THE INFLUENCE OF CO₂ ON SPATIOTEMPORAL FEATURES OF MECHANICALLY INDUCED COUGH IN ANESTHETIZED CATS

Abstract

Effective cough requires a significant increase in lung volume used to produce the shear forces on the airway to clear aspirated material. This increase in tidal volume during cough, along with an increase in tidal frequency during bouts of paroxysmal cough produces profound hyperventilation and thus reduces arterial CO_2. While there are several reports in the literature regarding the effects of hypercapnia, hyperoxia, and hypoxia on cough, there is little research quantifying the effects of hypocapnia on the cough reflex. We hypothesized that decreased CO₂ would enhance coughing. In 12 spontaneously breathing adult male cats, we compared bouts of prolonged mechanically stimulated cough, in which cough induced hyperventilation (CHV) was allowed to occur, with isocapnic cough trials where we maintained eupneic end-tidal CO₂ by adding CO₂ to the inspired gas. Isocapnia slightly increased cough number and decreased esophageal pressures with no change in EMG magnitudes or phase durations. The cough-to-eupnea transition was also analyzed between CHV, isocapnia, and a third group of animals that were mechanically hyperventilated to apnea. The transition to eupnea was highly sensitive to added CO₂, and CHV apneas were much shorter than those produced by mechanical hyperventilation. We suggest that the cough pattern generator is relatively insensitive to CHV. In the immediate post-cough period, the appearance of breathing while CO₂ is very low suggests a transient reduction in apneic threshold following a paroxysmal cough bout.

1. Introduction

Cough is a vital reflex for maintaining a functional respiratory system. Cough and breathing share several neuroanatomical substrates in the brainstem (Bolser, 2009; Bolser and Davenport, 2002; Shannon et al., 2004, 1996). The genesis of each behavior uses a shared set of respiratory muscles (Fontana and Lavorini, 2006; Leith, 1968; MacKlem, 1974; van Lunteren et al., 1989), but each behavior is regulated differentially by the respiratory control network (RCN) (Bolser et al., 1999; Bolser and Davenport, 2002; Chakravarty et al., 1956; Widdicombe and Singh, 2017). While breathing is primarily tasked with obtaining oxygen and maintaining homeostatic levels of the partial pressure of arterial CO₂ (PaCO₂), cough is tasked with defending the ability of the airway to maintain this homeostasis, as cough generates shear forces on the airway lumen to expel materials that interfere with gas exchange (Leith, 1968; MacKlem, 1974). The RCN must balance and prioritize airway protective behaviors such as cough with its baseline homeostatic functions, and studies of the underlying mechanisms of this management are ongoing.

Bouts of repetitive cough can produce profound hyperventilation, reducing arterial CO₂ well below eupneic levels (Tomori et al., 1973). The influence of this hyperventilation may be particularly important for patients suffering from intractable cough, particularly in the case of cough syncope. Cough syncope is a rare condition and its etiology contested (Dicpinigaitis et al., 2014). One proposed mechanism for cough syncope is hypoperfusion of the brain, which can be caused by elevated intrathoracic pressure or decreased arterial CO₂, both of which occur during bouts of paroxysmal cough (Dicpinigaitis et al., 2014; Kety and Schmidt, 1946; Sharpey-Schafer, 1953a; Tomori et al., 1973). It is unclear why some individuals are prone to cough syncope while others are not. It is possible that these patients have pathological responses to the hyperventilation, contributing to dysregulation of cerebral blood flow.

A primary role of the RCN is sensing and responding to changes in CO₂ (chemosensation). In its simplest form, maintenance of arterial CO₂ levels is controlled by a negative feedback process (Haldane and Priestley, 1905). While the actual mechanisms are more complicated than a simple feedback loop, the main effects of CO₂ on the respiratory pattern are profound and were reported over a century ago (Campbell et al., 1913; Douglas and Haldane, 1909; Haldane and Priestley, 1905). The activity of several brainstem regions likely underlies the central CO₂ response, including the retrotrapezoid nucleus/parafacial respiratory group (Erlichman et al., 2008; Kumar et al., 2015), midline raphe (Bernard et al., 1996; Silva et al., 2020; Smith et al., 2018), locus coeruleus (Hartzler et al., 2008; Nichols et al., 2008), nucleus of the solitary tract (Nichols et al., 2008), and cerebellum (Wang and Richerson, 2000; Xu et al., 2001, 1995). There is also a significant overlap with these loci and those implicated in the neurogenesis of cough (Jakus et al., 2008; Xu et al., 1997). The mechanisms by which these sites integrate chemosensory and tussigenic afferent information is unknown. Cough and breathing are very different behaviors, and the differential regulation necessary to generate their distinctive motor outputs complicates the interaction that is required for the central control areas to generate and modulate respiratory muscle motor output.

While cough has been investigated under the combination of varying levels of O₂ and CO₂ (Hanacek et al., 2006; Nishino et al., 1998, 1989; Tatár et al., 1987), the effects of hypocapnia (reduced CO₂) on the cough reflex in the absence of hypoxia are unknown. This is a particularly interesting gap in our understanding of cough reflex modulation as paroxysmal cough can result in a hypocapnic state evidenced by Tomori (1973) as mentioned above. This cough-induced hyperventilation (CHV) is due to large increases in tidal volume and thus alveolar ventilation (Bolser and Davenport, 2000). Furthermore, CHV presents a profound challenge for the RCN in returning to eupneic breathing after a bout of paroxysmal cough. As mentioned above, the RCN is highly sensitive to changes in CO2. This sensitivity can be particularly appreciated in sleeping or anesthetized subjects. In sleeping subjects, slight decreases in arterial CO2 (2–5 mmHg) result in a cessation of rhythmic ventilation and after some time, a ventilatory overshoot upon return to rhythmic ventilation (Dempsey, 2005; Skatrud and Dempsey, 1983; Xie et al., 2001). While this apneic threshold is not a fixed value and depending on the state of the animal, the decrease in arterial CO2 needed produce apnea can be slightly increased or decreased (Dempsey, 2005; Dempsey et al., 2004), CHV can cause arterial CO2 levels to fall many times below those needed to induce apnea.

The present investigation sought to understand the regulation of the cough reflex by CO₂ during a cough-induced hyperventilation protocol. We hypothesized that the reduced CO₂ that results from hyperventilation would enhance repetitive cough when compared to an isocapnic condition that we maintained by adding CO₂ to the inspired gas. We also sought to quantify how hyperventilation induced by cough or by mechanical ventilation affects the transition between cough and eupnea.

2. Methods

2.1. Animals

Experiments were performed on 12 spontaneously breathing adult male cats with an average age of 1.05 years (SD = 0.14 years) and weight of 4.95 kg (SD = 0.53 kg). The protocol was approved by the University of Florida Institutional Animal Care and Use Committee (IACUC) and all procedures were compliant with the Guide for the Care and Use of Laboratory Animals. Anesthesia was initially induced with sevoflurane (4.5%), then the animals were weaned onto sodium pentobarbital (25 mg/kg, i.v.). Based on forelimb withdrawal reflex and jaw tone, supplemental doses were given as needed in 0.1–0.3 mL i.v. boluses. A single dose of atropine sulfate (0.054 mg/kg, i.v.) was given at the beginning of surgery to minimize airway secretions. The femoral vein and artery were cannulated to deliver drugs and record blood pressure, respectively. Arterial blood pressure and end-tidal CO₂ (ETCO₂) were continuously recorded, and arterial blood gas samples were obtained every hour. Body temperature was maintained at 37.5 ± 0.5 °C with a TC-1000 homeothermic pad and rectal temperature probe (CWE Inc.; United States).

2.2. Surgical Procedure and EMG Placement

Animals were tracheostomized and a bifurcated tracheostomy tube was placed to permit simultaneous end-tidal gas recording, manipulation of inspired gases, and mechanical stimulation of the tracheobronchial airway to elicit coughing (Fig 1A). End-tidal gas was measured via a GEMINI O2 & CO2 Monitor (CWE Inc., Ardmore, PA). Electromyograms (EMGs) were recorded using bipolar fine wire electrodes placed intramuscularly to record respiratory related behaviors using the method of Basmajian and Stecko (1962). EMGs from the sternocostal diaphragm (dorsal to the xiphoid process), bilateral parasternal, bilateral internal abdominal oblique, thyroarytenoid, posterior cricoarytenoid (PCA), geniohyoid, mylohyoid, thyropharyngeus and the upper esophageal sphincter were recorded to classify respiratory behaviors and to evaluate breathing and cough. EMG signals were amplified using Grass P511 amplifiers and band-pass filtered (300–10,000 Hz) and subsequently digitized with a Power 1401 mk II and Spike 2 version 9 (Cambridge Electronic Design; United Kingdom). EMG channels were rectified and smoothed with a 50 ms time constant prior to analysis. A lead II electrocardiogram (ECG) was also placed on the animal (Einthoven, 1912), similarly amplified and digitized, with a band pass filter set to 3–3000Hz. Esophageal pressure was measured via a small balloon inserted through the mouth into the midthoracic esophagus. EMG and ECG signals were sampled at 25,000 Hz and pressure signals at 500 Hz.

Fig 1. Overview of surgical preparation, EMG placement, and cough measures.

A) A diagram illustrates EMG and tracheostomy setup. Parasternals and internal obliques were recorded bilaterally. For experiments without added CO₂, the end-tidal CO₂ (ETCO₂) line was connected to the Air/CO₂ port. Cough was mechanically stimulated with a flexible plastic cannula in the thoracic trachea. EMG placements were confirmed visually, corresponding to their activity during control cough and swallow trials. B) A representative cough trial shows cough phase duration and amplitude measurements. The top pairs of traces are integrated and raw electromyograms (EMGs), respectively, of respiratory muscles: the inspiratory diaphragm (DIA) and expiratory internal obliques (IO). Their pattern of activation generates an intrathoracic pressure that was measured by an esophageal balloon (ESPR). An end-tidal CO₂ (ETCO₂) recording provided additional information about respiratory state. Cardiovascular data were measured with a blood pressure (BP) transducer and electrocardiography (not shown).

2.3. Cough

Cough was stimulated via a flexible plastic cannula inserted into the tracheal tube. Mechanical stimulation of the intrathoracic trachea was accomplished by inserting and rotating this cannula at approximately 2 Hz. Prior to the experimental conditions, an adaptation protocol, consisting of 20 sequential 10–15 second mechanical stimulation trials separated by ~1 minute, was conducted (Kotmanova et al., 2018). Following this adaptation protocol, the cough response remains stable for several hours in this preparation.

Animals underwent both a cough-induced hyperventilation (CHV) condition and an experimentally-induced isocapnic condition. In the CHV trials, a cough stimulus was maintained for 60 seconds while the animal inspired room air. The isocapnic condition subjected the animals the same 60 second cough stimulus, however, after 10 seconds a 10% CO₂ gas mixture was introduced to the animal. This gas filled a reservoir bag with an outflow that was controlled with a one-way valve which was manually adjusted so that the end-tidal CO₂ of the animal approximated eupneic levels throughout the remainder of the cough stimulus. In both conditions, an arterial blood gas (ABG) sample was collected 45 seconds into the trial using an Element POC (Heska, Loveland, CO). Animals were allowed to recover at least 5 minutes between trials. Before a 60 second trial, two 10 second trials with a 60 second inter-trial interval were performed. The order of the condition was randomized, with 7 animals undergoing the CHV condition first.

Cough was confirmed by observing sequential and large ballistic-like EMG bursting patterns (Fig 1B) in inspiratory muscles, thyroarytenoid, and the internal abdominal obliques (Pitts et al., 2016; Poliacek et al., 2003; Wang et al., 2009), and rapidly rising inspiratory and expiratory esophageal pressure changes that were much larger than those observed during eupneic breathing (Fig 2A). The inspiratory (CT_I) phase of cough was defined as the period from the onset to the peak of activity of the diaphragm EMG. The active expiration component (CT_E1) began when the inspiratory phase ended and lasted until the ballistic activity on the internal abdominal oblique EMG subsided. The passive expiration (CT_E2) lasted from the end of the CT_E1 phase until the onset of the inspiratory phase of the following cough. No CT_E2 phase exists for the last cough in a paroxysmal cough bout, as there is no subsequent cough to define the end of this phase. The total cough cycle time (CT_TOT) was calculated by summating CT_I, CT_E1, and CT_E2. The cough bout duration was calculated by measuring the interval between the onset of the inspiratory phase (CT_I) of the first cough in a trial and the end of the active expiratory phase (CT_E1) of the last cough. For cough analysis in our experimental conditions, only coughs occurring in the second half of the cough bout were used. This ensured that coughs measured during this time were subject to a condition where both central and peripheral chemoreceptors were exposed to the change in blood gas, based on results reported in dogs by Smith et al. (2006). Additionally, by using only coughs occurring in the second half of the cough bout, we omitted the initial coughs during the period when changes in blood gas were more dynamic (Fig 2A), ensuring that measurements were obtained during the steady state. Fig 2A depicts the two conditions with relevant time points. A subset of 9 animals also had an arterial blood gas drawn approximately 45 seconds into the 60 second stimulus. Cough frequency was calculated by taking the number of coughs occurring in the second half of the cough bout and dividing by the summed total cycle time (CT_TOT) of these coughs.

Fig 2. The cough motor pattern was minimally affected by cough induced hyperventilation (CHV) while the cough-to-eupnea transition remained sensitive to changes in CO₂.

A) Esophageal pressure and end-tidal CO₂ (ETCO₂) recording traces of 60 second cough trials without and with added CO₂ added are shown from a single animal. Cough was stimulated for 60 seconds using a flexible plastic cannula inserted into the tracheostomy tube. In both conditions, an arterial blood gas (ABG) was recorded 45 seconds into the cough trial. During the control condition (top trace), no CO₂ was added. During the isocapnic condition (bottom trace), 10% CO₂ was added to the animal’s inspired gas 10s after the cough stimulus began; the flow was adjusted so that the end-tidal CO₂ matched levels seen during eupnea. B) Apnea duration was sensitive to the presence of cough and to CO₂ level. Recovery data from mechanical hyperventilation, cough-induced hyperventilation, and isocapnic repetitive coughing are shown. Animals were either mechanically hyperventilated to apnea (n = 5), exposed to cough-induced hyperventilation (CHV. n = 12), or to isocapnic cough conditions with added CO₂ (n = 12). Although both mechanical hyperventilation and CHV decrease CO₂, the apneic period after CHV was shorter than that following mechanically-induced hyperventilation. When CO₂ was added (isocapnia), the apnea was even shorter. C) The period of reduced breathing (hypopnea) before returning to eupneic-like breathing was influenced by arterial/end tidal CO₂ (isocapnia) but not affected by cough (CHV). ♦ = p < 0.05 (Mann-Whitney test). ## = p < 0.05 (Wilcoxon singed-rank test). ¤¤ = p < 0.01, ¤ p = < 0.05 (Welch t-test). CHV/Isocapnia n = 12, Ventilator n = 5.

2.4. Cardiovascular

Cardiovascular parameters were also measured during the 60 second cough trials. Of the 12 animals, cardiovascular data were analyzed for 6; the remaining 6 animals were excluded from this analysis because they had significant levels of artifact and noise on their ECG traces due to movement and strong thoracic muscle contractions occurring during coughing. Heart rate was measured using the interval between R wave peaks on the ECG (Einthoven, 1912). Systolic, diastolic, pulse, and mean arterial pressures were measured per beat. Mean arterial pressure was calculated by adding 1/3 of the pulse pressure to the diastolic pressure. The data collected during this period were aggregated for each condition in each animal, and like our method for analyzing cough, only data occurring in the second half of the trial were analyzed.

2.5. Recovery

Recovery from cough back to eupneic breathing was also investigated. Prior to each cough trial, a custom Spike2 script established a baseline level of activity on the integrated diaphragm EMG during a eupneic expiratory phase when the diaphragm is quiescent. The script detected when diaphragm EMG activity increased 5% from this baseline, and after visual confirmation, the crossing of this threshold defined our inspiratory period. We recorded both peak and area of the integrated diaphragm EMG during inspiration and averaged it over three eupneic breaths. After the last cough, our script would find two recovery time points. The first time point marked the end of any post-cough apnea, defined as the first inspiratory effort to occur after the end of the last CT_E1 phase of the last cough. We also measured a time point marking the end of a hypopneic period, defined as the first breath to generate an inspiratory integrated EMG burst that was at least 90% of the eupneic peak, including any post-cough apneas. To investigate whether the cause of hyperventilation had any influence, we mined data from prior experiments in our lab, and found a group of animals that were mechanically hyperventilated to apnea and maintained that that level of ventilation for 100 seconds. We compared recovery data from this separate group of 5 anesthetized and unparalyzed cats with the 12 animals mentioned earlier. Recovery from mechanical hyperventilation was measured in an identical fashion to our cough-induced hyperventilation and isocapnic trials.

2.6. Data Analysis

EMGs were normalized to the strongest amplitude recorded for that specific muscle during the CHV trial and represented as a percentage of that value. Data were then aggregated, resulting in a trial average for each measured parameter per condition. All data are reported as a mean and standard error unless otherwise stated. Differences between conditions were compared both with a visual assessment of their Q-Q plot and with a Shapiro-Wilk test to assess normality. Based on these outcomes, either a paired Student t-test or a Wilcoxon signed-rank test was used; alpha was set to 0.05. To analyze the effect of our perturbations on breathing recovery, three comparisons were made. Mechanically ventilated trials and cough-induced hyperventilation trials were compared with a Welch’s t-test. Mechanically ventilated trials and isocapnic trials were compared with a Mann-Whitney test. Cough-induced hyperventilation and isocapnic trials were compared with a Wilcoxon signed-rank test. The p-values from these tests were then corrected with a Holm-Bonferroni adjustment. To compare arterial (PaCO2 and alveolar (ETCO2) across our three conditions, a mixed effect model was fit using the lmerTest package (Kuznetsova et al., 2017) and a two-way repeated measures ANOVA table was generated using the Satterthwaite method to approximate degrees of freedom. Post-hoc analysis was performed using the emmeans (Lenth, 2020) package and p-values corrected with the Holm-Bonferroni adjustment. All statistical analysis and visualization was performed in R (R Core Team, 2020) with the RStudio IDE (RStudio Team, 2018) and the tidyverse package (Wickham et al., 2019).

3. Results

3.1. Cough bout features

Animals were subjected to two main conditions: a CHV condition, where cough-induced hyperventilation was not perturbed and the animal’s end-tidal CO₂ was allowed to fall without intervention; and an isocapnic condition, where CO₂ was added to inspired air such that end-tidal CO₂ during cough approximated eupneic levels.

Table 1 summarizes cough-related measurements from these trials. Initially, 600 coughs were measured across 24 trials (2 conditions per animal), but only coughs occurring in the last half of the cough bout were used for analysis, reducing the number of coughs to 265. These data were aggregated into two trial averages per measure based on the experimental condition. In the isocapnic condition, there was a 15% (p = 0.048) increase in cough number when compared to CHV. Cough frequency and cough bout duration did not change between the CHV and isocapnic conditions (p = 0.07 and p = 0.5, respectively).

Table 1.

Measures of cough during cough induced hyperventilation (CHV) and isocapnic trials.

	CHV	Isocapnia
Measure	Mean ± SE	Mean ± SE	Δ (%)	Test	T/Z-score	p -value
Cough bout features
Cough Number	10.3 ± 1.0	11.8 ± 1.1	15	P	−2.22	0.048 *
Cough Frequency (Hz)	0.34 ± 0.04	0.38 ± 0.04	13	P	−1.98	0.074
Bout Duration (sec)	61.7 ± 2.3	61.4 ± 1.2	0	W	−0.72	0.470
Amplitudes (% max CIHV)
Diaphragm	73.7 ± 3.0	73.5 ± 3.5	0	P	0.09	0.933
Parasternal Left	79.7 ± 2.6	78.3 ± 3.1	−2	P	0.94	0.367
Parasternal Right	75.4 ± 3.3	75.8 ± 4.2	1	W	−0.34	0.733
Internal Oblique	75.6 ± 2.8	73.8 ± 3.4	−2	P	0.71	0.490
Amplitudes (cmH2O)
Average ESPR	48.1 ± 9.2	41.7 ± 9.2	−13	P	2.33	0.040 *
Maximum ESPR	64.3 ± 10.6	56.1 ± 9.8	−13	W	−3.03	0.002 ##
Phase durations (ms)
CTI	1019 ± 75	958 ± 59	−6	P	1.78	0.103
CTE1	433 ± 24	444 ± 21	3	P	−0.63	0.540
CTE2	1854 ± 363	1400 ± 254	−24	W	−0.88	0.380
CTTOT	3307 ± 385	2802 ± 288	−15	W	−1.19	0.233

ESPR, esophageal pressure; CT_I Cough inspiration; CT_E1 cough active expiration; CT_E2 cough passive expiration; CT_TOT total cough cycle time.

Paired t-test (P)

^{*** =}p < 0.001,

^{** =}p < 0.01,

^{* =}p < 0.05

Wilcoxon singed-rank test (W)

^{### =}p < 0.001,

^{## =}p < 0.01,

^{# =}p < 0.05

n = 12

3.2. EMG magnitude and phase durations

When normalized to percent of the maximum effort recorded during the CHV condition, the magnitudes of diaphragm (p = 0.9), internal abdominal oblique (p = 0.5), parasternal left (p = 0.4), and parasternal right (p = 0.7) EMGs were unchanged between CHV and isocapnic conditions, as reported in

Table 1. Despite this, both average and maximal esophageal pressures generated in isocapnic trials were reduced 13% when compared to CHV trials (p = 0.04 and p = 0.002 respectively). There was no significant difference between CHV and isocapnia for cough phase durations CT_I (p = 0.1), CT_E1 (p = 0.5), CT_E2 (p = 0.4) or CT_TOT (p = 0.2).

3.3. Arterial and end-tidal CO₂

When compared to the CHV condition, end-tidal CO₂ was 168% greater and arterial CO₂ was 82% greater in the isocapnic condition (p < 0.001 and p = 0.009 respectively), as reported in Table 2. This was reflected by a statistically significant (p < 0.001) proportional change in pH from 7.52 ± 0.02 in our CHV condition to 7.42 ± 0.2 in our isocapnic condition. Additionally, to approximate ventilation and perfusion relationships, in the subset of animals that had an arterial blood gas analyzed (n = 9) we used a two-way repeated measures ANOVA compare the difference between source of our CO₂ measurement (alveolar/ETCO₂ and arterial/PaCO₂) within and between each of three conditions: CHV, isocapnia, and eupnea (no cough stimulus). This test revealed a significant effect of condition [F(2, 16) = 141.6, p < 0.001], CO₂ source [F(1, 8) = 21.8, p = 0.002], and their interaction [F(2, 16) = 5.9, p = 0.012]. A Post-hoc analysis was then used to compare PaCO₂ and ETCO₂ for each condition. In the CHV condition, the ETCO₂ was 32% lower than the PaCO₂ value (12.0 ± 0.9 vs 17.7 ± 0.9 mmHg), an arterial-alveolar difference of 5.7 mmHg; this was a significant difference (p < 0.001). In the isocapnic condition, ETCO₂ (31.5 ± 0.7 mmHg) and PaCO₂ (32.3 ± 1.4 mmHg) did not differ (p = 0.58), suggesting minimal arterial-alveolar difference (0.8 mmHg). During eupnea, ETCO₂ and PaCO₂ were 30.4 ± 1.1 and 36.6 ± 1.9 mmHg, respectively; ETCO₂ was 17% (p < 0.001) lower when compared to PaCO₂ during eupnea, an arterial-alveolar difference of 6.2 mmHg.

Table 2.

Cardiovascular measurements during cough induced hyperventilation (CHV) and isocapnia.

	CHV	Isocapnia
Measure	Mean ± SE	Mean ± SE	Δ (%)	Test	T/Z-score	p -value
Respiratory
ETCO2(mmHg)	11.8 ± 0.7	31.5 ± 0.7	168	P	−22.1	<0.001 ***
PaCO2 (mmHg)	17.7 ± 0.9	32.3 ± 1.4	82	W	−2.61	0.009 ##
Cardiovascular
Systolic (mmHg)	154.8 ± 7.4	163.0 ± 8.1	5	P	−3.08	0.028 *
Diastolic (mmHg)	106.6 ± 7.5	119.1 ± 8.1	12	P	−3.28	0.022 *
Pulse Pressure (mmHg)	48.2 ± 0.9	43.9 ± 1.9	−9	P	2.57	0.049 *
Mean Arterial Pressure (mmHg)	122.7 ± 7.4	133.7 ± 8.1	9	P	−3.28	0.022 *
Heart Rate (Beats/min)	208.6 ± 6.0	201.8 ± 6.5	−3	P	1.80	0.132
Acid-base
pH	7.52 ± 0.02	7.42 ± 0.02	−1.33	P	10.80	<0.001 ***

ETCO₂, end-tidal CO₂; PaCO₂, arterial CO₂.

Paired t-test (P)

^{*** =}p < 0.001,

^{** =}p < 0.01,

^{* =}p < 0.05

Wilcoxon singed-rank test (W)

^{### =}p < 0.001,

^{## =}p < 0.01,

^{# =}p < 0.05

ETCO2 n = 12, PaCO2 n = 9, pH n = 9, Cardiovascular n = 6

3.4. Cardiovascular effects

Cardiovascular features in a subset of 6 animals were measured during bouts of repetitive cough and compared between CHV and isocapnic conditions. Like the other parameters reported for cough above, the last half of the cough bout was used for this analysis, and these results are summarized in Table 2. When compared to CHV, isocapnic trials resulted in increased systolic (5% increase; p = 0.03) and diastolic (12% increase; p = 0.02) pressures during cough. The greater increase in diastolic pressure compared to the increase in systolic pressure resulted in reduced pulse pressure 9% (p = 0.049) between the isocapnic and CHV conditions. Mean arterial pressure was 9% higher (p = 0.02) during isocapnic compared to CHV conditions. There was no change in heart rate between CHV and isocapnia (p = 0.1).

3.5. Recovery from cough and mechanical hyperventilation

The recovery duration after the last cough was quantified by two measures: the duration of the post-cough apnea and the post-cough hypopnea. We also measured 5 animals that were mechanically hyperventilated to the point of apnea, wherein end-tidal CO₂ fell on average to 17 mmHg (SD = 2.4) and was maintained below apneic threshold for approximately 100 seconds (M = 96, SD = 4.9), and then the ventilator removed. This recovery period was measured similarly to the post-cough recovery, with the onset of each duration indicated by removal from the ventilator. These results are displayed in Fig 2B and Table 3. When compared to mechanical hyperventilation, apneas following CHV were 56% shorter (p = 0.02) and apneas following isocapnia were 86% shorter (p = 0.002). The apneas for the isocapnic condition were also 68% shorter than the CHV condition (p = 0.004). The duration of the post-cough hypopnea was only significantly different when comparing the isocapnic trials to the CHV trials (75% shorter; p = 0.006), and the isocapnic trials to mechanical ventilation (64% shorter; p = 0.04). There was no significant difference in post-cough hypopnea between cough-induced hyperventilation and mechanical hyperventilation conditions (p = 0.2).

Table 3.

Recovery from mechanical hyperventilation, cough induced hyperventilation (CHV), and isocapnic coughing.

		Compared to ventilator			Compared to cough
Condition	Mean ± SE	Δ (%)	Test	p -value	Δ (%)	Test	p -value
Apnea Duration (s)
Ventilator	40 ± 6	-	-	-	-	-	-
CHV	17 ± 3	−56	T	0.017 ♦	-	-	-
Isocapnia	6 ± 2	−86	M	0.002 ¤¤	−68	W	0.004 ##
Hypopnea Duration (s)
Ventilator	45 ± 8	-	-	-	-	-	-
CHV	63 ± 12	42	T	0.213	-	-	-
Isocapnia	16 ± 5	−64	M	0.039 ¤	−75	W	0.006 ##

Welch’s t-test (T)

^{♦♦♦ =}p < 0.001,

^{♦♦ =}p < 0.01,

^{♦ =}p < 0.05

Mann-Whitney test (M)

^{¤¤¤ =}p < 0.001,

^{¤¤ =}p < 0.01,

^{¤ =}p < 0.05

Wilcoxon singed-rank test (W)

^{### =}p < 0.001,

^{## =}p < 0.01,

^{# =}p < 0.05

CHV/Isocapnia n = 12, Ventilator n = 5.

Figure 3 shows mean traces of end-tidal CO₂ values, mean arterial carbon dioxide (PaCO₂) measures (also reported in Table 2), and mean arterial pressure (MAP) traces during CHV and isocapnic cough trials. MAP dropped during the beginning of both cough conditions, but the drop was attenuated in the isocapnic condition. In both conditions, MAP then gradually increased after 30 seconds, but that increase was more dramatic during this recovery period in the CHV condition. A Paired t-test of aggregated BP data from this period is reported in Table 2.

Fig 3. Mean arterial pressure fluctuates during cough and the subsequent recovery back to breathing, and this fluctuation was more pronounced during cough induced hyperventilation than during isocapnia.

The upper panel shows traces of group means of end-tidal CO₂ values during cough induced hyperventilation (CHV, dashed) and isocapnic (ISO, solid) cough trials. Mean arterial carbon dioxide (PaCO₂) measures are also shown (as colored dots) at the approximate time they were obtained during each trial. In the isocapnic condition, the alveolar (ETCO₂) and arterial (PaCO₂) carbon dioxide measures are nearly identical, suggesting no ventilation/perfusion mismatch. However, in the CHV condition the ETCO₂ was 33% lower than the PaCO₂ value, suggesting a mismatch due to low perfusion. The bottom panel shows group means of mean arterial pressure (MAP) recorded via femoral arterial cannula. MAP drops during the beginning of both cough conditions; matching the CO₂ to control levels in the isocapnic condition attenuated this decrease. Both conditions resulted in a gradual increase in MAP after 30 seconds, but the CHV condition produced a more dramatic rise in MAP during the recovery period, throughout the transition back to eupneic breathing. A Paired t-test of aggregated BP data from this period is reported in Table 2. n = 12. Grey bands represent a 95% confidence interval.

4. Discussion

The major findings of this study are that coughing in this model induces significant hyperventilation that did not profoundly reduce cough-related motor drive to respiratory muscles. The CO2 levels produced by this cough-induced hyperventilation were substantially lower than eupneic values and elicited hypopnea in the immediate post-cough period. Coughing during isocapnea produced small increases in cough motor drive and ameliorated the hypopnea following cough bouts. In separate animals, mechanical hyperventilation to similar CO2 levels elicited hypopneas and apneas that were substantially longer than those that followed coughing. The results support a minimal role of CO2 in the regulation of coughing relative to breathing and a reduction in apneic threshold following cough bouts. We propose that reduced apneic threshold following bouts of coughing functions to buffer significant oscillations in arterial CO2 induced by cough-induced hyperventilation.

4.1. Cough

Previous studies have investigated the effect of hypercapnia on cough (Nishino et al., 1998; Tatar et al., 1986; Tatár et al., 1987), but this is the first reported study on the effects of hypocapnia alone on cough in the cat. Hypercapnia inhibits a variety of airway protective behaviors including cough (Nishino et al., 1989; Tatár et al., 1987), expiration reflex (Nishino et al., 1998; Tatár et al., 1987), swallow (Nishino et al., 1989), and laryngospasm (Nishino et al., 1981). Tatar et al. (1986) found that severe hypoxia, leading to hyperventilation and thus hypocapnia, enhanced cough response in cats, but no data were reported for hypocapnia in normoxic conditions. Paroxysmal cough does not seem to cause hypoxia, as arterial oxygen tension slightly increased from approximately 101 mmHg to 113 mmHg in one study (Tomori et al., 1973). Nishino et al. (1989) also reported a similar relationship between hyperoxic hypocapnia and airway defensive reflexes in humans. However, that study was conducted postoperatively under a volatile anesthetic with a sustained hyperoxic physiology, which may have influenced peripheral cough receptors. The relative level of hypocapnia is another caveat. Understandably, the human subjects appeared to be under a relatively deep level of anesthesia, and the reported hypocapnic end-tidal CO₂ levels would be considered normocapnic in most experimental models. Nonetheless, our data suggest that cough induced hyperventilation and the resulting hypocapnia have no effect on cough motor pattern and minimal effect on cough sensitivity.

Alterations of end-tidal CO₂ produced no significant changes in cough phase timing or EMG amplitudes. This suggests that the cough motor pattern is robust and is resilient to the hyperventilation that results from increased ventilation during paroxysmal bouts of coughing. Additionally, this insensitivity to profound hyperventilation and the resulting hypocapnia provides further evidence for the differential regulation of cough and breathing by the RCN. Due to a small but significant difference between the maximal and average esophageal pressures produced, it is possible that there were changes in cough related chest wall muscle activity that were not captured by our measurements. While not measured, changes in laryngeal motor output between the conditions are a putative explanation for this difference, as hypocapnia increases upper airway resistance (Campbell et al., 1963; Nishino et al., 1981). Patients with congenital central hypoventilation syndrome displayed highly variable compression phase durations during maximal voluntary cough when compared to healthy individuals (Lavorini et al., 2007), however, the mean durations were similar between the two populations.

There was a small increase in cough number with isocapnia, but this minimal change may not be physiologically significant at the time scale that cough bouts typically occur (Table 1). It is possible that the mechanism responsible for the gain and patterning of the cough motor output is resistant to CHV, whereas the cough gating mechanism (Bolser et al., 2003, 1999; Canning and Mori, 2010) remains sensitive to changes in CO₂. This may contribute to the increase in cough frequency and reduction in citric acid cough threshold reported during both real and simulated ascents to high altitude, with the latter controlling for temperature and humidity (Barry et al., 1997; Mason et al., 1999; Mason and Barry, 2007).

4.2. Cardiovascular

While neural mechanisms governing cough phase timing and magnitude were relatively insensitive to CHV, the cardiovascular system appears to remain sensitive to changes in arterial CO₂ during CHV. The increases in systolic, diastolic, and mean arterial pressures during isocapnic conditions (Table 2) suggest that net sympathetic drive is higher with increased arterial CO₂. This is also consistent with a study that reported a reduction of sympathetic-related neuron oscillations during hyperventilation-induced hypocapnic apnea (Nuding et al., 2015). Heart rate did not change during isocapnia, which suggests other mechanisms such as alterations in baroreflex gain.

While the specific physiological underpinnings of cough syncope are unknown, a cough-associated reduction in blood pressure has been reported and implicated in cough syncope (Dicpinigaitis et al., 2014; Sharpey-Schafer, 1953b). We observed that coughing immediately reduced blood pressure, which reached a nadir at the 30 second mark of our cough stimulus, and then gradually increased to approximate the MAP level that had been observed during the eupnea immediately prior to the cough stimulus (Fig. 3). The immediate blood pressure decrease was smaller in the isocapnic condition than the CHV condition (Fig. 3; Table 2). As the cough stimulus was withdrawn and the respiratory system transitioned back to eupnea, the blood pressure appeared to overshoot its eupneic levels in the CHV condition, resulting in an overall profound oscillation in blood pressure across the trial and recovery. While this oscillation in blood pressure also occurred in the isocapnic condition, the overall oscillation was dampened. One possible explanation for this dampening is the lower intrathoracic pressures reached during the isocapnic condition compared to the CHV condition (Table 1). Peak intrathoracic pressure has been implicated in some individuals suffering from cough syncope, which may explain why cough syncope is more common in men than women (Dicpinigaitis et al., 2014). The compensatory rise in blood pressure that occurs in the latter half of the cough trial, and the subsequent overshoot during the recovery period suggest a more complicated relationship between blood pressure regulation and cough than previously assumed. We also cannot rule out a neurogenic component to the cardiovascular changes we observed during cough and the subsequent transition back to eupnea.

4.3. Cough-to-Eupnea Transition

Coughing results in a shorter apneic duration, but a similar hypopnea duration, for the transition back to breathing when compared to hyperventilation. This supports the idea that the cough pattern generator and/or cough receptors provide excitatory input to the respiratory pattern generator. The relatively brief apnea seen after 60 seconds of cough-induced hyperventilation was much shorter than that seen after hyperventilation induced by mechanical ventilation. This increased inspiratory excitability during the transition between cough and breathing could prevent or reduce extreme oscillations in blood gases as the system switches back to [H⁺]/CO₂ as the major source for respiratory drive.

These data also reveal that, while cough is relatively insensitive to its self-induced hypocapnia, the transition period back to breathing remains sensitive, as evidenced by the shorter apnea periods during the isocapnic trials. The hypopnea duration also seems to be influenced by arterial CO₂, with higher levels of CO₂ minimizing the duration of reduced breathing, however, there were no statistically significant differences between cough and ventilator induced hyperventilation on this measure of recovery. Our findings regarding the influence of CO₂ on the recovery from cough support data reported by Nishino et al. (1989), who saw a similar shortening of the post-cough recovery period with added CO₂ under hyperoxic conditions.

The increased sympathetic activity that may underlie the increase in blood pressure during coughing could result in pulmonary vasoconstriction(Liu, 1997; Shirai et al., 1994). This possibility is supported by our observation that the difference between PaCO₂ (representing arterial perfusion) and ETCO₂ (representing alveolar ventilation) was significantly higher during cough in CHV compared to isocapnic conditions. This represents a high estimated ventilation to perfusion ratio for the CHV condition. Tidal volume is 4–5 times higher during cough than eupnea in our experimental model(Bolser and Davenport, 2000), so there would be ample ventilation in both the CHV and isocapnic cough conditions to allow for offloading of CO₂ from the pulmonary capillaries. Thus, the difference in ratio can be ascribed to reduced perfusion. Pulmonary vasoconstriction would increase physiological dead space, producing units that are ventilated but not perfused, and would explain the values that we observed. This increase in physiological dead space would prevent PaCO₂ from dropping even further during cough-induced hyperventilation, allowing greater retention of arterial CO₂. This could further blunt apnea, as reflected in the differences in apnea durations between the CHV and mechanical hyperventilation conditions. Additionally, a small portion of the unresponsiveness of cough to hyperventilation may be attributed to the attenuated drop in PaCO₂, but we speculate that the main cause of cough insensitivity to hyperventilation is mediated centrally. We also cannot rule out the possibility that the reduced arterial-alveolar difference in our isocapnic condition is a result of gas mixing in the tracheostomy tube near where the ETCO₂ gas was sampled.

4.4. Limitations

The preparation used here to study cough is dependent on anesthetized animals. Pentobarbital is the ideal anesthetic for studying airway protective reflexes in larger mammals due to its relative sparing of airway protective reflexes when compared to other available anesthetic agents (Knociková, 2011; May and Widdicombe, 1954). However, barbiturates such as pentobarbital have been shown to depress chemosensitive reflexes (Wagner et al., 1991). It is likely that pentobarbital is influencing chemosensation during cough, and future studies using awake or decerebrate animals would be valuable for understanding these phenomena. Furthermore, this study focused on reflex cough occurring in an anesthetized state, but the cough is a highly complex reflex that is also mediated by and can be elicited via higher brain centers during periods of consciousness (Lee et al., 2002), therefore the data here can only speak to reflexive, and not evoked or volitional cough. Additionally, our methodology focused on CO2 chemosensitive areas in the brainstem. We are unable to differentiate the influence of peripheral chemoreceptors on the phenomena we report here. While peripheral chemoreceptors are thought to primarily sense O2, and may erroneously be deemed secondary to their central counterparts, this is not the case and they are certainly capable of detecting changes in CO2 (Carroll et al., 1991) as well as modulating the sensitivity of central chemoreceptors (Blain et al., 2010; Dahan et al., 2008; Smith et al., 2010). Our preparation also focused on mechanically induced cough, primarily mediated through small myelinated “cough receptors”, however cough can also be elicited through a variety of chemical stimuli mainly governed by unmyelinated c-fibers (Bolser et al., 1991; Widdicombe, 2001, 1998). Our results cannot determine whether changes in these peripheral receptors are occurring, nor if one type of receptor would be more or less sensitive to our perturbation than the other.

There could have been significant differences between the mechanical hyperventilation method and cough hyperventilation regarding CO2 washout. However, the ETCO2 value for mechanical hyperventilation was higher (17 mm Hg, section 3.5, Results) than that during cough hyperventilation (approximately 12 mm Hg, Table 2), but the apnea duration was longer in the mechanically hyperventilated animals.

5. Summary

Coughing in spontaneously breathing animals induces significant hyperventilation. This cough-induced hyperventilation does not reduce motor drive to chest wall inspiratory and abdominal expiratory muscles during this behavior, suggesting that eupneic levels of CO2 are not required for cough to be produced. Spatiotemporal features of cough are unaffected by cough-induced hyperventilation, but this hyperventilation influences the cough-to-eupnea transition period. When compared to mechanical hyperventilation, cough facilitates a faster transition to eupnea by reducing apneic periods under hypocapnic states, potentially blunting oscillations in blood gasses that might occur due to prolonged apneas.