Hypercapnia‐induced active expiration increases in sleep and enhances ventilation in unanaesthetized rats

Abstract

Key points

• Expiratory muscles (abdominal and thoracic) can be recruited when respiratory drive increases under conditions of increased respiratory demand such as hypercapnia.

• Studying hypercapnia‐induced active expiration in unanaesthetized rats importantly contributes to the understanding of how the control system is integrated in vivo in freely moving animals.

• In unanaesthetized rats, hypercapnia‐induced active expiration was not always recruited either in wakefulness or in sleep, suggesting that additional factors influence the recruitment of active expiration.

• The pattern of abdominal muscle recruitment varied in a state‐dependent manner with active expiration being more predominant in the sleep state than in quiet wakefulness.

• Pulmonary ventilation was enhanced in periods with active expiration compared to periods without it.

Abstract

Expiration is passive at rest but becomes active through recruitment of abdominal muscles under increased respiratory drive. Hypercapnia‐induced active expiration has not been well explored in unanaesthetized rats. We hypothesized that (i) CO₂‐evoked active expiration is recruited in a state‐dependent manner, i.e. differently in sleep or wakefulness, and (ii) recruitment of active expiration enhances ventilation, hence having an important functional role in meeting metabolic demand. To test these hypotheses, Wistar rats (280–330 g) were implanted with electrodes for EEG and electromyography EMG of the neck, diaphragm (DIA) and abdominal (ABD) muscles. Active expiratory events were considered as rhythmic ABD_EMG activity interposed to DIA_EMG. Animals were exposed to room air followed by hypercapnia (7% CO₂) with EEG, EMG and ventilation (${\dot{V}}_{\mathrm{E}}$) recorded throughout the experimental protocol. No active expiration was observed during room air exposure. During hypercapnia, CO₂‐evoked active expiration was predominantly recruited during non‐rapid eye movement sleep. Its increased occurrence during sleep was evidenced by the decreased DIA‐to‐ADB ratio (1:1 ratio means that each DIA event is followed by an ABD event, indicating a high occurrence of ABD activity). Moreover, ${\dot{V}}_{\mathrm{E}}$ was also enhanced (P < 0.05) in periods with active expiration. ${\dot{V}}_{\mathrm{E}}$ had a positive correlation (P < 0.05) with the peak amplitude of ABD_EMG activity. The data demonstrate strongly that hypercapnia‐induced active expiration increases during sleep and provides an important functional role to support ${\dot{V}}_{\mathrm{E}}$ in conditions of increased respiratory demand.

Key points

• Expiratory muscles (abdominal and thoracic) can be recruited when respiratory drive increases under conditions of increased respiratory demand such as hypercapnia.

• Studying hypercapnia‐induced active expiration in unanaesthetized rats importantly contributes to the understanding of how the control system is integrated in vivo in freely moving animals.

• In unanaesthetized rats, hypercapnia‐induced active expiration was not always recruited either in wakefulness or in sleep, suggesting that additional factors influence the recruitment of active expiration.

• The pattern of abdominal muscle recruitment varied in a state‐dependent manner with active expiration being more predominant in the sleep state than in quiet wakefulness.

• Pulmonary ventilation was enhanced in periods with active expiration compared to periods without it.

Abbreviations

ABD_EMG

abdominal EMG

DIA_EMG and ABD_EMG

diaphragm and abdominal EMG

∫DIA_EMG and ∫ABD_EMG

integrated DIA_EMG and ABD_EMG

Neck_EMG

neck muscle EMG

NREM

non‐rapid eye movement sleep

OSA

obstructive sleep apnoea

Peak Amp

peak amplitude

pFRG

parafacial respiratory group

REM

rapid eye movement

RF

respiratory frequency

RTN

retrotrapezoid nucleus

T_E

expiratory time

${\dot{V}}_{\mathrm{E}}$

ventilation

T_I

inspiratory time

T_TOT

total cycle duration

V_T

tidal volume

V_T/T_I

mean inspiratory flow

Introduction

Respiration is a rhythmic process initiated spontaneously in the brainstem respiratory network. Under resting conditions expiration is passive, but expiratory muscles (abdominal and thoracic) can be recruited when respiratory drive increases under conditions such as hypercapnia, hypoxia or exercise (Iscoe, 1998; Iizuka, 2011; Feldman et al. 2013; Jenkin & Milsom, 2014). Recent evidence indicates that the parafacial respiratory group (pFRG), which overlaps to some extent with the retrotrapezoid nucleus (RTN), is a crucial brainstem site for generating active expiration (Janczewski et al. 2002; Mellen et al. 2003; Janczewski & Feldman, 2006; Huckstepp et al. 2015, 2016). Disinhibition of pFRG neurons and stimulation of the RTN/pFRG elicit active expiration in anaesthetized adult rats (Pagliardini et al. 2011), and the expiratory activity induced by hypercapnia is eliminated after chemical inhibition of the RTN/pFRG in in situ preparations (Abdala et al. 2009). This has led to the suggestion that the pre‐Bötzinger complex generates the inspiratory rhythm, while the pFRG generates the expiratory rhythm, and each functions as distinct, but coupled, oscillators (Huckstepp et al. 2016).

The central respiratory control system is strongly influenced by vigilance state (Nattie, 2001; Feldman et al. 2003; Kuwaki et al. 2010; Nattie & Li, 2010; Pagliardini et al. 2012). Burke et al. (2015) recently reported that hypercapnic ventilatory response is highly dependent on vigilance state in rats. Moreover, the activity of most respiratory muscles is reduced in sleep [rapid eye movement (REM) and non‐REM (NREM) sleep] (Sherrey & Megirian, 1977, 1980). The main inspiratory muscle, the diaphragm, is less affected but other accessory muscles exhibiting respiratory activity, such as the genioglossus, are strongly affected (Horner et al. 2002; Horner, 2008; Pagliardini et al. 2012). Interestingly, however, under eupnoeic conditions, where active expiration is absent normally, abdominal activity emerges in the REM‐like sleep state in urethane‐anaesthetized rats as well as in REM sleep, providing stability to breathing (Pagliardini et al. 2012; Andrews & Pagliardini, 2015). Evidence of hypercapnia‐induced expiratory abdominal activity in rats extends from in situ preparations (Abdala et al. 2009; de Britto & Moraes, 2016; Jenkin et al. 2017) to in vivo anaesthetized animals (Iizuka & Fregosi, 2007; Lemes & Zoccal, 2014), but whether active expiration occurs in unanaesthetized animals under hypercapnic conditions in a state‐dependent manner, i.e. differently in sleep and wakefulness, remains to be seen. Finally, active expiration has also been hypothesized to enhance ventilation by producing a more effective breathing pattern. In this context, the impact of hypercapnia‐induced abdominal expiratory activity on the functional aspects of ventilation across vigilance states has been poorly explored. The characterization of CO₂‐induced active expiration in unanaesthetized animals may advance our understanding of a rhythmic respiratory‐related behaviour in freely moving animals that is essential to organismic homeostasis by enhancing sustained ventilation in conditions of increased respiratory demand.

Here we characterized active expiration in unanaesthetized, adult Wistar rats during hypercapnia in sleep and wakefulness. Moreover, we evaluated the functional impact of active expiration on the breathing pattern (balance of tidal volume and breathing frequency) in freely moving rats. To do so, combined measurements of EEG, EMG and ventilation (V _E) were performed in unanaesthetized, freely moving adult rats. Our data support the view that hypercapnia‐induced active expiration occurs in a state‐dependent manner (in wakefulness and more so in NREM sleep) and improves pulmonary ventilation when recruited.

Methods

Ethical approval and animals

Animal handling and experimental protocols were carried out in compliance with the guidelines of the National Council for Animal Care (CONCEA: Conselho Nacional de Controle de Experimentação Animal) and approved by the local Animal Care and Use Committee (Protocol no. 019289/14) of Sao Paulo State University (UNESP/FCAV). Animals (adult male Wistar rats; 280–330 g) were provided by the animal facility of UNESP (Botucatu campus), and, after transportation to the Jaboticabal campus of UNESP, they were acclimatized to the animal facility for at least 7 days before any procedures were done. Animals were housed under a 12 h dark–light cycle (lights on at 07.00 h) and had free access to water and food. All experiments were performed in the light phase between 09.00 and 16.00 h at room temperature 24–25°C.

Surgical procedures

Electrodes for EMG and EEG recordings were implanted surgically. Bipolar intramuscular EMG electrodes were made of multi‐strand, coated stainless steel wire (A‐M Systems, Sequim, WA, USA) and EEG screw electrodes were commercially acquired (Plastics One Inc., Roanoke, VA, USA). Animals were anaesthetized with ketamine (100 mg kg⁻¹; i.p.) and xylazine (15 mg kg⁻¹; i.p.) to have the EMG and EEG electrodes implanted. For the insertion of EMG electrodes in the diaphragm (DIA_EMG) and abdominal (ABD_EMG) muscles, animals were placed in the supine position, and after local asepsis, a small horizontal incision was made in the lateral portion of the abdomen. For the ABD_EMG, the oblique abdominal muscle was targeted and a pair of electrodes were inserted. On the opposite side, for the DIA_EMG, a diagonal incision was made at the superior–lateral portion of the abdomen (from lateral to midline direction), following the ribcage border, from where the diaphragm muscle was accessed and electrodes were inserted. EMG electrode wires were tunnelled under the skin, and attached to an electrical socket secured between the animal's shoulder blades. For implanting the EEG electrodes and EMG electrodes in the neck muscle (Neck_EMG), the head and dorsal region of the neck were shaved and animals were placed in a stereotaxic apparatus. Next, the local asepsis was performed (dorsal region of the neck and head) followed by a subcutaneous injection of local anaesthetic (3% lidocaine hydrochloride). A skin midline incision was performed and the skull exposed. Three EEG electrodes were then screwed into the skull (Nattie & Li, 2002; da Silva et al. 2010), as follows: (i) the frontal electrode was placed 2 mm anterior to the bregma and 2 mm lateral to the midline (right side), (ii) the parietal electrode was placed 4 mm anterior to lambda and 2 mm lateral to the midline (right side) and (iii) the ground electrode was placed between the frontal and parietal electrodes (on the left), forming a triangle. A pair of electrodes was inserted into the dorsal cervical neck muscle to record the Neck_EMG. A plastic pedestal was used as an electrical socket and covered with a dust cap (Plastics One Inc.). The electrodes were mounted and fixed to the skull using acrylic resin. After surgery, the animals receive injections of antibiotic (enrofloxacin, Baytril; 10 mg kg⁻¹ s.c., Bayer S.A., S.o Paulo, SP, Brazil) and analgesic (Flunixina Meglumina, Banamine; 2.5 mg kg⁻¹ s.c., Schering‐Plough, Kenilworth, NJ, USA)

Physiological recordings

For recordings, an insulated and shielded cable was connected to the plugs/sockets on the animal's head and neck. The cable was attached to an electrical swivel, which permitted free movement of the animal. A custom‐made lid for the experimental chamber contained a hole through which the swivel was inserted and attached. From the swivel in the chamber lid a cable was connected to a four‐channel amplifier (A‐M Systems, model 1700). Signals were amplified (1000×), with 60 Hz notch filters, and band‐pass filtered (low and high cut‐off: 10 and 1000 Hz for EMG signals and 0.1 and 100 Hz for EEG signals, respectively). Signals were acquired (sample rate: 1 kHz) on a PowerLab (ADInstruments, Bella Vista, NSW, Australia) acquisition system and recorded using LabChart8 software. Online digital filtering was applied when necessary.

Ventilation (${\dot{V}}_{\mathrm{E}}$) was obtained by using a well‐established barometric method (whole body plesthymography; for a detailed description see: Nattie & Li, 2002; da Silva et al. 2010, 2011, 2013). Briefly, before connecting the EEG and EMG wires, unanaesthetized rats were placed in the recording chamber and allowed to move freely (∼30 min) while the chamber was flushed with humidified air (1500 ml min⁻¹). The inflow gas was controlled using a gas‐mixing pump. The flow rate through the animal chamber was maintained at 1500 ml min⁻¹. The temperature inside the chamber was continuously monitored. The breathing‐related pressure oscillations were detected by a differential pressure transducer and amplified (MLT141 spirometer, Power Lab, ADInstruments). A volume calibration was performed by injecting a known air volume (1 ml) inside the chamber. Tidal volume (V _T) was calculated using the formula described by Drorbaug & Fenn (1955) and Bartlett & Tenney (1970). ${\dot{V}}_{\mathrm{E}}$ was calculated as the product of V _T and respiratory frequency (RF). From the ${\dot{V}}_{\mathrm{E}}$ trace, the inspiratory and expiratory time (T _I and T _E), total respiratory cycle duration (T _TOT) and the mean inspiratory flow (V _T/T _I) were calculated.

Experimental protocol

Experiments were performed a minimum of 8 days after the surgical procedures. All experiments began at 09.00 h (light phase) and were carried out at room temperature (24–25°C). Animals were placed in a cylindrical plethysmograph chamber (5 litres) and allowed to acclimatize for 30 min before connecting EEG and EMG wires. During this period the chamber was flushed with room air at a flow rate of 1500 ml min⁻¹. Then, small wires for recording EMG and EEG signals were connected to the implanted wires in the animal and connected to an electrical swivel in the internal face of the lid, and from the external face of the lid to the amplifiers (A‐M Systems, model 1700). The use of a swivel was critical to keep the animal undisturbed and the wires untangled throughout the experimental protocol, and this wire arrangement allowed the animal to move freely inside the chamber. With all cables in place, further time (∼30 min) was given before starting to record signals. Room air was continuously flushed during the whole period of adaption. After that, EEG and EMG signals were recorded during room air exposure for a time interval of usually 40–60 min that yielded recording signals in different sleep–wake cycles. The gas was then switched to a hypercarbic gas mixture (7% CO₂, 21% O₂, N₂ balance) provided by a gas mixing system (GSM‐3 Gas Mixer; CWE Inc., Ardmore, PA, USA) at a flow rate of 1500 ml min⁻¹. Animals were exposed to hypercapnia for 60 min, and physiological parameters described above were measured throughout the time of exposure. After the experiments all animals were killed with an overdose of anaesthesia.

Data and statistical analyses

Data were collected electronically for subsequent analyses. The following criteria were applied for data analysis: all periods containing artifacts of animals’ movements, exploratory behaviours and grooming were excluded from analysis, whereas periods of quiet wakefulness and sleep were all analysed. EEG and Neck_EMG recordings were used to define sleep or wakefulness states (Nattie & Li, 2002; da Silva et al. 2010). The EMGs from respiratory muscles were integrated (0.05 s), and the muscle activity was calculated as the peak amplitude (Peak Amp) of the integrated DIA_EMG (∫DIA_EMG) and ABD_EMG (∫ABD_EMG), respectively. Event detection and Peak Amp analysis were performed on Clampfit software on a peak‐to‐peak basis during different alert states (sleep and wakefulness). The detection and selection of ∫DIA_EMG and ∫ABD_EMG during the data analysis were all performed within the exact time window for each signal. This allowed the assessment of DIA‐to‐ABD ratio within a given time window of the trace. This ratio expresses the occurrence of ABD events in relation to the DIA. A 1:1 ratio indicates a high occurrence of ABD activity (each DIA event is followed by an ABD event). Hence, the lower the ratio, the higher the occurrence of active expiration. All data were exported and further processed in an Excel spreadsheet where the ∫DIA_EMG and ∫ABD_EMG data were normalized to their largest integrated (∫) EMG Peak Amp obtained during the wakefulness state for each muscle. The normalized Peak Amp values were used for comparisons. For the purpose of analysis, active expiration was defined by the presence of ABD_EMG expiratory activity, i.e. rhythmic burst of ABD_EMG activity (between inspiratory DIA_EMG bursts) above tonic levels (Pagliardini et al. 2011; Huckstepp et al. 2015). Data were analysed in a state‐dependent manner, separated into two states – quiet wakefulness and non‐REM sleep – both observed consistently throughout the experimental protocol both during room air and hypercapnia exposures. REM sleep periods were very short and in most of the experiments were not present. Hence, all analysis during sleep is reported for NREM sleep. Arousal states were determined based on EEG and neck EMG and categorized as previously described (Nattie & Li, 2001, 2002; Taylor et al. 2006; da Silva et al. 2010). All active expiration events (according to the criteria) in each animal data set were averaged. Distributions (histograms) of the ∫ABD_EMG Peak Amp (absolute number of events) were determined individually for each rat in different states during hypercapnia, and these data were also reported in box plots. For ventilatory variables, the analyses during hypercapnia were run in periods with or without active expiration in a state‐dependent manner. Data were expressed as mean ± SEM. Student's t tests and two‐way ANOVA for repeated measures followed by post hoc tests were performed accordingly. A Pearson test was used for correlation analysis. P was set at 0.05 to indicate a significant difference.

Results

Room air

EEG and Neck_EMG recordings allowed identification of sleep–wake states. Figure 1 A shows a representative spectrogram, EEG and Neck_EMG (top, middle and bottom, respectively) of an animal during room air and hypercapnia exposure. During room air exposure, the percentage of time spent in wakefulness or in NREM sleep in the group data averaged 33.7 and 63.3%, respectively. Figure 1 B and C depict the recorded physiological variables during room air exposure in both wakefulness and sleep. The mean breathing frequency measured by ∫DIA_EMG events was 101 ± 5 and 86 ± 4 cycles min⁻¹ in wakefulness and sleep (n = 8), respectively. As for the ABD_EMG activity, regardless of the state, no active expiration was observed during room air exposure (n = 8), i.e. no rhythmic expiratory ABD_EMG activity was present, either in wakefulness (Fig. 1 B) or in sleep (Fig. 1 C). The absence of active expiration was consistent among animals breathing room air.

Figure 1. State‐dependent activity of abdominal and diaphragm activity during room air and hypercapnia exposure.

A, EEG spectrogram, EEG and Neck_EMG raw recordings (top, middle and bottom, respectively) of a representative animal during a typical experimental protocol in room air (left) and hypercapnia exposure (right). In the EEG spectrogram, the colour intensity, as indicated by the colour calibration bar to the right side, corresponds to the amount of signal power at a given time and frequency. On the left side, frequency calibration bar: 2 Hz. Note that the animal cycles between wake and sleep in both conditions: room air and hypercapnia. Above the spectrogram, the letters and arrows indicate for each of the respective panels where their traces were withdrawn and expanded. The expanded EEG and EMGs [raw and integrated abdominal (∫ABD_EMG) and diaphragm (∫DIA_EMG) signals] during room air are shown in periods of wakefulness (B) and sleep (C), and during hypercapnia in wakefulness (D) and sleep (E). The horizontal rectangles in D and E depict the expanded trace shown in F and G, respectively. Note the absence of active expiration during room air (B and C) and its recruitment during hypercapnia (D–G). [Color figure can be viewed at http://wileyonlinelibrary.com]

Hypercapnia exposure

During hypercapnia exposure (7% CO₂) the percentage of time in wakefulness or in NREM sleep averaged 55.8 and 44.5%, respectively. The hypercapnia‐induced decrease in sleep time was the result of a longer time spent in wakefulness. These effects of hypercapnia in the arousal state have been well documented in the literature (Taylor et al. 2006; Buchanan & Richerson, 2010; Dias et al. 2010; Buchanan et al. 2015; Vicente et al. 2016). The recruitment of abdominal muscles occurred during exposure to CO₂ (Fig. 1 D–G). Hypercapnia‐induced active expiration was not present during the entire period of CO₂ exposure. Figure 1 D is an example showing that during hypercapnia in a wakefulness period active expiration was not present initially, but that expiratory ABD_EMG activity emerged after a brief period of movement. The ∫ABD_EMG Peak Amp associated with abdominal expiratory activity during wakefulness was relatively constant (Fig. 1 F). During sleeping periods, however, active expiration was more consistent, and the ∫ABD_EMG Peak Amp associated with expiratory ABD_EMG activity was often highly variable (Fig. 1 E and G). This general pattern was also observed in other animals (Figs 2 and 3). It is noteworthy that 1 out of 9 rats recorded in the present study did not show any active expiration while awake, but only while sleeping.

Figure 2. Distribution of active expiration (A.E.) events (∫ABD_EMG Peak Amp) during hypercapnia in wakefulness and sleep.

Top panel shows the group data as box plots in which the centre line shows the median; box limits indicate the 25th and 75th percentiles; whiskers extend to 5th and 95th percentiles; and crosses represent sample means. Each histogram (row) represents the distribution for one individual animal in both vigilance states: wakefulness (left column) and sleep (right column). Note that one animal (bottom row) did not present active expiration in wakefulness, but only in sleep. [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 3. Respiratory muscle EMG and breathing frequency during hypercapnia in different vigilance states: wakefulness and sleep.

Box plots: centre line shows the median; box limits indicate the 25th and 75th percentiles; whiskers extend to 5th and 95th percentiles; and crosses represent sample means. The panels show the DIA‐to‐ADB ratio (A), ∫ABD_EMG Peak Amp (B), breathing frequency (C) and ∫DIA_EMG Peak Amp (D) (n = 8). The P values are indicated for the sleep data set compared to wakefulness. ^*Difference (P < 0.05) of sleep state compared to wakefulness state.

Figure 2 illustrates the group data (box plot) and the distribution (histograms) of ∫ABD_EMG Peak Amp (i.e. active expiration events) for each individual animal in different states. The box plot (Fig. 2, top) shows an apparent larger variability of ∫ABD_EMG data during sleep. The greater variability during sleep is confirmed by the coefficient of variation (CV) of ∫ABD_EMG Peak Amp, which was 96.4 and 65% during sleep and wakefulness, respectively. The ∫DIA_EMG had lower variability compared to abdominal activity and was not influenced by vigilance state (the CV of ∫DIAEMG was 35.5 and 28.8% during sleep and quiet wakefulness, respectively). Moreover, the histograms demonstrate that, for most animals, the number of events of active expiration as well as the number of a given ∫ABD_EMG Peak Amp tended to be greater during sleep (right column, Fig. 2) when compared to wakefulness (left column, Fig. 2).

The DIA‐to‐ADB ratio, ∫ABD_EMG Peak Amp, ∫DIA_EMG Peak Amp and breathing frequency during hypercapnia are presented in Fig. 3. The increased occurrence of active expiration during sleep is evidenced by the decreased DIA‐to‐ADB ratio (a 1:1 ratio means that each DIA is followed by an ABD event, indicating a high occurrence of ABD activity). Figure 3 A shows that this ratio decreased during sleep (∼2:1) compared to wakefulness (∼7:1) (mean values: 1.92 ± 0.2 vs. 6.87 ± 3; wakefulness and sleep, respectively; P = 0.02, n = 8). In Fig. 3 B, the amplitude of ∫ABD_EMG was 33.2% larger during sleep when compared to wakefulness (0.205 ± 0.02 vs. 0.307 ± 0.05; P = 0.06, n = 8; normalized Peak Amp for wakefulness and sleep, respectively) (Fig. 3 B). Breathing frequency had a ∼6% reduction (147 ± 5 vs. 138 ± 2 cycles min⁻¹, P = 0.066, n = 8) and ∫DIA_EMG Peak Amp (0.43 ± 0.4 vs. 0.45 ± 0.06, n = 8) was not different between wakefulness and sleep, respectively (Fig. 3 C and D).

Active expiration and assessment of ventilatory variables

Hypercapnia consistently recruited active expiration, but periods in which active expiration failed to occur were also observed (Fig. 1). To assess the functional role of active expiration during hypercapnia in unanaesthetized freely moving rats, ventilatory measurements were recorded along with EEG and Neck_EMG, DIA_EMG and ABD_EMG. First, ventilatory variables were assessed in the trace periods with active expiration and compared to the same variables in periods without active expiration. Figure 4 shows the relative change of the ventilatory variables analysed separately in periods with or without active expiration. The ventilatory response to hypercapnia was enhanced by active expiration. V _T, RF and ${\dot{V}}_{\mathrm{E}}$ were higher (P = 0.04, P = 0.026, P = 0.003, n = 7; respectively) when active expiration was present, and this effect was not dependent on sleep–wake cycle. T _I was not significantly affected by the presence of active expiration, whereas T _E and T _TOT decreased (P = 0.02 and P = 0.011, respectively) in periods where active expiration was present. Therefore, in unanaesthetized rats during hypercapnia, whenever present, active expiration enhanced ${\dot{V}}_{\mathrm{E}}$ through both increased V _T and RF; the reduction in T _TOT was mainly achieved by a diminished T _E. These effects were independent of vigilance state.

Figure 4. Comparison of ventilatory variables during hypercapnia in periods with or without recruitment of active expiration.

V _T and T _E top panels, RF and T _I middle panels, and VE and T _TOT bottom panels. (*) indicates difference (P < 0.05) comparing the ventilatory variables with and without active expiration (A.E.) (n = 7). The P values are indicated.

Figure 5 shows a representative trace of EEG, Neck_EMG, DIA_EMG and ABD_EMG recorded across a sleep–wake cycle during hypercapnia. Note that during the natural transition from sleep to wakefulness there was increasing ABD_EMG activity (also seen in Fig. 1 E and G) within the sleeping period, but ABD_EMG rhythmic activity ceased completely as soon as the animal woke up. As mentioned earlier, the ∫ABD_EMG Peak Amp associated with abdominal expiratory activity during wakefulness was relatively constant (see Fig. 1 F), whereas this type of increasing ∫ABD_EMG Peak Amp occurred more during sleep (see Figs 1 E and G and 5).

Figure 5. Functional impact of active expiration with increasing ABD_EMG activity on ventilatory variables during hypercapnia.

Top traces depict EEG, Neck_EMG, barometric V _E recording and raw and integrated DIA_EMG and ABD_EMG. Two arrows below the ∫ABD_EMG trace show the interval time with increasing ABD_EMG activity from which correlation analyses between active expiration (∫ABD_EMG Peak Amp) and ventilation were performed. A breath‐by‐breath analysis was run in 60 events. Scatter plots show the correlation between increasing ∫ABD_EMG Peak Amp and T _E, T _I and V _T/T _I (top row) and V _T, RF and V _E (bottom row).

The ventilatory variables were measured breath‐by‐breath over this recording to assess the overall functional impact of the increasing expiratory activity on ${\dot{V}}_{\mathrm{E}}$. A breath‐by‐breath analysis was run by taking each breath and the corresponding ∫ABD_EMG Peak Amp (representative trace and scatter plots, Fig. 5). The ∫ABD_EMG Peak Amp was positively correlated with RF (r = 0.56, P < 0.001), ${\dot{V}}_{\mathrm{E}}$ (r = 0.37, P = 0.003) and V _T/T _I (r = 0.45, P < 0.001) and negatively correlated with T _I (r = −0.47, P < 0.001), while no significant correlation was observed with T _E and V _T (P = 0.5 and P = 0.11, respectively). Therefore, once active expiration was present, the increasing ABD_EMG activity improved ${\dot{V}}_{\mathrm{E}}$ through inspiratory actions, mainly due to increases in RF (decrease T _I) and V _T [augmented mean inspiratory flow (V _T/T _I)].

Discussion

In the present study, while no active expiration was observed in unanaesthetized animals breathing room air (regardless of sleep–wake state), abdominal muscle expiratory activity was recruited during hypercapnia exposure with an increased occurrence during NREM sleep compared to wakefulness. Moreover, whenever present, active expiration was associated with an increase in ${\dot{V}}_{\mathrm{E}}$ compared to periods without it.

Active expiration has gained attention recently arising from data indicating the existence of a brainstem conditional expiratory oscillator, the pFRG, distinct from (but coupled to) the inspiratory oscillator, the pre‐Bötzinger complex (Pagliardini et al. 2011; Huckstepp et al. 2015, 2016). Under eupnoeic conditions (rats breathing room air), urethane‐anaesthetized rats increased abdominal activity switching from tonic to expiratory rhythmic activity (i.e. active expiration) with the transition from NREM to REM‐like sleep (Pagliardini et al. 2012). This has also been reported to occur during natural REM sleep in unanaesthetized rats (Andrews & Pagliardini, 2015). In the present study, we did not observe active expiration (expiratory ABD_EMG activity) during sleep in animals breathing room air. The source of this difference is not clear. However, in the urethane‐anaesthetized rats, in ∼46% of recordings (6 out of 13) ABD_EMG activity was tonic (no active expiration) and did not change across sleep–wake cycle (Pagliardini et al. 2012). Likewise, in unanaesthetized rats in room air, events of active expiration (that lasted for more than three consecutive breaths as defined by the authors) occurred in only ∼49% of REM epochs (147 out of 300 REM epochs) (Andrews & Pagliardini, 2015). The amount of REM sleep seen in our study was small, suggesting that long‐lasting recording experiments are needed to yield sufficient REM episodes to accurately examine the recurrent activation of ABD_EMG activity.

In the present study, abdominal oblique muscles were targeted as representative of hypercapnia‐induced expiratory abdominal activity. Abdominal muscles are innervated by branches from the ventral rami of intercostal nerves of thoracic segments and from lumbar segments, and there is a variable overlap of the abdominal motoneurons in regions of the ventral horn (Iscoe, 1998). The rectus abdominis has little or no respiratory‐related activity in humans and conscious rats (Abe et al. 1996; Iscoe, 1998). In contrast, and in agreement with our data, a consistent response of the abdominal obliques to hypercapnia has been reported in both humans and rats (Jeffries et al. 1984; Takasaki et al. 1989; Abe et al. 1996; Iscoe, 1998; Iizuka, 2011; Pagliardini et al. 2012; Andrews & Pagliardini, 2015).

An increased respiratory drive leads to emergence of expiratory abdominal activity in humans (Badr et al. 1990; Wakai et al. 1992; Abe et al. 1996), dogs (Oliven et al. 1985; Smith et al. 1989, 1990) and rodents (Sherrey et al. 1988; Iizuka & Fregosi, 2007; Lemes & Zoccal, 2014). In rats, evidence of hypercapnia‐induced expiratory ABD_EMG activity extends from in situ preparations (Abdala et al. 2009; de Britto & Moraes, 2016; Jenkin et al. 2017) to in vivo anaesthetized animals (Iizuka & Fregosi, 2007; Lemes & Zoccal, 2014). Fewer studies, however, have examined its occurrence in unanaesthetized rats in different vigilance states. In agreement with the literature, we reported that the CO₂ drive to breathe (hypercapnia exposure) recruits active expiration in unanaesthetized adult rats. Two observations are noteworthy: (i) in unanaesthetized rats, the expiratory ABD_EMG activity was not always recruited during exposure to hypercapnia either in wakefulness or in sleep, suggesting that other factors also influence the recruitment of active expiration; and (ii) the pattern of this recruitment varied in a state‐dependent manner with active expiration being more predominant in NREM sleep state compared to quiet wakefulness.

The observation (i) that active expiration failed to be consistently recruited by hypercapnia in both sleeping and awake rats is consistent with the observation that active expiration was not consistently recruited in REM sleep (Andrews & Pagliardini, 2015). It seems that factors other than levels of chemical drive and state influence abdominal expiratory activity. Some evidence suggests that the posture of the animals may affect the activity of ABD_EMG. Although Sherrey et al. (1988) reported that hypercapnia elicited active expiration in rats regardless of posture, when the rat curled its body oppositely from the side of the recording electrodes, the ABD_EMG activity was large and decreased to a minimum when the rat was curled toward the electrodes. In humans, body position is also suggested to play a role in the emergence of active expiration, as the abdominal recruitment elicited by increased inspiratory resistance was seen in upright but not supine subjects (Martin & De Troyer, 1982; Badr et al. 1990). Moreover, De Troyer et al. (1989) suggested that, in conscious dogs, proprioceptive inputs from the abdominal wall could promote or modulate ABD activity during expiration. To our knowledge, how the proprioceptive components act to inhibit or facilitate the emergence of active expiration remains uncertain. It is worth mentioning that, in unanaesthetized, freely moving rats, active expiration seems to be a complex respiratory motor behaviour influenced by different components (i.e. chemical drive, state and proprioception); however, how these are computed in the brainstem expiratory generating network to finally express active expiration remains unknown.

The underlying mechanism of the second observation (ii), that active expiration is more predominant in NREM sleep, is also elusive. Recent evidence supports the role of neurons (different from the RTN Phox2b neurons) in the pFRG as the brainstem site that generates active expiration (Huckstepp et al. 2015, 2016). Burke et al. (2015) reported that RTN stimulation caused active expiration only when the rats were awake. In contrast, other reports in the literature showed that active expiration is recruited during sleep in rats and humans (Pagliardini et al. 2012; Andrews & Pagliardini, 2015; and in our present study). The fact that RTN neurons drive active expiration only during wakefulness suggests that different mechanisms to recruit active expiration may be involved during sleep. Alternatively, as suggested by Burke et al. active expiration could yet be triggered in NREM sleep by stimulating a larger fraction of RTN neurons. Serotoninergic (Lemes et al. 2016) and cholinergic (Boutin et al. 2017) neurotransmission in the RTN/pFRG have now been reported to induce active expiration in anaesthetized rats. Disinhibition of these neurons has been shown to underlie the recruitment of active expiration (Pagliardini et al. 2011) and it is quite possible that this disinhibition is influenced by vigilance state. The modulation of pFRG neurons across state as well as the state‐dependent involvement of neurotransmitters is currently not well known. It has also been hypothesized that the activity of different brainstem CO₂ chemoreceptor sites varies in a state‐dependent manner (Nattie, 2000; da Silva et al. 2010; Nattie & Li, 2010), and hence it is also possible that the different CO₂‐sensitive cell types input into and modulate pFRG neurons to generate active expiration differently in sleep and wakefulness.

The state‐dependent enhancement of CO₂‐induced active expiration may also be related to a change in peripheral airway resistance. In humans during NREM sleep, expiratory abdominal muscles are recruited by sustained inspiratory resistive loading (Badr et al. 1990; Henke et al. 1991), obstructive sleep apnoea (Jeffries et al. 1984; Praud et al. 1989) and acute bronchospasm in asthmatics (Issa & Sullivan, 1985), and an interaction of increased airway resistance and chemoreceptor stimuli have been suggested as necessary to recruit abdominal muscles during NREM sleep (Badr et al. 1990; Henke et al. 1992). This suggestion emphasizes the importance of studying active expiration in hypercapnic conditions. In rodents, it has been shown that the activity of airway muscles such as the cricothyroid, nasolabial (Sherrey & Megirian, 1977) and genioglossal muscles (Horner, 2008) changes across state, which implies changes in airway resistance. In the present study, we did not record from other airway respiratory muscles, but it remains possible that the increase in ABD_EMG activity during NREM sleep (Fig. 1 E and F) was related to fluctuation of airway resistance during sleep.

Another important aspect of the present study is the description of the functional impact of active expiration on the ventilatory variables. We demonstrate here that, regardless of vigilance state, the presence of active expiration elicited an enhancement of ventilation achieved by increases in both V _T and RF (with decrease in T _E and T _TOT; Fig. 4). This is in agreement with the notion that active expiration can increase V _T by recruiting the expiratory reserve volume and increase RF by shortening the expiratory time constant (shorter than the passive time constant of the system) with a decreased T _E and T _TOT (Jenkin & Milsom, 2014). In humans, the main mechanical consequence of abdominal recruitment, specifically during sustained respiratory loading, was a shortened T _E and increased RF (Badr et al. 1990; Henke et al. 1991).

When active expiration was already present and the ABD_EMG activity was recruited further in a progressive manner, this was achieved mainly by an increase in RF (i.e. with diminished T _I) and increased mean inspiratory flow (V _T/T _I), although inspiratory DIA_EMG activity did not increase to support the increased V _T/T _I. The lack of further change in V _T could reflect maximum recruitment of the expiratory reserve volume. At this point further recruitment of the expiratory muscles can only serve to produce complete exhalation within the same T _E. Associated with this may be changes in compliance or resistance (lengthening tau, the passive time constant) reflecting state‐dependent changes in airway muscle activity (and hence resistance). Also, ABD_EMG activity could place the diaphragm at a better length for generating tension and improving mechanical output during inspiration. Indeed, these mechanical consequences of expiratory muscle recruitment have been reported during transient events in sleeping humans (Henke et al. 1991, 1992). The observed effect of increasing ABD_EMG activity on inspiration (increased V _T/T _I and decreased T _I) could also be associated with a parallel increase in activity of other inspiratory accessory muscles that were not recorded in the present study.

The respiratory control system is dedicated to maintaining the rhythmic activity of respiratory muscles via respiratory motor output in a coordinated fashion. Their rhythmic activities have to ultimately match metabolic demand and maintain organismic homeostasis of blood gases. Using a simple, yet elegant approach, the present study brings fundamental knowledge about control of breathing through the following integrative perspectives on active expiration: the brainstem expiratory generating network, which is influenced by sleep state (although the underlying mechanisms remain to be further explored), and, simultaneously, active expiration enables optimal ventilatory strategies during ventilatory challenge, further supporting important physiological function. By showing the state dependence of the hypercapnia‐induced active expiration, we also have another (patho)physiological aspect. The hypercapnia exposure used in the present study elicits a ∼15 mmHg rise in arterial $P_{\mathrm{C}{\mathrm{O}}_2}$, reaching values of about 50 mmHg (for detailed values see da Silva et al. 2011). A rise in blood CO₂ can be life threatening, and hypercapnia is a significant factor in diseases such as obstructive sleep apnoeas (OSAs) with consequent variable degrees of morbidity (and mortality). For example, an increased occurrence of active expiration during sleep was reported in patients with OSA, compared to subjects without OSA (Jeffries et al. 1984). Hypercapnia was present in patients with OSA and some of them who had significant blood gas abnormalities, i.e. $P_{\mathrm{C}{\mathrm{O}}_2}$ values above 50 mmHg, presented the highest ABD expiratory activity during sleep (Jeffries et al. 1984). Therefore, appropriate respiratory motor outputs as well as ventilatory adjustments to hypercapnia are highly desired.

In conclusion, the present study in unanaesthetized adult rats indicates that while no active expiration is present under normocapnic conditions in any vigilance state, ABD_EMG expiratory activity is recruited during hypercapnia predominantly during NREM sleep. We also show that active expiration is associated with improved ventilation achieved by both increased V _T and RF, with decreased T _E and T _TOT.

Additional information

Competing interests

The authors declare no competing interests.

Author contributions

I.P.L. and C.A.S. contributed equally to acquisition, analysis, interpretation of data and manuscript revision; L.H.G. contributed to the design of the work and manuscript revision; G.S.F.S contributed to acquisition, analysis and interpretation of data, conception, study design, drafting and manuscript revision. All authors revised and approved the final version of the manuscript.

Funding

This study was funded by Sao Paulo Research Foundation (FAPESP): Young Investigator Award (Grant nos. 2013/17606‐9 and 2014/12190‐1) and undergraduate scholarships (2016/11061‐9 and 2015/04926‐0).