Pharmacology of Bradykinin-Evoked Coughing in Guinea Pigs

Abstract

Bradykinin has been implicated as a mediator of the acute pathophysiological and inflammatory consequences of respiratory tract infections and in exacerbations of chronic diseases such as asthma. Bradykinin may also be a trigger for the coughing associated with these and other conditions. We have thus set out to evaluate the pharmacology of bradykinin-evoked coughing in guinea pigs. When inhaled, bradykinin induced paroxysmal coughing that was abolished by the bradykinin B₂ receptor antagonist HOE 140. These cough responses rapidly desensitized, consistent with reports of B₂ receptor desensitization. Bradykinin-evoked cough was potentiated by inhibition of both neutral endopeptidase and angiotensin-converting enzyme (with thiorphan and captopril, respectively), but was largely unaffected by muscarinic or thromboxane receptor blockade (atropine and ICI 192605), cyclooxygenase, or nitric oxide synthase inhibition (meclofenamic acid and N^G-nitro-L-arginine). Calcium influx studies in bronchopulmonary vagal afferent neurons dissociated from vagal sensory ganglia indicated that the tachykinin-containing C-fibers arising from the jugular ganglia mediate bradykinin-evoked coughing. Also implicating the jugular C-fibers was the observation that simultaneous blockade of neurokinin₂ (NK₂; SR48968) and NK₃ (SR142801 or SB223412) receptors nearly abolished the bradykinin-evoked cough responses. The data suggest that bradykinin induces coughing in guinea pigs by activating B₂ receptors on bronchopulmonary C-fibers. We speculate that therapeutics targeting the actions of bradykinin may prove useful in the treatment of cough.

Introduction

Bradykinin is a peptide autacoid formed from precursor kininogens by tissue and plasma peptidases. Multiple chemical insults and pathologic conditions result in bradykinin generation in tissues and on mucosal surfaces. The ability of bradykinin to initiate vasodilatation, plasma exudation, leukocyte activation, and reflexes and sensations attributed to the stimulation of visceral and somatic nociceptors established bradykinin as a mediator of many acute and chronic inflammatory conditions (Joseph and Kaplan, 2005; Leeb-Lundberg et al., 2005; Kaplan and Joseph, 2014).

Kinins have been implicated in inflammatory responses of the airways and lungs initiated by allergen, airway acidification, cold, dry air inhalation, viral infections, and gram negative bacterial infections, and in other inflammatory conditions promoting recruitment of neutrophils and/or eosinophils to the airways (Proud et al., 1983, 1988; Bertrand et al., 1993; Ricciardolo et al., 1994a, 1999; Coyle et al., 1995; Featherstone et al., 1996; Yoshihara et al., 1996; Grünberg et al., 1997; Folkerts et al., 2000; Scuri et al., 2000; Turner et al., 2001; Abraham et al., 2006; Arndt et al., 2006; Broadley et al., 2010; Hewitt and Canning, 2010; Taylor et al., 2013; Sahoo et al., 2014). The bronchospasm, mucus secretion, airway microvascular dilatation, and plasma exudation evoked by exogenously administered bradykinin are primarily mediated by bradykinin B₂ receptor activation (Nakajima et al., 1994; Abraham et al., 2006; Broadley et al., 2010). The B₁ receptor–dependent effects have also been implicated in respiratory diseases, but bradykinin has little or no affinity for B₁ receptors. The actions of bradykinin are limited by metabolizing peptidases including the angiotensin-converting enzyme (ACE) and neutral endopeptidase, and by rapid B₂ receptor desensitization (Wolsing and Rosenbaum, 1993; Leeb-Lundberg et al., 2005; Broadley et al., 2010; Zimmerman et al., 2011).

When inhaled, bradykinin causes coughing in humans and guinea pigs (Choudry et al., 1989; Katsumata et al., 1991; Canning et al., 2004; Grace et al., 2012; Smith et al., 2012). This peptide autacoid may also cause the coughing associated with ACE inhibitor therapy (Fox et al., 1996; Morice et al., 1997; Hirata et al., 2003; Dicpinigaitis, 2006; Cialdai et al., 2010; Mutolo et al., 2010; Mahmoudpour et al., 2013). Bradykinin-induced cough likely results from its direct effects on bronchopulmonary C-fibers (Kaufman et al.; 1980; Bergren, 1997; Kajekar et al., 1999). However, the indirect effects of B₂ receptor activation might also contribute to its capacity to initiate cough (Grace et al., 2012). For example, bradykinin induces eicosanoid formation in the airways, including prostaglandin E₂, thromboxane, and 15-hydroxyeicosatetraenoic acid (Salari and Chan-Yeung, 1989; Arakawa et al., 1992). Prostanoids can both enhance and directly initiate coughing (Shinagawa et al., 2000; Liu et al., 2001; Xiang et al., 2002; Gatti et al., 2006; Maher et al., 2009; Ishiura et al., 2014). These effects of the bronchoconstrictor prostanoids could result from the activation of mechanically sensitive vagal afferent nerves innervating the airways and lungs (Bergren, 1997; Canning et al., 2001).

We have addressed the hypothesis that bradykinin evokes coughing by activating bronchopulmonary C-fibers as well as the mechanically sensitive vagal afferents innervating the airways and lungs. We further hypothesized that these vagal afferent nerve subtypes act synergistically to promote coughing in response to bradykinin. On the contrary, the results suggest dissociation of the prostanoid-dependent bronchospasm evoked by bradykinin and the prostanoid-independent effect of this peptide on the cough reflex.

Materials and Methods

Our institutional Animal Care and Use Committees approved all of the experiments described in this study. Male Hartley strain guinea pigs (200–400 g, Hilltop Lab Animals, Scottdale, PA) were purchased pathogen free and housed in accredited housing facilities with food and water provided ad libitum.

Bradykinin-induced coughing was studied in awake guinea pigs placed in a recording chamber continuously filled with fresh, room temperature air. Bradykinin was delivered by aerosol to the chambers using an ultrasonic nebulizer (particle size: <5 μm). Breathing patterns, respiratory rate, and cough were monitored visually and by measuring pressure changes within the chambers, which were recorded digitally (Biopac Systems, Goleta, CA). Bradykinin was delivered as a single dose (0.1–10 mg·ml⁻¹) for 10 minutes, or with cumulatively increasing doses (1–10 mg·ml⁻¹), delivered for 5 minutes, with 5 minutes in between each dose. In some animals, citric acid (0.01–0.1 M) was also used to evoke cough. Results are presented as the mean ± S.E.M. cumulative coughs evoked. Differences among treatment groups were assessed by analysis of variance, with differences among treatment groups evaluated by Scheffé’s F test for unplanned comparisons. Statistical significance was set at P < 0.05.

We attempted to modify bradykinin-evoked coughing by drug pretreatments given intraperitoneally or by aerosol. These interventions were chosen for their known ability to modify bradykinin-evoked responses in the airways and lungs and were administered at doses that were selected based on the results of previous studies or following validation studies performed prior to the cough experiments. The neutral endopeptidase inhibitor thiorphan and the ACE inhibitor captopril were delivered as aerosols prior to bradykinin challenge to evaluate the role of metabolism in regulating bradykinin-evoked coughing. These peptidase inhibitors were dissolved in saline and administered at doses of 1 mg·ml⁻¹, with vehicle control experiments carried out in parallel. The bradykinin B₂ receptor antagonist HOE 140 (1 mg·ml⁻¹) was also administered by aerosol. The thromboxane receptor antagonist ICI 192605 (1 mg·kg⁻¹ i.p. or 10 μM delivered as an aerosol for 10 minutes) and the cyclooxygenase inhibitor meclofenamic acid (1 mg·kg⁻¹ i.p.), were used to measure the contribution of prostanoids in the response to bradykinin. The nitric oxide synthase inhibitor N^G-nitro-L-arginine (L-NNA) (0.1 mM) and the muscarinic receptor antagonist atropine (1 mg·ml⁻¹) were administered to modify bronchospasm during bradykinin-evoked cough. Both were administered by aerosol for 10 minutes prior to bradykinin challenge. Atropine was dissolved in saline, while L-NNA was first dissolved in 0.1N HCl at a concentration of 0.1 M, and then diluted 1000-fold in saline to the concentration used for aerosol delivery. The role of neurokinin (NK) receptors in bradykinin-evoked coughing was determined by pretreating the animals with various combinations of NK₁ (SR14033 and CP99994), NK₂ (SR48968), and NK₃ (SR142801 and SB223412) receptor antagonists, administered at 3 mg·kg⁻¹ each by i.p. injection or by aerosol (1 mg·ml⁻¹ each). All studies were designed as parallel group, unpaired experiments. Vehicle control experiments for each intervention were carried out in parallel. Drugs used in an attempt to modify bradykinin-evoked coughing were administered 10–30 minutes prior to bradykinin challenge.

We also studied bradykinin-induced bronchospasm and coughing evoked by mechanical stimulation of the airway mucosa in anesthetized guinea pigs. Guinea pigs were anesthetized using urethane (1.5 g·kg⁻¹, i.p.), which produces a stable anesthesia lasting well beyond the duration of these experiments. The absence of withdrawal responses or cardiovascular responses to a sharp pinch of a hind limb was used to determine the adequacy of the anesthesia. Although no animals required additional anesthesia in these experiments, supplemental urethane would have been given had arousal been noted. Once the animals were anesthetized and placed supine on a warming pad, a midline incision exposed the trachea, which was cannulated at its caudal-most end. To study mechanically induced cough, we either probed the laryngeal mucosa with a von Frey filament (producing >1 mN of force), or mechanically stimulated the intrathoracic trachea and carina by threading a length of 6-0 suture through the tracheal cannula and toward the carina (Canning et al., 2004). These stimuli typically evoke a single cough, and no more than two coughs; therefore, the results are presented as a percentage of the animals coughing.

Bradykinin-induced bronchospasm was studied as described previously (Canning et al., 2001). Once the trachea had been cannulated, the animals were paralyzed with succinylcholine (2.5 mg·kg⁻¹, s.c.). Guinea pigs were then mechanically ventilated (60 breaths/minute, 6 ml·kg⁻¹ b.wt. tidal volume, 3–5 cm/H₂O of positive end-expiratory pressure (to limit the prominent gas trapping that occurs during bronchospasm in guinea pigs) (Stengel et al., 1995). These ventilation parameters created a baseline peak pulmonary inflation pressure (PIP) of 8–12 cm/H₂O. The abdominal aorta and vena cava were exposed by abdominal incision and cannulated. Blood pressure and heart rate were monitored using a pressure transducer connected to the cannula in the aorta. To assess adequacy of anesthesia following paralysis, we monitored changes in heart rate and blood pressure in response to a sharp pinch of a forelimb. Additional anesthetic would have been provided if responses to the pinches were noted (no animals required additional anesthetic). Bradykinin (0.1–2 nmol·kg⁻¹) was administered intravenously to evoke bronchospasm, which was monitored by recording the PIP with a pressure transducer connected to a side port of the tracheal cannula. We use the PIP as a surrogate measure of bronchospasm and interpret changes in PIP as the net effects of interventions on airway resistance and lung compliance (Arakawa et al., 1992; Broadley et al., 2010; Keir et al., 2015). Heart rate, blood pressure, and PIP were recorded digitally (Biopac Systems). Doses were administered at 5-minute intervals, with volume histories (1 to 2 tidal breath holds given by preventing consecutive lung deflations) used to reverse any residual airway obstruction 1 minute prior to the administration of each dose. At the end of these experiments, guinea pigs were asphyxiated by carbon dioxide, followed by exsanguination.

The effects of bradykinin on [Ca⁺⁺] measured intracellularly were recorded in vitro in retrograde-labeled vagal afferent neurons acutely dissociated from the nodose and jugular ganglia. Guinea pigs (150–200 g) were anesthetized with ketamine and xylazine (60 and 6 mg·kg⁻¹, s.c.). Once anesthetized and placed supine on a warming pad, the neck was shaved and a small incision (5–10 mm) was made in the neck to expose the trachea. The neuronal tracer 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate was injected (3–4 μl) into three locations in the extrapulmonary airways (larynx, extrathoracic trachea, and carina) using a Hamilton syringe with a 30 gauge needle. The incisions were sutured shut, coated with betadine, and the animals allowed to recover under close observation. After 2 to 3 weeks, the animals were euthanized and the nodose and jugular ganglia were removed. The ganglia neurons were dissociated and adhered to coverslips in culture media overnight. Retrograde-labeled neurons from the airways were visualized by fluorescent microscopy. The Ca⁺⁺ influx was recorded in these retrograde-labeled neurons using a Fura-based assay as described elsewhere (Lee et al., 2005). Responses were normalized to the effects of ionomycin.

Reagents.

Atropine, bradykinin, captopril, citric acid, ionomycin, L-NNA, meclofenamic acid, succinylcholine, thiorphan, urethane, and xylazine were purchased from Sigma (St. Louis, MO). ICI 192605 was purchased from Tocris (Minneapolis, MN). U46619 was purchased from Cayman (Ann Arbor, MI). Glaxosmithkline (King of Prussia, PA) and Sanofi-Aventis (Montpellier, France) kindly provided CP99994, SB223412, SR48968, SR142801, and SR140333. HOE 140 was a generous gift from Hoechst (Kansas City, MO). Drugs were dissolved in 0.9% saline solutions except the NK receptor antagonists, which were dissolved initially in dimethylsulfoxide (30 mg·ml⁻¹) and then further diluted into saline. Drugs were delivered by aerosol (thiorphan, captopril, L-NNA, atropine, and HOE 140) or were administered by i.p. injection, with injection volumes <500 μl.

Results

Bradykinin Evokes Paroxysmal Coughing: Role of B₂ Receptors and Modulation by Peptidases.

When delivered as an aerosol to awake guinea pigs, bradykinin (1–10 mg·ml⁻¹) evoked coughing in a dose-dependent manner. With no pretreatments (e.g., peptidase inhibitors) at the outset of experimentation, 10-minute aerosol challenges with single doses of 0.1, 1, and 3 mg·ml⁻¹ bradykinin were in most animals subthreshold for initiating cough. Only aerosol doses of 5 mg·ml⁻¹ (9 ± 4 coughs; n = 7) and 10 mg·ml⁻¹ (14 ± 3 coughs; n = 20) bradykinin reliably evoked cough in control animals. The bradykinin B₂ receptor antagonist HOE 140 (1 mg·ml⁻¹ delivered as an aerosol) completely abolished the cough responses evoked by bradykinin (Fig. 1).

Fig. 1.

Bradykinin evokes paroxysmal coughing in conscious guinea pigs via bradykinin B₂ receptor activation. (A) The representative trace of bradykinin-evoked coughing (10 mg·ml⁻¹) illustrates the typical pattern of cough evoked by the inflammatory peptide. Within minutes of initiating the aerosol challenge, multiple coughs in rapid succession occur, often ending abruptly despite continued bradykinin challenge. Expiratory efforts manifest as upward deflections in chamber pressure (see Materials and Methods for further details). (B) The bradykinin B₂ receptor antagonist HOE 140 (1 mg·ml⁻¹, delivered as an aerosol 10 minutes prior to bradykinin challenge) prevented bradykinin-evoked coughing (0.1–10 mg·ml⁻¹, delivered cumulatively as aerosols; *P < 0.05). The results are presented as the mean ± S.E.M. cumulative coughs evoked in 4–9 experiments.

As shown previously (Canning et al., 2004; Smith et al., 2012), bradykinin-induced coughing occurred in repetitive, paroxysmal patterns, often with few or no tidal breaths separating each cough effort. This occasionally produced expiration reflexes in lieu of coughing, which occurred when the animals had initiated a cough before much (if any) inspiratory efforts had been completed at the end of the preceding cough. Initial, subthreshold challenges with bradykinin appeared to sensitize the airways to higher concentrations of the kinin. Thus, as illustrated in Fig. 2, when delivered as a single challenge without preceding or subsequent challenges, 3 mg·ml⁻¹ bradykinin was typically subthreshold for initiating cough; however, in animals first challenged with 1 mg·ml⁻¹ bradykinin, 3 mg·ml⁻¹ was the optimal dose for cough challenge (Fig. 2).

Fig. 2.

Bradykinin both sensitizes and desensitizes its ability to evoke coughing in awake guinea pigs. The peptide was delivered as an aerosol either as single doses to individual animals or in cumulatively increasing concentrations, with 5 minutes in between each challenge dose. When administered as a single dose, only concentrations of 5 mg·ml⁻¹ (7 ± 3 coughs; n = 14; data not shown) or 10 mg·ml⁻¹ reliably evoked coughing. However, when administered in cumulatively increasing concentrations, 3 mg·ml⁻¹ bradykinin was optimal for evoking cough. Increasing the concentration of bradykinin from 3 to 10 mg·ml⁻¹ evoked only one cough in one out of eight animals studied. An asterisk (*) indicates that the number of coughs evoked by bradykinin was significantly less than the number of coughs evoked by 3 mg·ml⁻¹ bradykinin in the cumulative response curves (P < 0.05). These results are the mean ± S.E.M. of 4–20 experiments. The data for coughing evoked by cumulatively administered bradykinin were regraphed from Fig. 1B, showing only the cough responses evoked by each dose studied and not the cumulative responses.

Cough responses to bradykinin quickly desensitized. In animals challenged with cumulatively increasing doses, 3 mg·ml⁻¹ bradykinin evoked coughs (9.0 ± 2.5), while subsequent challenge (just 5 minutes later) with 10 mg·ml⁻¹ bradykinin resulted in one cough in just one out of eight animals studied (0.1 ± 0.1 coughs overall; n = 8). Desensitization could also be seen by quantifying cough responses after the initial paroxysm of cough evoked by the peptide. In the nine (out of 20) vehicle-treated animals challenged with 10 mg·ml⁻¹ bradykinin that had paroxysmal bouts of ≥10 coughs in any 30-second interval of the 10-minute challenge (16.8 ± 2.6 coughs in ≤30 seconds; n = 9), just 1.4 ± 0.6 coughs occurred over the ensuing 5.1 ± 0.7 minutes of continuous bradykinin challenge. Similar results were seen in animals in all other treatment groups challenged with 10 mg·ml⁻¹ bradykinin (15.0 ± 0.5 coughs in ≤30 seconds of paroxysmal coughing, 1.8 ± 0.5 coughs over the ensuing 5.1 ± 0.6 minutes of challenge; n = 21) and in animals provoked with 5 mg·ml⁻¹ bradykinin (15.6 ± 1.7 coughs in ≤30 seconds of paroxysmal coughing, 1.6 ± 0.5 coughs over the ensuing 4.9 ± 0.8 minutes of challenge; n = 7). Overall, only 12 out of 107 animals studied had multiple bouts of 10 or more coughs in any 30-second intervals during bradykinin challenge, with eight out of the 12 additional paroxysms occurring less than 2 minutes after the preceding bout, suggestive of a continuation/second phase of an ongoing response.

Pretreatment with thiorphan or captopril alone (1 mg·ml⁻¹ each delivered as aerosols) did not evoke cough or alter cough responsiveness to 1 mg·ml⁻¹ bradykinin challenges. However, when administered in combination captopril and thiorphan markedly potentiated the cough evoked by bradykinin (Fig. 3). This combination of peptidase inhibitors decreased the time to first cough (1.07 ± 0.01 minutes after initiating a 1 mg·ml⁻¹ bradykinin challenge with peptidase pretreatment versus 3.9 ± 0.8 minutes after initiating a 10 mg·ml⁻¹ bradykinin challenge in the absence of peptidase inhibitors; n ≥ 15; P < 0.05). However, the paroxysmal pattern of coughing was unchanged and desensitization was still apparent, with no coughing observed over an average of the last 4.3 ± 0.7 minutes of the 10-minute, 1 mg·ml⁻¹ bradykinin challenge.

Fig. 3.

Peptidases regulate bradykinin-evoked cough in awake guinea pigs. Bradykinin (1 mg·ml⁻¹) was administered as an aerosol after aerosol pretreatment with vehicle (saline), the neutral endopeptidase inhibitor thiorphan (1 mg·ml⁻¹), the ACE inhibitor captopril (1 mg·ml⁻¹), or the combination of thiorphan and captopril (1 mg·ml⁻¹ each). The drugs administered alone had little effect on 1 mg·ml⁻¹ bradykinin-evoked coughing, but when administered in combination, the cough responses were markedly potentiated. However, even following peptidase inhibition, a lower dose of bradykinin (0.1 mg·ml⁻¹) was still largely ineffective in evoking cough (median = 1; n = 14). The data are presented as a mean ± S.E.M. of 12–21 experiments. An asterisk (*) indicates a statistically significant potentiation of bradykinin-evoked cough relative to vehicle-treated animals (P < 0.05).

Bradykinin-Evoked Bronchospasm Does Not Initiate Coughing.

Bradykinin-induced bronchospasm occurs indirectly and is thought to result from the net effects of dilating nitric oxide and constricting thromboxanes (Arakawa et al., 1992; Ricciardolo et al., 1994b, 1996, 1999; Figini et al., 1996; Canning et al., 2001; Keir et al., 2015). We have thus evaluated the effects of a thromboxane receptor antagonist (ICI 192605) and a nitric oxide synthase inhibitor (L-NNA) on the cough responses evoked by bradykinin. Using a pretreating dose and delivery scheme (0.1 mM aerosol for 10 minutes), identical to that used previously to enhance bradykinin-induced bronchospasm (Ricciardolo et al., 1994b), we found that L-NNA did not significantly modify bradykinin-induced coughing (Fig. 4). We then confirmed the adequacy of dosing of ICI 192605 (10 μM delivered as an aerosol) by showing that this thromboxane receptor antagonist prevented the enhancement of citric acid–induced coughing evoked by the thromboxane receptor agonist U46619 (1 μM delivered as an aerosol 30 minutes prior to cough challenge) (Xiang et al., 2002), with citric acid (0.01–0.1M) evoking 2 ± 1, 25 ± 6, and 6 ± 3 cumulative coughs in animals pretreated with vehicle, U46619, or U46619 with ICI 192605 pretreatment, respectively (n = 8–13; P < 0.05 for control versus U46619). However, aerosolized ICI 192605 failed to significantly inhibit bradykinin-evoked cough (15 ± 6 and 11 ± 5 coughs in control and ICI 192605 pretreated animals, respectively; n = 5 to 6/treatment group; P > 0.1). Moreover, at twice the dose (0.5 mg·kg⁻¹) that completely abolished bradykinin-induced increase (0.1–2 nmol·kg⁻¹ i.v.) in PIP in anesthetized, mechanically ventilated guinea pigs, intraperitoneally administered ICI 192605 (1 mg·kg⁻¹) failed to significantly inhibit bradykinin-evoked cough (Fig. 4). Incidentally, the thromboxane receptor agonist U46619 failed to directly evoke coughing upon aerosol administration but did induce labored breathing, suggestive of bronchoconstriction, which was prevented by aerosolized ICI 192605.

Fig. 4.

Bradykinin-induced bronchospasm does not influence the number of coughs evoked by bradykinin. (A) The thromboxane receptor antagonist ICI 192605 (0.5 mg·ml⁻¹ i.v.; n = 3) abolishes bradykinin-evoked bronchoconstriction (measured as an increase in PIP) in anesthetized guinea pigs (*: P < 0.05). (B) In contrast to bradykinin-evoked bronchospasm, 1 mg·ml⁻¹ bradykinin-evoked coughing in awake guinea pigs was not inhibited by prior pretreatment with ICI 192605 (1 mg·ml⁻¹ i.p.; n = 8). Similarly, administered in a way shown previously to enhance bradykinin-evoked bronchospasm (0.1 mM aerosol for 10 minutes) (see Ricciardolo et al., 1994b), the nitric oxide synthase inhibitor L-NNA (n = 8) also failed to alter the number of coughs evoked by bradykinin. The results are presented as a mean ± S.E.M. of 3–15 experiments. Bradykinin challenges were delivered 5 minutes after aerosol pretreatments with the peptidase inhibitors captopril and thiorphan (1 mg·ml⁻¹ each, 10-minute aerosols). Control animals received either the vehicle for ICI 192605 or the vehicle for L-NNA. The cough responses in these two control groups were similar and the data were therefore pooled.

Prostanoids in addition to thromboxane may regulate the respiratory reflexes evoked by bradykinin (Canning et al., 2001; Chou et al., 2008). Parasympathetic-cholinergic nerves and NKs released in the airways through axonal reflexes also play a role in responses to bradykinin (Fuller et al., 1987; Nakajima et al., 1994; Canning et al., 2001). Their roles in the cough responses evoked by bradykinin were evaluated in the present study. Neither meclofenamic acid nor atropine reduced the number of bradykinin-evoked coughs. However, atropine did change the time course of the coughing evoked. Thus, in control animals the time elapsed until ≥50% of the total number of coughs occurred averaged 7.0 ± 0.5 minutes (median = 7.25 minutes; range: 2.5–10 minutes; n = 16). This was significantly reduced by atropine (3.4 ± 0.7 minutes; n = 5; P < 0.05) but not by meclofenamic acid (5.6 ± 1.2 minutes; n = 5). The combination of meclofenamic acid and atropine still failed to produce a statistically significant inhibition of cough evoked by 10 mg·ml⁻¹ bradykinin (8 ± 4 coughs; n = 10; P > 0.05) (compare with the results in Fig. 5); however, as with atropine alone it reduced the time elapsed until ≥50% of the total number of coughs occurred (3.5 ± 0.6 minutes; n = 4; P < 0.05).

Fig. 5.

NK receptor antagonists (but not atropine or meclofenamic acid) reduced the number of coughs evoked by aerosolized bradykinin (10 mg·ml⁻¹) in awake guinea pigs. Thirty minutes after administration of vehicle (i.p.; n = 20), atropine (1 mg·ml⁻¹ aerosol; n = 7), meclofenamic acid (1 mg·kg⁻¹ i.p.; n = 7), or the combination of SR140333, SR48968, and SB223412 (3 mg·kg⁻¹ each i.p.; n = 5), bradykinin was delivered as an aerosol for 10 minutes and the total number of coughs evoked was counted. Similar to the NK receptor antagonists administered alone, coadministering NK receptor antagonists (CP99994, SR48968, and SB223412; 3 mg·kg⁻¹ each i.p.) along with either meclofenamic acid (1 mg·kg⁻¹ i.p.; n = 8) or ICI 192605 (1 mg·kg⁻¹ i.p.; n = 5) also markedly inhibited the bradykinin-evoked coughing (4 ± 2 and 4 ± 2 coughs, respectively; P < 0.05). The vehicle for atropine aerosol delivery (saline) was without effect on bradykinin-evoked coughing (14 ± 4 coughs; data not shown; n = 7). Some animals pretreated with the i.p. vehicle (4/20), the vehicle for atropine aerosol (2/7), meclofenamic acid (2/7), atropine (2/7), or the NK receptor antagonists (4/5) coughed once or not at all in relation to the bradykinin challenges. These animals were still included in the mean data, which are presented as a mean ± S.E.M. of 5–20 experiments. An asterisk (*) indicates that the NK receptor antagonists significantly inhibited bradykinin-evoked coughing relative to vehicle control (P < 0.05).

In contrast to the effects of atropine and meclofenamic acid, a combination of NK₁, NK₂, and NK₃ receptor antagonists (SR140333, SR48968, and SB223412, respectively; 3 mg·kg⁻¹ each, given i.p.) markedly inhibited bradykinin-evoked cough (Fig. 5). However, when administered directly to the airways by aerosol (1 mg·ml⁻¹ each) a combination of NK₁ [SR140333 (n = 3) or CP99994 (n = 2)] and NK₂ (SR48968) receptor antagonists was without effect on bradykinin-evoked cough (13 ± 5 coughs; n = 5) (compare with the results in Fig. 5).

Evidence that Bronchopulmonary C-fibers Regulate Bradykinin-Evoked Coughing.

Bradykinin is thought to be a relatively selective stimulant of C-fibers; however, bradykinin can activate other airway afferent nerves in guinea pigs, including rapidly adapting receptors (RARs) and a poorly defined subset of capsaicin-sensitive myelinated afferents arising from the jugular ganglia (Bergren, 1997; Kajekar et al., 1999). We have quantified the effects of bradykinin on airway vagal afferent nerve subtypes using a Ca⁺⁺ influx assay in retrograde-labeled neurons with 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate injected into the larynx, trachea, and main stem bronchi. Among the labeled afferent neurons recovered from the vagal sensory ganglia, only those taken from the jugular ganglia were activated by bradykinin (Fig. 6).

Fig. 6.

Bradykinin fails to evoke intracellular Ca⁺⁺ increases in laryngeal/tracheal/bronchial vagal sensory neurons arising from the nodose ganglia (a–c) but does activate airway afferent neurons arising from the jugular ganglia (d–f). Neurons were visualized in bright-field microscopy (a and d) and by fluorescent microscopy at baseline (b and e) and at the peak of their responses to 1 μM bradykinin (c and f). Images in the upper and lower panels are from the same neurons visualized under differing conditions. Red and yellow coloring in these neurons indicates higher concentrations of Ca⁺⁺ in comparison with the green fluorescence at baseline. Retrograde neuronal tracing with 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate was used to identify vagal sensory neurons projecting to the larynx, trachea, and main stem bronchi. Neurons were dissociated from the ganglia, cultured on coverslips, and loaded with Fura for measurements of Ca⁺⁺ influx. All 24 retrograde-labeled jugular ganglia neurons studied responded to 1 μM bradykinin challenge (Ca⁺⁺ influx ≥10% of the response to ionomycin), while none of the six labeled nodose ganglia neurons responded (even though many of the 63 nonlabeled nodose ganglia neurons studied were activated by bradykinin). The micrographs shown are representative of experiments performed on neurons recovered from eight guinea pigs.

The sensitivity of bradykinin-evoked coughing to blocking all three NK receptors with combinations of SR140333, SR48968, and SB223412 or CP99994, SR48968, and SB223412 (Fig. 5) is consistent with a role for jugular C-fibers in cough (Ricco et al., 1996). We have further characterized the specific NK receptors involved in these responses. The combinations of only NK₁ and NK₂ (CP99994 and SR48968) or NK₁ and NK₃ (CP99994 and SB223412) receptor antagonists were without significant effect on bradykinin-evoked coughing (Table 1). However, in contrast to the combinations of NK₁ and NK₂ and NK₁ and NK₃ receptor antagonists, which failed to modify responses to bradykinin, a combination of NK₂ (SR48968) and NK₃ (SR142801) receptor antagonists markedly inhibited bradykinin-evoked coughing, approximating the effects of blocking all three NK receptors. The NK₃ receptor antagonist SR142801 administered alone did not significantly inhibit bradykinin-evoked coughing [15.2 ± 7.3 and 13.0 ± 3.5 coughs in matched control and SR142801 (3 mg/kg) treated animals, respectively; n = 4 to 5; P > 0.1]. Respiratory rate was unchanged by these compounds (Table 1).

TABLE 1.

Effect of NK receptor antagonists on bradykinin-evoked coughing in awake guinea pigs

NK receptor antagonists were administered simultaneously at a dose of 3 mg·kg⁻¹ by i.p. injection. Coughing was evoked by aerosol challenges with 1 mg·ml⁻¹ bradykinin 30 minutes subsequent to drug pretreatments. Basal respiratory rates were measured just prior to the bradykinin challenges. Bradykinin challenges were delivered 5 minutes after aerosol pretreatments with the peptidase inhibitors captopril and thiorphan (1 mg·ml⁻¹ each, 10-minute aerosols). The results are presented as the mean ± S.E.M. of 8–15 nonpaired experiments. An asterisk (*) indicates that the combination of SR48968 and SR142801 inhibited bradykinin-evoked coughing (P < 0.05).

Treatment	Total Number of Coughs	Basal Respiratory Rate
Vehicle Control	21 ± 3	113 ± 4
CP99994 and SR48968	17 ± 3	113 ± 4
CP99994 and SB223412	22 ± 6	108 ± 8
SR48968 and SR142801	6 ± 3*	115 ± 8

The inhibitory effects of the NK receptor antagonists studied here were selective for bradykinin-evoked cough. Single, explosive coughing events evoked by mechanical stimulation of the airway mucosa in anesthetized guinea pigs—reflexes we attribute to activation of the bradykinin-insensitive nodose Aδ fibers innervating the trachea, larynx, and main stem bronchi (Mazzone et al., 2009; Muroi et al., 2013)—were still present in all animals treated with a combination of the NK₁, NK₂, and NK₃ receptor antagonists (SR140333, SR48968, and SB223412, respectively; 3 mg·kg⁻¹ each, i.p.; n = 3) or the vehicle for these antagonists (n = 3).

Discussion

We have described the physiologic and pharmacologic basis of bradykinin-induced cough in guinea pigs. Similar to many other effects of bradykinin in the lung, we anticipated that the indirect effects of this inflammatory peptide on the structural cells of the airways would contribute to the coughing we observed upon aerosol challenge. For example, the bronchospasm induced by bradykinin occurs via indirect mechanisms, either secondary to thromboxane formation (perhaps from platelets) or by reflex activation of airway parasympathetic-cholinergic nerves (Fuller et al., 1987; Arakawa et al., 1992; Hulsmann et al., 1994; Arvidsson et al., 2001; Canning et al., 2001; Keir et al., 2015). NK receptor–dependent axonal reflexes may also contribute to the airway responses evoked by bradykinin in guinea pigs (Nakajima et al., 1994; Joad et al., 1997). However, we observed that drugs that prevent bradykinin-induced bronchospasm did not prevent bradykinin-induced coughing, while drugs that would be expected to enhance bradykinin-induced bronchospasm (e.g., captopril, L-NNA, and thiorphan) (Ichinose and Barnes, 1990; Ricciardolo et al., 1994b) did not, on their own, enhance bradykinin-evoked coughing.

Cough resulting from bradykinin challenge likely depends upon the direct effects of the inflammatory peptide on bronchopulmonary C-fibers (Kaufman et al.; 1980; Bergren, 1997; Kajekar et al., 1999). Bradykinin can also activate RARs, perhaps due to its direct physiologic effects (e.g., bronchospasm, vascular engorgement, vascular leakage, and mucus secretion), and there is also a bradykinin-sensitive, capsaicin-sensitive, myelinated afferent nerve subtype innervating the extrapulmonary airways of guinea pigs (Kaufman et al.; 1980; Bergren, 1997; Kajekar et al., 1999). However, RAR activation by bradykinin is largely abolished by cyclooxygenase inhibition or by isoproterenol (Bergren, 1997; Canning et al., 2001), and neither meclofenamic acid nor the thromboxane receptor antagonist ICI 192605 (both of which abolish bradykinin-evoked bronchospasm) inhibited bradykinin-induced cough. Also arguing against a role for RARs in bradykinin-evoked cough is the observation that thromboxane inhalation fails to evoke coughing in guinea pigs (present study; Shinagawa et al., 2000; Xiang et al., 2002), although prostaglandin E₂, via EP3 receptor activation and prostaglandin D₂ via prostaglandin D₂ receptor 1 (DP1) activation, can induce coughing in guinea pigs and humans (Choudry et al., 1989; Maher et al., 2009, 2015).

The effects of the NK receptor–selective antagonists on bradykinin-evoked cough and the results of previous studies argue in favor of jugular C-fiber involvement and a central site of action for these agents in cough suppression (Bolser et al., 1997; Canning et al., 2001; Mazzone and Canning, 2002; Mazzone et al., 2005). The primary local effects of bradykinin occur through parasympathetic-cholinergic reflex bronchospasm and mucus secretion, prostanoid formation, and perhaps NK₁ receptor–dependent vasodilation and plasma exudation initiated by axonal reflexes (Arakawa et al., 1992; Bertrand et al., 1993; Nakajima et al., 1994; Joad et al., 1997; Canning et al., 2001). However, neither cyclooxygenase inhibition nor atropine prevented bradykinin-evoked cough. NK₁ receptor antagonism also failed to modulate bradykinin-evoked cough when given in combination with either NK₂ or NK₃ receptor antagonists. Only when both NK₂ and NK₃ receptors were blocked was bradykinin-evoked cough inhibited. Indeed, in four separate sets of experiments, where both NK₂ and NK₃ receptor antagonists were administered and two structurally unrelated NK₃ receptor antagonists were used, bradykinin-evoked coughing was markedly reduced relative to that observed in matched control animals. All NK receptor subtypes have been localized to mammalian brainstem (Geraghty and Mazzone, 2003). There is precedence for NK₂ and NK₃ receptor–dependent cough in guinea pigs, as well as precedence for a lack of effect of NK₁ receptor antagonists in some (but not all) studies of cough in guinea pigs (Advenier et al., 1993; Bolser et al., 1997; Emonds-Alt et al., 1997, 2002; Daoui et al., 1998; Hay et al., 2002; El-Hashim et al., 2004). The involvement of NKs also supports our previous assertions of jugular C-fiber involvement and against a role for either jugular Aδ-fibers or nodose C-fibers (Ricco et al., 1996; Undem et al., 2004; Muroi et al., 2013).

Despite the profound antitussive effects of the NK receptor antagonists reported here and elsewhere, their promise as cough suppressants in patients remain unclear. In favor of their utility in cough and in controlling other aberrant respiratory reflexes and sensations, we and others have shown previously that, in guinea pigs, NK receptor antagonists inhibit capsaicin-evoked and citric acid–evoked coughing, reflex bronchospasm evoked by tracheal/laryngeal C-fiber activation, and the reflex bronchospasm evoked by bradykinin following cyclooxygenase inhibition (Girard et al., 1995; Bolser et al., 1997; Canning et al., 2001; Mazzone and Canning, 2002; El-Hashim et al., 2004). NK receptor antagonists also inhibit coughing evoked in rabbits, cats, and dogs (reviewed in Canning, 2009), and a recent preliminary report suggests modest cough suppression in lung cancer patients by the NK₁ receptor antagonist aprepitant (Harle et al., 2015). However, we have also described how NK receptor antagonists are without effect on coughing evoked by acid or mechanical stimulation of the airways of anesthetized guinea pigs, or the reflex bronchospasm evoked by histamine in anesthetized guinea pigs (present study; Canning et al., 2001; Mazzone et al., 2005), each of which being reflexes that are unlikely to depend upon C-fiber activation. The utility of these agents in human cough will thus depend upon the predictive value of studies performed in animals but also on the relative contribution of afferent nerve subtypes to cough in human disease.

We observed both sensitization and desensitization of the cough responses to bradykinin. The desensitization to bradykinin-induced cough (and bradykinin-induced bronchospasm) (see Fig. 4A) is consistent with previous studies and likely reflects the rapid and pronounced desensitization of B₂ receptors with sustained receptor occupancy (Wolsing and Rosenbaum, 1993; Leeb-Lundberg et al., 2005; Zimmerman et al., 2011). However, we cannot rule out other possible inhibitory mechanisms, including local inhibitory effects induced by autacoids formed in response to bradykinin, adaptation at the central synapses for C-fibers, or perhaps negative feedback regulation of the cough reflex. The sensitization of cough apparent when comparing the responses to bradykinin-evoked cough with or without preceding subthreshold challenge doses has not been previously described. However, a sensitizing effect of subthreshold doses of bradykinin on subsequently evoked cough responses has been noted (Fox et al., 1996; El-Hashim and Amine, 2005; Mazzone et al., 2005). These sensitizing effects may be both central nervous system and NK₁ receptor dependent (Joad et al., 2004; Mazzone et al., 2005; Canning and Mori, 2011; Cinelli et al., 2015) or could depend on local actions of eicosanoids released from structural cells following bradykinin B₂ receptor activation (Salari and Chan-Yeung, 1989; Arakawa et al., 1992; Gatti et al., 2006; Petho and Reeh, 2012).

The paroxysmal pattern of coughing evoked by bradykinin is unique to this peptide and characteristic of certain pathologies, leading to the speculation that conditions that enhance bradykinin actions or slow the rate of bradykinin receptor desensitization may result in bradykinin-dependent cough (Fox et al., 1996; Morice et al., 1997; Dicpinigaitis, 2006; Hewitt and Canning, 2010; Mutolo et al., 2010). It is also interesting that captopril alone was unable to enhance bradykinin-evoked coughing in this study. Only when the ACE and neutral endopeptidase were both inhibited was the cough response enhanced. This result suggests a redundant effect of these enzymes in limiting the actions of bradykinin and may also suggest that the small subset of patients that cough when on ACE inhibitor therapy have an additional decrease/defect in neutral endopeptidase activity that predisposes them to coughing evoked by bradykinin (Dicpinigaitis, 2006; Mahmoudpour et al., 2013).

Implications for Cough in Disease.

The results of our study suggest that neither bradykinin-evoked bronchospasm nor eicosanoid formation within the airways, nor even central nervous system activation of NK₁ receptors, are necessary for the initiation of cough evoked by this peptide. However, this does not rule out modulatory roles for these secondary effects of bradykinin in cough (Malini et al., 1997; Shinagawa et al., 2000; Liu et al., 2001; Xiang et al., 2002; Mazzone et al., 2005; Gatti et al., 2006; Canning and Mori, 2011). Thus, we saw that atropine altered the kinetics of bradykinin-evoked cough and thromboxane receptor activation enhances cough responsiveness, and bradykinin certainly induces thromboxane release and receptor activation in the airways. We speculate that the effects of atropine may highlight the importance of mucosal barrier function and clearance of inhaled irritants on cough responses, while autacoids such as thromboxane may act directly on the afferent nerve terminals to enhance their excitability. Precisely how the many secondary effects of bradykinin might modulate cough awaits further study. The primary focus of the present study was defining how bradykinin directly evokes this airway defensive reflex.

The most logical approach to preventing the actions of bradykinin in the lungs is B₂ receptor antagonism. It may also be possible to inhibit bradykinin-evoked cough by preventing its effects on the ion channels TRPA1 and TRPV1 (Shin et al., 2002; Carr et al., 2003; Bandell et al., 2004; Kollarik and Undem, 2004; Lee et al., 2005; Grace et al., 2012). Other approaches targeting the central and peripheral terminals of bradykinin-sensitive C-fibers, preventing bradykinin formation (e.g., kallikrein inhibitors), hastening its degradation (e.g., soluble neutral endopeptidase), or promoting bradykinin B₂ receptor desensitization or downregulation may also prevent bradykinin-evoked coughing (Joseph and Kaplan, 2005; Leeb-Lundberg et al., 2005; Zimmerman et al., 2011; Smith et al., 2012). No matter the approach, we speculate that targeting bradykinin may prevent the coughing associated with several respiratory diseases.

Abbreviations

ACE

angiotensin-converting enzyme

L-NNA

N^G-nitro-L-arginine

NK

neurokinin

PIP

pulmonary inflation pressure

RAR

rapidly adapting receptor

Authorship Contributions

Participated in research design: Canning.

Conducted experiments: Hewitt, Adams, Mori, Mazzone, Yu.

Performed data analysis: Hewitt, Adams, Mori, Mazzone, Yu, Canning.

Wrote or contributed to the writing of the manuscript: Hewitt, Adams, Mori, Mazzone, Yu, Canning.