Neuromechanical control of the isolated upper airway of mice

Abstract

We characterized the passive structural and active neuromuscular control of pharyngeal collapsibility in mice and hypothesized that pharyngeal collapsibility, which is elevated by anatomic loads, is reduced by active neuromuscular responses to airflow obstruction. To address this hypothesis, we examined the dynamic control of upper airway function in the isolated upper airway of anesthetized C57BL/6J mice. Pressures were lowered downstream and upstream to the upper airway to induce inspiratory airflow limitation and critical closure of the upper airway, respectively. After hyperventilating the mice to central apnea, we demonstrated a critical closing pressure (Pcrit) of −6.2 ± 1.1 cmH₂O under passive conditions that was unaltered by the state of lung inflation. After a period of central apnea, lower airway occlusion led to progressive increases in phasic genioglossal electromyographic activity (EMG_GG), and in maximal inspiratory airflow (V̇i_max) through the isolated upper airway, particularly as the nasal pressure was lowered toward the passive Pcrit level. Moreover, the active Pcrit fell during inspiration by 8.2 ± 1.4 cmH₂O relative to the passive condition (P < 0.0005). We conclude that upper airway collapsibility (passive Pcrit) in the C57BL/6J mouse is similar to that in the anesthetized canine, feline, and sleeping human upper airway, and that collapsibility falls markedly under active conditions. Active EMG_GG and V̇i_max responses dissociated at higher upstream pressure levels, suggesting a decrease in the mechanical efficiency of upper airway dilators. Our findings in mice imply that anatomic and neuromuscular factors interact dynamically to modulate upper airway function, and provide a novel approach to modeling the impact of genetic and environmental factors in inbred murine strains.

obstructive sleep apnea is a common disorder with a wide spectrum of neurocognitive, metabolic, and cardiovascular consequences (28, 37, 39, 52, 61). Its clinical sequelae stem from frequent nocturnal arousals and/or oxyhemoglobin desaturations, which result from recurrent episodes of upper airway obstruction during sleep (41). Obstruction has been related to elevations in pharyngeal collapsibility during sleep (12, 36, 47, 53), although the mechanism for these elevations is not well understood.

Structural alterations and disturbances in neuromuscular control account for elevations in upper airway collapsibility in sleep apnea patients compared with normal controls (36). Boney and soft tissue defects can increase pharyngeal collapsibility by elevating the pressure in tissues around the pharynx (15–19, 59). The impact of structural alterations on upper airway collapsibility may be mitigated by increases in lung volume and caudal traction on upper airway structures (18, 42, 54–56), which decreases extraluminal tissue pressures (18) and/or stiffens the pharynx (42, 54). Neuromuscular responses to negative intraluminal pressures (25, 26), pulmonary mechanoreceptors, and disturbances in gas exchange (43, 49, 51) can also lower upper airway collapsibility and relieve the obstruction during sleep (36). Thus both structural and neuromuscular factors can modulate upper airway collapsibility during sleep (36) and contribute to the development of obstructive sleep apnea (7, 60).

Animal models have been developed to explore the impact of structural and neuromuscular factors on upper airway collapsibility. In isolated canine, feline, rat, and rabbit upper airway preparations, the mechanical effects of lingual displacement, tracheal traction, and extraluminal tissue pressure (18–20, 42, 42, 54, 56) and the neuromuscular effects of alterations in gas exchange and airway pressure (1, 6, 8–10, 43, 51) on pharyngeal collapsibility have been demonstrated. Studies in pigs and bulldogs highlight the impact of structural narrowing on upper airway patency, particularly when neuromuscular activity wanes during rapid eye movement (REM) sleep (14, 24, 58). Recently, investigators have demonstrated that obesity can also impose structural loads on the airway (5) that elevate pharyngeal collapsibility in the fa/fa rat and that serotonergic mechanisms can offset this mechanical effect (27). We have demonstrated that a similar defect in leptin signaling in obese ob/ob mice accounts for the development of the obesity-hypoventilation syndrome in a murine model (29, 38), suggesting that mouse strains may ultimately serve to probe the genetic determinants of sleep-related breathing disorders. Nevertheless, a mouse model of upper airway function will be required to explore the genetic control of pharyngeal collapsibility.

The primary objective of the present study was to characterize the passive structural and active neuromuscular control of upper airway collapsibility in mice. We hypothesized that pharyngeal collapsibility, which is elevated by anatomic/structural loads, is reduced by active neuromuscular responses to airflow obstruction. To address this hypothesis, we developed an approach for assessing upper airway mechanical and neuromuscular control in the isolated upper airway of anesthetized mice. Our findings provide new evidence that mechanical loads and neuromuscular mechanisms interact to control upper airway collapsibility dynamically and offer an approach for probing genetic and environmental determinants of pharyngeal function in inbred murine strains.

METHODS

Mice

Male C57/BL6J mice were obtained from Jackson Laboratory (Bar Harbor, ME) and housed in a microisolation facility. Temperature and humidity were continuously regulated at 20–22°C and 40–60% relative humidity, respectively. Food and water were available ad libitum throughout the study. Male mice were utilized exclusively to minimize any variability in upper airway characteristics related to the estrous cycle. All study protocols were approved by the Johns Hopkins Animal Care and Use Committee (JHACUC), and all animal experiments were conducted in accordance with JHACUC guidelines.

Experimental Procedures

Anesthesia protocol.

The murine upper airway was isolated as previously described for larger animals (49, 50). Briefly, surgical anesthesia was induced with 2–3% isoflurane in a closed chamber and was maintained with urethane (1 g/kg ip) and etomidate (10 mg/kg ip) (11). Respiratory rate was continuously monitored, and additional doses of urethane (0.2 g/kg ip) and etomidate (2 mg/kg ip) were administered as required to maintain a stable plane of surgical anesthesia. The depth of anesthesia was continuously monitored based on a targeted respiratory rate of 120–160 breaths/min and the absence of pedal withdrawal to interdigital space pinch. At experiment completion, the animals were euthanized by an overdose of pentobarbital (60 mg ip). The rectal temperature was monitored continuously with a temperature probe, and body temperature was maintained at 36.5–37.5°C with a variable-temperature heating pad throughout the experiments.

Isolated murine upper airway surgery.

Mice were instrumented with pressure catheters at the trachea and the nose as previously described (Fig. 1) (49, 50). In brief, a midline neck incision from the sternum to the larynx was made while the mouse was supine. The trachea was transected, and the esophagus was ligated. Curved stainless steel 19-gauge catheters were placed in the caudal tracheal stub toward the lungs, and through the rostral tracheal stub and glottic structures to the level of the aryepiglottic folds, where it was fixed in place. The mice were mechanically ventilated through the caudal tracheal catheter (Harvard Instruments, model 683, Holliston, MA). The rostral upper airway catheter was connected in series to a pneumotachometer and negative pressure source, which was utilized to maintain a negative pressure downstream in the hypopharynx (P_DS). One nostril was cannulated (27-gauge PE tubing). The mouth was sewn shut with 6-0 silk, and the mouth and nostrils were sealed with glue and putty. The cannulated nostril was utilized to deliver a variable upstream pressure at the nose (P_US). The position of the rostral tracheal cannula was confirmed to be at the level of the glottic opening at the conclusion of the experiment. In all experiments, the head was fixed in place at an angle of 10–20° from the horizontal plane.

Fig. 1.

Experimental set-up. The upper airway was isolated by bisecting the trachea. The nasal and tracheal ends of the upper airway were cannulated. The rostral trachea was connected to a negative pressure source, and the caudal segment was connected to a ventilator (∼). A pneumotach was placed in-line with the nasal cannula to measure airflow through the isolated upper airway. The pressures upstream (nasal) and downstream (tracheal) to the isolated upper airway (P_US and P_DS) were monitored, as was the tracheal pressure (P_TRACH) in the lower airway.

Monitoring pressure-flow dynamics.

The upstream nasal (P_US) and downstream hypopharyngeal (P_DS) pressures were monitored (P23XL; Statham Laboratories), and inspiratory airflow (V̇i) through the isolated upper airway was monitored with a pneumotachometer (no. 0.771, Fleisch) and a differential pressure transducer (Validyne, Northridge, CA; MP 45–1; ±2 cmH₂O). The pneumotachometer was calibrated by applying steady-state levels of flow through a mass flowmeter (model FMA-A2303, Omega Engineering, Stamford, CT). The upstream nasal cannula was connected to a blow-by circuit that was attached on either side to a positive and negative pressure source. The pressure sources were regulated separately to apply varying levels of positive and negative pressure to the nose. The downstream tracheal cannula was connected to a negative pressure source. The P_US, P_DS, and V̇i were digitized (WinDaq DI-720; DATAQ Instruments, Akron, OH) for real-time display and storage and for later analysis. Throughout the experiments, secretions inside the catheter were flushed and aspirated through the catheter as needed.

EMG.

Two Teflon-insulated hooked-wire EMG electrodes (stainless steel wire, 0.005-in. bare, 0.007-in. coated, A-M Systems, Carlsborg, WA) were inserted using separate 26-gauge needles into the genioglossus muscle group bilaterally. The genioglossal EMG (EMG_GG) signal was amplified, band pass-filtered from 30 to 1,000 Hz (P511K, AC Preamplifier, Grass Instruments, Quincy, MA), and digitized at a sampling rate of 1,000 Hz (WinDaq DI-720; DATAQ Instruments). The EMG_GG was rectified, and a 200-ms time constant was utilized to compute the moving average (Advanced Codas, DATAQ Instruments). The moving average was employed for all subsequent data analysis (see below).

Mechanical ventilation.

Mice were mechanically ventilated through the caudal tracheal tube with a tidal volume of approximately 200 μl and respiratory rate of 120–140 breaths/min (model 687 Small Animal Ventilator, Harvard Apparatus, Holliston, MA). Respiratory rate was adjusted to induce a passive state for the pharyngeal musculature by suppressing phasic EMG_GG activity and spontaneous respiration (as assessed by negative tracheal pressure swings). Positive end-expiratory pressure (PEEP) was applied at the exhalation port (0, 5, and 10 cmH₂O) as described in Experimental Protocol below. Before each passive and active run (see below), the tidal volume and respiratory rate were increased to approximately 250 μl and 180 breaths/min, respectively, to induce a period of central apnea. The mouse was hyperventilated for several minutes before passive runs or for ∼15 s before active runs to induce relatively prolonged or short periods of central apnea when mechanical ventilation was discontinued.

Experimental Protocol

Characterizing passive upper airway properties.

Initially, P_DS was lowered to approximately −50 cmH₂O to induce a flow-limited state for the upper airway, as defined by a plateau in V̇i at a maximal level (V̇i_max) despite further reductions in P_DS (42, 43, 48, 49, 51, 54). Thereafter, the level of airflow (V̇i_max) through the isolated upper airway was determined at several P_US levels. At each P_US level, mechanical ventilation was discontinued to maintain a constant lung volume when measuring V̇i_max during the ensuing central apnea. P_US was lowered to defined levels of 4, 0, −4, and −8 cmH₂O in successive runs to encompass the level of nasal pressure at which flow ceased (the pharynx occluded when P_US fell below a critical pressure). At each level of P_US, V̇i_max was measured at a PEEP of 0, 5, and 10 cmH₂O to determine the effect of lung inflation on upper airway properties. Runs at each level of P_US were performed in triplicate in the passive and active protocols (see below).

Characterizing active upper airway properties.

As in the passive protocol, P_DS was lowered to induce a flow-limited state, as defined above. Mechanical ventilation was discontinued and the caudal tracheal cannula was occluded for ∼10 s, or until a sigh or swallow occurred. Following a brief period of central apnea, spontaneous breathing resumed, and inspiratory efforts increased progressively. Peak phasic and tonic levels of V̇i_max and EMG_GG were monitored throughout the period of tracheal occlusion during inspiration and expiration, respectively. Tracheal occlusion led to progressive increases in inspiratory effort (P_TRACH), V̇i_max, and EMG_GG from the start to end of the run. Active responses to tracheal occlusion were characterized by the changes in V̇i_max and EMG_GG over the course of each run (as described in results; see Fig. 4). The tracheostomy was then reopened, and mechanical ventilation was resumed until spontaneous respiratory efforts and phasic EMG_GG abated. The nasal pressure was lowered in stepwise fashion, and active runs were repeated at the P_US levels described above (see Characterizing passive upper airway properties). During active runs, the trachea was occluded for 9.9 ± 1.3 s across all nasal pressure levels and mice.

Fig. 4.

Representative recordings of active airflow (V̇i_max) and genioglossal EMG responses to tracheal occlusion at upstream (nasal) pressures of −4.0 cmH₂0 (left) and 0.0 cmH₂O (right). In each panel, the trachea was occluded following the discontinuation of positive pressure ventilation (see positive P_TRACH swings at left). After a period of central apnea, spontaneous breathing efforts resumed at the onset of the run (t_O) and increased progressively through the end of the run (t_f). Phasic levels of airflow through the isolated upper airway (V̇i_max) and raw EMG_GG increased progressively during the run. Phasic (open symbols) and tonic (closed symbols) EMG_GG and V̇i_max are illustrated in bottom graphs. While the phasic EMG_GG increased similarly at low and high nasal pressures, phasic increases in V̇i_max were markedly attenuated at the high compared with low nasal pressure (see bottom left vs. right panels).

Data Analysis

In the passive protocol, V̇i_max and P_US were measured during periods of central apnea, and separate pressure-flow relationships were constructed at each PEEP level. In the active protocol, V̇i_max and EMG_GG responses to tracheal occlusion were assessed at each level of P_US. These parameters were measured repeatedly for each inspiration and expiration during the entire run. The maximal EMG_GG activity was measured for each active (tracheal occlusion) run, and the mean maximal activity was calculated for each mouse. Subsequently, the peak phasic and tonic (minimum) levels of EMG_GG were measured for each inspiration and expiration, respectively, and were expressed as a percentage of the mean maximal EMG_GG activity in each mouse. Least-squares linear regression was utilized to characterize the rate of increase in EMG_GG and V̇i_max over time during periods of tracheal occlusion. To examine the phasic control of pharyngeal patency during periods of tracheal occlusion, phasic responses were represented as changes in V̇i_max (ΔV̇i_max) and EMG_GG (ΔEMG_GG) by subtracting the corresponding level of V̇i_max and EMG_GG activity during the preceding central apnea from phasic and tonic levels of these parameters during the period of spontaneous breathing (see Fig. 5).

Fig. 5.

Phasic changes in airflow (ΔV̇i_max) and EMG_GG (ΔEMG_GG) in response to tracheal occlusion at high and low P_US. Group mean (±SE) phasic increases in ΔV̇i_max vs. ΔEMG_GG are plotted for the highest (closed symbols) and lowest (open symbols) levels of nasal pressure applied (low P_US: −3.4 ± 1.1; high P_US: 5.7 ± 0.6 cmH₂O). The phasic component of the EMG_GG response to tracheal occlusion is represented at the start (bottom left symbols) and end (rightmost symbols) of the runs by subtracting the tonic level at the onset of the run from the peak phasic EMG_GG at each time point. Increases in ΔEMG_GG over the course of the runs were only associated with a significant increase in ΔV̇i_max at low but not high levels of upstream nasal pressure (P < 0.0001).

Separate pressure-flow relationships were constructed from data generated at each of several P_US levels (between 10 and 30 points in total) in both the passive and active protocols. Least-squares linear regression was applied over the linear portion of each V̇i_max vs. P_US relationship to determine the critical pressure (Pcrit) and upstream resistance (R_US), as previously described (36). In the active protocol, a passive V̇i_max vs. P_US relationship was constructed from data obtained at the start of each run (before neuromuscular activation), from which the passive Pcrit and R_US were derived. Pressure-flow data from the point of maximal neuromuscular activation at the end of each run were utilized to generate a separate active pressure-flow relationship. Neuromuscular modulation of upper airway function was quantified as the degree of leftward displacement of the active curve from the passive pressure-flow relationship at the passive Pcrit level. This pressure difference between the active and passive curves at the passive Pcrit level (ΔP_A-P) was taken to be a measure of the strength of the neuromuscular response (36). The sum of the passive Pcrit and the pressure difference, ΔP_A-P, was calculated to estimate the active Pcrit.

Values are expressed as means ± SE. Linear mixed-effects regression analysis (XTMIXED, Intercooled Stata 9.2, Statacorp, College Station, TX) was utilized to model the effect of PEEP on Pcrit and R_US (passive protocol), the time course of responses in V̇i_max and peak phasic and tonic EMG_GG to tracheal occlusion, and the effect of tracheal occlusion on Pcrit and R_US (active protocol). When significant differences were detected, post hoc comparisons were performed (Tukey) to determine the source of these differences. Statistical significance was inferred at a P < 0.05 level.

RESULTS

Passive Upper Airway Properties

Upper airway passive properties were characterized in six mice (weight 26.1 ± 1.8 g; age 94.0 ± 4.1 days). Representative P_TRACH, EMG_GG, and V̇i_max recordings are illustrated in one mouse at three levels of nasal pressure in separate panels (Fig. 2), while a markedly negative pressure was maintained at the downstream end of the isolated upper airway to induce inspiratory airflow limitation. At each nasal pressure level, three levels of PEEP (0, 5, and 10 cmH₂O) were applied (Fig. 2, see step changes in P_TRACH signals in each panel). At the start of each run in each panel, positive pressure ventilation was applied to the lower tracheostomy to hyperventilate the mouse (see positive P_TRACH swings from ventilator on left side of each panel). When the ventilator was stopped, a prolonged central apnea ensued (EMG_GG and P_TRACH remained quiescent for the remainder of the run, see right side of each panel). During this apnea, airflow through the isolated upper airway remained constant at each level of nasal pressure (see panels for each P_US level). As the nasal pressure was reduced from 4 to 0 to −4 cmH₂O, the level of airflow (V̇i_max) decreased progressively and ultimately ceased at the lowest nasal pressure applied (see left, center, and right panels, respectively). Moreover, the level of V̇i_max remained constant as PEEP was increased stepwise from 0 to 5 and 10 cmH₂O at each nasal pressure level, indicating no effect of lung inflation on upper airway flow. V̇i_max vs. P_US for the isolated upper airway in the representative mouse shown in Fig. 2 demonstrated a linear relationship with a passive Pcrit of −5.9 cmH₂O (Fig. 3), and no influence of PEEP on this relationship.

Fig. 2.

Representative recordings from 1 mouse illustrating the effects of 3 levels of upstream (nasal) pressure (P_US) and positive end-expiratory pressure (PEEP) on airflow (V̇i_max) through the isolated flow-limited upper airway. In each panel, positive-pressure ventilation is delivered (see positive swings in P_TRACH on left) and then abruptly discontinued (Vent Off), leading to a prolonged central apnea (spontaneous P_TRACH swings absent). During central apnea, PEEP was increased from 0 to 5 and 10 cmH₂O (downward arrows). V̇i_max decreased as P_US was lowered, but was not influenced by PEEP (the state of lung inflation). Note that phasic genioglossal EMG (EMG_GG) activity was absent, indicating suppression of central respiratory drive.

Fig. 3.

Representative passive pressure-flow relationship for the isolated upper airway at varying levels of lung inflation (positive end-expiratory pressure, PEEP) in the mouse illustrated in recording examples in Fig. 2. A linear relationship between maximal inspiratory flow (V̇i_max) and upstream (nasal) pressure (P_US) was demonstrated from which the critical pressure (Pcrit, −5.9 cmH₂O) and upstream resistance (R_US, 4.9 cmH₂O·ml⁻¹·s) were derived. PEEP did not influence Pcrit or R_US significantly.

In Table 1, passive Pcrit and R_US are reported at PEEP of 0, 5, and 10 cmH₂O for the entire group. The mean passive Pcrit was −6.19 ± 1.11 cmH₂O and R_US was 13.8 ± 1.8 cm H₂O·ml⁻¹·s and did not change significantly as a function of the level of PEEP.

Table 1.

Effects of PEEP on passive upper airway characteristics (Pcrit and R_US)

Variable	PEEP Level, cmH2O			Mean ± SE
	0	5	10
Pcrit, cmH2O	−6.93±1.07	−6.92±1.23	−6.53±1.16	−6.79±0.62
RUS, cmH2O·ml−1·s	14.4±3.6	13.8±3.0	14.4±3.6	13.8±1.8

Values are means ± SE. PEEP, positive end-expiratory pressure; Pcrit, critical pressure; R_US, upstream resistance.

Active Upper Airway Properties During Lower Tracheal Occlusion

Active responses to lower airway occlusion were assessed in a separate group of 7 mice (weight 26.4 ± 0.8 g; age 83.6 ± 11 days). Representative recordings are illustrated for one mouse at −4.0 cm H₂O and 0.0 cm H₂O level of nasal pressure (Fig. 4, left and right, respectively). As in the passive protocol, P_DS was lowered to induce a flow-limited state for the isolated upper airway, and positive pressure ventilation was applied to the lower airway (Fig. 4; see augmenting positive P_TRACH swings on left side of panels). After discontinuing mechanical ventilation, an initial period of central apnea was followed by a progressive increase in inspiratory efforts (Fig 4; see negative P_TRACH swings in middle and right half of each panel). At the start of the run (t_O), baseline levels of airflow through the isolated upper airway were greater at a high (∼1.1 ml/s) compared with a low (∼0.6 ml/s) nasal pressure level, as expected. As tracheal pressure swings increased, phasic changes in EMG_GG and airflow ensued and increased progressively over time (see flow and EMG_GG vs. time, two lower signals, Fig. 4). Comparing EMG_GG responses to tracheal occlusion at high and low nasal pressures, we found that levels of tonic EMG_GG were similar at the start and end of these runs (at t_O and t_f) and that phasic EMG_GG activity increased similarly over time. By the end of the run (t_f), phasic changes in airflow were greater at the low compared with high nasal pressure level (Fig. 4, left vs. right).

As illustrated in Fig. 4, bottom graphs, the peak phasic (inspiratory) and tonic (expiratory) EMG_GG and V̇i_max rose linearly over time in each mouse during periods of tracheal occlusion (P < 0.0003). The rate of rise in V̇i_max was significantly greater during inspiration than expiration (3.18 ± 0.67 vs. 1.25 ± 0.30 ml·s⁻¹·s⁻¹, P < 0.003), as was the rate of rise in EMG_GG (0.06 ± 0.01 vs. 0.01 ± 0.003 μ V/s, P < 0.0001). Since V̇i_max and EMG_GG increased linearly over time during periods of tracheal occlusion, the initial and final breaths of each run were utilized to examine the neuromuscular control of upper airway patency over the range of upstream pressures applied.

In Fig. 5, the relationship between changes in V̇i_max and EMG_GG are illustrated for the highest and lowest levels of P_US applied (low P_US: −3.4 ± 1.1; high P_US: 5.7 ± 0.6 cmH₂O). Despite comparable increases in EMG_GG during tracheal occlusion over the entire range of P_US applied, V̇i_max increased markedly only at the low (P < 0.001) but not high P_US level.

In Fig. 6, the effects of tracheal occlusion on the passive and active pressure-flow relationships are illustrated for the mouse whose recording examples appear in Fig. 4. V̇i_max at two levels of P_US are plotted for data obtained from the start of the run, demonstrating a passive Pcrit of −4.2 cm H₂O. During tracheal occlusion, peak inspiratory V̇i_max increased by the end of the run (see Fig. 4) more at the lower compared with higher level of P_US, decreasing the slope of the active compared with passive pressure-flow relationship (see Fig. 6, active curve). The shift in the active pressure-flow relationship from the passive curve, ΔP_A-P, provided a measure of the strength of neuromuscular activation at the passive Pcrit level, as shown. Utilizing ΔP_A-P to estimate the active Pcrit (see methods), we found significant decreases in Pcrit and increases in R_US at maximum activation during inspiration compared with expiration, and compared with the passive condition at the onset of tracheal occlusion for the entire group (Fig. 7). Compared with the passive pressure-flow relationship, the active curve was displaced leftward and reduced in slope, leading to underestimates of the active Pcrit (as represented by the nasal pressure at zero flow) and the Pcrit response to tracheal occlusion. The passive Pcrit did not differ significantly in the active compared with the passive protocol (P = nonsignificant).

Fig. 6.

Passive and active relationships between V̇i_max and P_US are derived from representative recordings illustrated in Fig. 4. The passive curve demonstrated a passive Pcrit of −4.2 cm H₂O. Tracheal occlusion resulted in phasic increases in V̇i_max, particularly at low P_US levels, and an overall reduction in slope of the active curve. The pressure difference, ΔP_A-P, between active and passive curves at passive Pcrit represents the strength of neuromuscular activation. See text for details.

Fig. 7.

Pcrit (A) and R_US (B) are represented for the passive condition and for the active condition during inspiration and expiration. Pcrit was significantly lower in the active condition during inspiration than expiration (P < 0.005) and was significantly lower than the passive Pcrit (P < 0.0005). R_US was significantly higher in the active condition during inspiration than expiration (P = 0.011) and was significantly higher than the passive R_US (P < 0.025).

DISCUSSION

In this study, we examined the mechanical and neuromuscular control of pharyngeal collapsibility in the isolated upper airway of the mouse. Passive pressure-flow relationships demonstrated a moderately negative passive Pcrit in the range previously found in sleeping humans (2, 35, 36, 45). In contrast to studies in larger mammals and human, the passive Pcrit did not vary with alterations in lung volume (PEEP), suggesting that it was primarily determined by pharyngeal soft tissues and boney structures. These passive properties were modulated dynamically by tracheal occlusion, which elicited progressive neuromuscular responses in phasic genioglossal activity and upper airway flow. Of note, responses in airflow were most marked at low compared with high levels of nasal pressure, leading to a reduced slope of the pressure-flow relationship with a concomitant decrease in active Pcrit and increase in R_US compared with passive conditions. In contrast, EMG_GG responses to tracheal occlusion did not vary with nasal pressure and were only associated with substantial phasic increases in airflow at lower levels of nasal pressure. These differences in EMG_GG and airflow responses can be attributed to improved neuromechanical coupling when the pharyngeal transmural pressure is reduced. Our findings imply that passive and active elements interact dynamically to modulate the collapsibility of the isolated murine upper airway and establish an approach for modeling the impact of these components on sleep apnea susceptibility.

Our approach to modeling murine upper airway function under passive conditions was based on principles of pressure-flow dynamics in simple collapsible conduits (Starling resistor). The isolated murine upper airway demonstrated two essential characteristics of a Starling resistor (46, 47, 53). First, inspiratory airflow limitation was produced by lowering the downstream (tracheal) pressure until airflow plateaued at a maximal level (V̇i_max) and became independent of further decreases in downstream pressure. Second, V̇i_max was linearly related to the upstream (nasal) pressure, and ceased when nasal pressure fell below a critical pressure, as described for sleeping humans. Pcrit and R_US for the isolated murine upper airway were derived from the V̇i_max vs. P_US relationship. In contrast to previous methods for assessing these parameters in larger animals (1, 8–10, 27, 30, 40), the present approach obviated the need for an intraluminal pressure-monitoring cannula in the pharynx, which might have produced artifactual changes in our measurements of upper airway properties in the mouse. Despite this methodological difference, pressure-flow dynamics in the isolated murine upper airway resembled those previously demonstrated in the larger anesthetized animals and sleeping humans and yielded comparable levels of passive Pcrit (27, 30, 46, 47, 49, 51, 53), suggesting similarities in the mechanical determinants of pharyngeal patency across species.

The passive Pcrit can be modulated by pharyngeal tissue pressure (rabbit) and caudal traction (dog, cat) on upper airway structures (13, 42, 50). The latter is produced by elevations in lung volume and transmitted by mechanical attachments between the chest and upper airway structures (55, 56). In contrast to prior studies in larger animals and humans, PEEP, a surrogate for increasing lung volume in our preparation, did not influence the passive Pcrit in our isolated murine upper airway. This finding suggests that mechanical linkage between chest and upper airway structures is lacking in this murine model. Such linkage may be diminished by the elevated compliance of the murine chest wall (23) or by fixing the position of the rostral tracheal stump in our preparation (54). Since lung volume can still influence upper airway function in dogs even when the tracheal position is fixed (56), our findings in the murine upper airway suggest that pharyngeal collapsibility under passive conditions is largely determined by upper airway rather than thoracocervical structures.

Relative to passive conditions, tracheal occlusion elicited progressive increases in phasic genioglossal activity over the entire range of upstream pressures applied, although phasic increases in upper airway flow were only observed at lower levels of upstream nasal pressure. These responses can be attributed to a loss of phasic inhibition from pulmonary mechanoreceptor and progressive disturbances in gas exchange (51), leading to marked increases in pharyngeal neuromuscular activity (49). Our methods allowed us to quantify the strength of the pharyngeal musculature by referencing active responses to the passive state. In the passive condition, the pharynx occluded when the nasal pressure was lowered to the passive Pcrit level, at which point the pharyngeal transmural pressure became zero. At this level of nasal pressure, any increase in airflow in the active condition could be attributed to increased dilator activity (31, 33, 34, 44, 50), which can decrease the extraluminal tissue pressure (19) and increase transmural pressure accordingly. Alternatively, airway patency may have been stabilized by coactivation of tongue protrusor and retrusor muscles (1, 8–10), pharyngeal constrictors (22), and/or cervical strap muscles (33, 57). Thus the pressure difference between the active and passive curve is caused by differences in transmural pressure and may reflect the strength of pharyngeal dilator and constrictor muscles, tongue protrusor and retrusor muscle, and/or cervical strap muscle responses to tracheal occlusion. Differences in airflow responses to tracheal occlusion at the passive Pcrit level suggest that the mechanical efficiency of the pharyngeal musculature increases as the upstream pressure is lowered. Nevertheless, measurements of muscle force during tracheal occlusion would be required to confirm this conclusion.

An intriguing finding was that in contrast to EMG_GG responses to tracheal occlusion, phasic changes in airflow increased progressively with reductions in nasal pressure. Responses in airflow and neuromuscular activity dissociated at higher nasal pressures, suggesting a loss of dilating action in pharyngeal muscles (see Fig. 8). The mechanical efficiency of these dilator muscles depends on their resting length (3, 4, 32), which may vary with the size of the pharyngeal lumen relative to its boney enclosure (59). At lower levels of nasal pressure, the pharynx will narrow, and dilators will elongate toward their optimal length (L_O) (32). Neuromuscular activation at L_O would yield the greatest increase in tension, transmural pressure, and airflow. Alternatively, overdistension of the pharyngeal lumen relative to the size of the boney enclosure (21, 59) may limit the mechanical action of these dilating muscles.

Fig. 8.

“Tube in box” model of upper airway and boney enclosure at low (left) and high (right) P_US in active (bottom) and passive (top) conditions (see arrows representing dilator muscle action, active conditions). In passive condition, the airway is occluded at low and open at high P_US. In active condition, dilator muscle activity improves airway patency (reduces its collapsibility) at the low but not high P_US. Dilators restore patency at low P_US in active condition because muscles are operating from optimal length in passive condition (L_O). Alternatively, patency does not improve at high P_US because distension of the tube is limited by the size of the boney enclosure (box) when dilators contract.

Several limitations should be considered when interpreting our data. First, our approach to characterizing the passive and active properties of the upper airway requires that a flow-limited state be achieved by lowering the downstream tracheal pressure below inspiratory levels. This downstream pressure was confined to the hypopharynx, and exposure was limited in duration to avoid mucosa drying. Second, active responses were assessed over a relatively narrow range of nasal pressures that were applied to define the passive pressure-flow relationship. Nevertheless, our protocol elicited a marked neuromuscular response and allowed us to isolate the active component of this response from the passive baseline (36). Third, we recognize that dynamic responses during anesthesia cannot be extrapolated to sleep. Nonetheless, the isolated murine upper airway exhibited pressure-flow dynamics, a passive Pcrit, and dynamic neuromuscular responses that were remarkably similar to those previously demonstrated across species (27, 30, 43, 49, 51) and in sleeping subjects (36). Despite differences in upper airway anatomy and size, it is nonetheless remarkable that inspiratory airflow limitation and critical closure occur at similar levels of pharyngeal critical pressure in normal sleeping humans and in anesthetized dogs, rabbits, cats, rats, and mice (10, 27, 42, 49, 51, 54). Finally, we recognize that our genioglossal EMG signal may have included activity from adjacent intrinsic tongue muscles, dilator muscles (e.g., geniohyoid), tongue retrusors (e.g., stylohyoid and geniohyoid), and cervical strap muscles due to close proximity of these muscles to the recording electrodes.

Our approach to analyzing upper airway pressure-flow relationships under active and passive conditions has yielded new insights into mechanisms of upper airway neuromuscular control. In contrast to prior work in larger animals, our strategy for probing neuromuscular control mechanisms in mice was based on comparisons of pressure-flow dynamics over a range of upstream pressure. Utilizing this approach, we found evidence that active neuromuscular control of pharyngeal collapsibility is greatest at low levels of upstream (transmural) pressure. These findings suggest that the pharyngeal transmural pressure influences the mechanical efficiency of the musculature for any given level of neural output.

Our findings have also laid a foundation for studies examining the impact of genetic and environmental factors on murine upper airway control. They serve to establish a conceptual foundation for modeling passive and active upper airway properties across species. This experimental approach can be utilized to capitalize on the power of inbred murine strains to explore the genetic basis for alterations in upper airway control. In addition, inbred murine strains can be employed to examine the impact of environmental exposure on upper airway control (e.g., caloric excess leading to obesity, local pharyngeal inflammation, and interactions with underlying cardiopulmonary disease). Finally, further work will be required to characterize and evaluate the impact of these factors on upper airway function during sleep.

GRANTS

This research was supported by National Heart, Lung, and Blood Institute Grants HL-50381 and HL-37379.