.

Action-Potential Induction and Conduction in Bronchopulmonary C Fibers (NaVs)

Irrespective of whether the generator potential is evoked by activation of ligand-gated ion channel or through GPCR or cytokine receptor signaling to an ion channel, it will be of no physiological relevance if it does not reach the voltage threshold for activation of NaV channels. The activation of NaVs is a sine qua non of action-potential generation at the terminals and its conduction along the C fibers to the central terminals in the brain stem.

The pore of the NaVs comprise a large α subunit with four homologous domains. A major breakthrough in NaV1 research has been the discovery that the α subunits are encoded by nine distinct genes leading to nine distinct NaV1 subtypes referred to as NaV 1.1–1.9 (9). Particularly relevant to potential therapeutics is the finding that NaV1 subtypes are differentially expressed in different tissue and nerve types. For example, striated muscle expresses mainly NaV1.4, whereas cardiac muscle expressed mainly NaV1.5. An analysis of gene expression in airway-specific jugular and nodose C-fiber neurons has revealed that they express primarily NaV 1.7, 1.8, and 1.9 (37, 63) (FIGURE 4A). These same NaV subtypes have been found to be expressed in the somatosensory system as well. In the somatosensory system, however, these NaV1s are somewhat selectively expressed in small-diameter nociceptive neurons. (For a recent thorough and informative review of NaV1 subtypes in somatosensory nociception, see Ref. 4.) In the vagal sensory system, NaV 1.7, 1.8, and 1.9 are not only expressed in C-fiber nociceptive neurons but also in the large A-fiber neurons, including low-threshold mechanosensitive nerves (44).

FIGURE 4.

Expression of NaV1 subtype mRNA in TRPV1-expressing nodose neurons

A: expression of NaV1 subtype mRNA in TRPV1-expressing nodose neurons. Data were obtained from Ref. 89. B: excitability of patch-clamped nodose neurons obtained from untreated mice (wild type), mice treated with AAv NaV1.7 scrambled shRNA (control), or AAv NaV 1.7-targeting shRNA. The amount of current required to evoke an action potential was ~20 pA in untreated and scrambled shRNA-treated neurons; approximately four times more current was needed to evoke action potentials when NaV1.7 mRNA was silenced (see Ref. 63). The action potential in the shRNA-treated neurons are likely due to NaV1.8 activaiton. C: when AAv NaV1.7 shRNA was injected bilaterally into the vagal ganglia (red), the animals failed to cough in response to inhaled citric acid; naive mice coughed repetitively to the same challenge (black).

With respect to that NaV subtype most important for action-potential initiation and conduction, attention can be focused on NAV1.7 and 1.8. Although NaV1.9 is extensively expressed in nearly all vagal bronchopulmonary sensory neurons, the kinetic characteristics of this channel render it unlikely to contribute directly to the rapid upstroke of the spike and the conduction of the action potential (4). When the expression of NaV1.7 was selectively silenced in guinea pig vagal sensory ganglia using AAV-shRNA methodology, the excitability of the neurons was substantially reduced (much more depolarizing current was required to evoke an action potential) (FIGURE 4B), and the conduction of action potentials along the vagal axons was largely blocked. Consistent with this was the observation that NaV1.7-deficient guinea pigs failed to cough in response to tussive stimuli (62, 63) (FIGURE 4C).

Over the past decade, advances have been made in the development of compounds that can selectively block NaV1.7 versus NaV1.8. By using a pharmacological approach, it was discovered that the dependence on NaV1.7 and NaV1.8 may be different for the initiation of action potentials at the C-fiber endings in the tissue versus the conduction of action potentials along the vagus nerves. When tetrodotoxin, a drug that blocks NaV1.7 but not NaV1.8, was limited to the trachea or to the lungs (as would occur with an inhaled sodium channel blocker), it had little effect on action-potential initiation in jugular C fibers, yet when tetrodotoxin was applied selectively to vagus nerve outside of the airways, action-potential conduction in the jugular C fiber was blocked. Consistent blockage of action-potential formation in jugular C fibers within the airways required the blockade of NaV1.7 and NaV1.8. This was not the case with nodose C fibers, where NaV1.7 blockers effectively inhibited both induction of action potentials within the airway tissue as well as their conduction along the nerve fibers in the vagi (37).

If NaV1.7 and in some cases NaV1.7 and 1.8 can account for action-potential initiation and conduction, what is the role of NaV1.9 in the nodose and jugular C fibers? NaV1.9 is expressed in virtually every nodose and jugular C fiber innervating the respiratory tract. NaV1.9 has slow kinetics and is activated at more negative voltages than NaV1.7 or NaV1.8; in fact, NaV1.9 can be activated at membrane potentials only slight more depolarized than the resting potential (21). One might speculate that opening NaV1.9 may occur during a generator potential that then further drives the potential to the thresholds for activation of NaV1.7 and NaV1.8. In other words, NaV1.9 may play an important role as a “threshold channel” in regulating the susceptibility of the C fibers to chemical and mechanical activation.

Conclusions and Therapeutic Implications

A feature common to vagal and spinal sensory C fibers is their direct activation by chemical mediators associated with inflammation. During chronic inflammatory diseases, this may lead to a persistent action-potential discharge and a central sensitization that can ultimately lead to sensations and reflexes that go beyond providing for a useful nocifensive system (57, 58, 92). In the airways, chronic activation of bronchopulmonary C fibers may lead to non-productive, persistent coughing and dyspnea, as well as reflex bronchospasms and excessive secretions. Therapeutic strategies aimed at quieting bronchopulmonary C fibers would in theory work additively or synergistically with existing anti-inflammatory therapies.

There are basically three general approaches to quiet the C fibers in the airways. First, block the receptors for a given activating mediator; second, block the generator potential that occurs downstream from the receptor activation; third, inhibit NaVs, causing a loss (or change in discharge pattern) of action potentials reaching the brain stem. Recent studies showing the efficacy of blocking ATP P2X receptors in the treatment of chronic cough (1) support the mediator receptor antagonist approach. The utility of inhibiting generator potentials or action-potential initiation and conduction awaits trials with respective selective and potent channel blockers or activators. Similar strategies have been long considered in attempts to find peripheral acting analgesics to replace opioids in the treatment of various pain disorders. A potential advantage in the treatment of airway disorders is that the generator potential modulators and NaV1 antagonists can be delivered topically (by inhalation), thereby reducing the risk of systemic side effects.