Journal of Applied Physiology
Vol. 81, No. 6, pp. 2428-2435, December 1996
CONTROL OF BREATHING, CIRCULATION, AND TEMPERATURE

Effect of hypoxia on abdominal motor unit activities in spontaneously breathing cats

J. H. Mateika, E. Essif, and R. F. Fregosi

Department of Physiology, The University of Arizona Health Sciences Center, Tucson, Arizona 85721

ABSTRACT

INTRODUCTION

METHODS

RESULTS

DISCUSSION

ACKNOWLEDGEMENTS

FOOTNOTES

REFERENCES

ABSTRACT

Mateika, J. H., E. Essif, and R. F. Fregosi. Effect of hypoxia on abdominal motor unit activities in spontaneously breathing cats. J. Appl. Physiol. 81(6): 2428-2435, 1996.These experiments were designed to examine the behavior of external oblique motor units in spontaneously breathing cats during hypoxia and to estimate the contribution of recruitment and rate coding to changes in the integrated external oblique electromyogram (iEMG). Motor unit activities in the external oblique muscle were identified while the cats expired against a positive end-expiratory pressure (PEEP) of 1-2.5 cmH₂O. After localization of unit activity, PEEP was removed, and recordings were made continuously for 3-4 min during hyperoxia, normoxia, and hypoxia. A total of 35 single motor unit activities were recorded from 10 cats. At each level of fractional concentration of end-tidal O₂, the motor unit activity was characterized by an abrupt increase in mean discharge frequency, at ~30% of expiratory time, which then continued to increase gradually or remained constant before declining abruptly at the end of expiration. The transition from hyperoxia to normoxia and hypoxia was accompanied by an increase in the number of active motor units (16 of 35, 20 of 35, and 29 of 35, respectively) and by an increase in the mean discharge frequency of those units active during hyperoxia. The changes in motor unit activity recorded during hypoxia were accompanied by a significant increase in the average peak amplitude of the abdominal iEMG. Linear regression analysis revealed that motor unit rate coding was responsible for close to 60% of the increase in peak iEMG amplitude. The changes in abdominal motor unit activity and the external oblique iEMG that occurred during hypoxia were abolished if the arterial PCO₂ was allowed to fall. We conclude that external oblique motor units are activated during the latter two-thirds of expiration and that rate coding and recruitment contribute almost equally to the increase in expiratory muscle activity that occurs with hypoxia. In addition, the excitation of abdominal motor units during hypoxia is critically dependent on changes in CO₂ and/or tidal volume.

electrophysiology; external oblique; rate coding; recruitment

INTRODUCTION

THE DISCHARGE PATTERN of a whole abdominal expiratory nerve or muscle is characterized by an initial steeply rising phase, followed by a phase of constant neural activity, which abruptly declines before the onset of inspiratory activity (4, 5, 10, 13, 16, 20). This pattern remains relatively constant during either hyperoxic hypercapnia or isocapnic hypoxia. However, the rate of rise of the initial phase and the mean activity of the plateau phase increase in vagally intact spontaneously breathing animals (5, 9, 16, 20) and in paralyzed animals that are either ventilated (7, 8) or apneic (10).

Although the effect of hypercapnia and hypoxia on abdominal nerve or muscle discharge has been studied extensively, only one investigation to date has been designed to examine the discharge characteristics of individual abdominal motoneurons to changes in chemical stimuli. Fregosi et al. (7) showed that abdominal motoneurons recorded from decerebrate, vagotomized, thoracotomized, paralyzed, and artifically ventilated cats were characterized by a decrementing discharge pattern. Furthermore, the results from this investigation demonstrated that these motoneurons are not recruited throughout the expiratory cycle. Moreover, the increase in peak integrated whole nerve activity that accompanied the transition from normocapnia to hypercapnia was caused by an increase in the discharge of individual motoneurons at the start of expiration, but the recruitment or rate coding of motoneurons was not observed (7).

It is difficult to envision how the discharge characteristics of these single units could be responsible for the whole abdominal nerve or muscle discharge pattern described above and for the increase in amplitude that has been recorded from the abdominal muscles of vagally intact spontaneously breathing cats in hypoxia (5). It is possible that the pattern of motoneuron activity that was recorded in the Fregosi et al. investigation was caused by the removal of descending inhibition from cortical structures, or by inhibitory spinal and vagal reflex mechanisms that were removed during surgical preparation of the animal.

This hypothesis is supported indirectly by the results obtained by Russell et al. (12), which showed that an increase in abdominal nerve activity, elicited by the administration of continuous positive airway pressure or expiratory occlusion, was produced by rate coding and recruitment of external oblique motoneurons in spontaneously breathing cats. However, no investigation completed to date has examined the behavior of external oblique motor units during chemoreceptor stimulation in the spontaneously breathing cat. Although it is probable that rate coding and recruitment are responsible for changes in whole abdominal muscle activity during chemoreceptor stimulation, the relative contribution of these two strategies has not been determined. Furthermore, because abdominal motoneurons are the site of convergence of a variety of afferent inputs the behavior of the motoneuron pool will be highly dependent on the experimental conditions imposed. Therefore, the discharge pattern of abdominal motoneurons during chemoreceptor stimulation likely is not identical to that recorded in response to stimulation of lung, chest wall, diaphragm, and abdominal proprioceptors (12).

The purpose of the present investigation was to examine the behavior of external oblique motor units in response to alterations in central and peripheral chemoreceptor activity in the spontaneously breathing animal. Specifically, we were interested in 1) estimating the relative contribution of recruitment and rate coding to changes in the abdominal muscle discharge pattern; 2) quantifying the temporal pattern of external oblique motor unit discharge throughout the respiratory cycle; and 3) comparing the behavior of external oblique motor units in the spontaneously breathing cat with that recorded from the decerebrate cat.

METHODS

Animal preparation. Acute experiments were performed on 10 supine cats of either sex, with weights ranging from 1.5 to 4.1 kg. All of the experimental procedures were approved by the Institutional Animal Care and Use Committee of the University of Arizona. The animals were anesthetized initially with a mixture of 2-4% halothane in oxygen. The trachea was cannulated for the measurement of airflow, and the femoral arteries and veins were cannulated for blood pressure monitoring, withdrawal of arterial blood samples, and the intravenous infusion of anesthetic agents and fluids. Subsequently, halothane anesthesia was discontinued, and a bolus of -chloralose (35-50 mg/kg) and urethan (200 mg/kg) was administered intravenously. Supplemental doses of -chloralose (5 mg/kg) and urethan (20 mg/kg) were administered throughout the experiment to maintain a desired level of anesthesia, which was determined by ensuring that deep pressure applied to the animal's paw did not elicit a withdrawal reflex or cause blood pressure or ventilation to increase. The right or left external oblique muscle was exposed after surgical anesthesia was achieved.

Measurement of respiratory and cardiovascular parameters. Bidirectional airflow was measured with a pneumotachometer that was attached to the tracheal cannula. To deliver the desired mixture of O₂, CO₂, and N₂ to the animals during the experimental protocol, the outflow port of a rotameter was attached to the pneumotachmeter with a t-tube system, which has been described previously (4). The fractional concentrations of end-tidal carbon dioxide (FET_CO2) and oxygen (FET_O2) were sampled from the tracheal cannula by using rapidly responding CO₂ (Ametek CD-3A) and O₂ (Beckman OM-11) analyzers. Rectal temperature was monitored and maintained at 37.5 ± 0.5°C with the use of a servo-controlled heating lamp (Yellow Springs Instruments model 73A). Blood pressure was measured with a pressure transducer (Gould P23XL) that was attached to a femoral artery cannula.

Arterial blood samples were withdrawn during the steady state of each experimental intervention (see Experimental protocol) and analyzed for PO₂, PCO₂, and pH (Cameron Instrument model BGM). The blood gas and pH values were corrected using the rectal temperature that was recorded at the time of sampling. Bicarbonate values were calculated by using measured pH and PCO₂ values, and if a base deficit existed, it was corrected by infusing sodium bicarbonate intravenously.

Electromyographic (EMG) and single motor unit recordings. The EMG was recorded from the external oblique muscle using a pair of platinum alloy subdermal electrodes (Grass Instruments, Quincy, MA), which were placed 3-4 mm apart in the muscle belly. The EMG signals were amplified, filtered (30-3,000 Hz) with alternate current (AC)-coupled differential amplifiers (Grass Instruments), rectified, and integrated (Coulbourn S76-01, time constant 100 ms).

Single motor unit recordings were made from the ipsilateral external oblique muscle by using monopolar tungsten electrodes (A-M Systems, impedance 5 M). These recordings were amplified (Grass Instruments) and filtered (300 Hz-10 kHz) by using an AC-coupled differential amplifier.

The analog signals were converted to digital signals (Vetter Digital, model 4000) before storage on videocassette tape, so that off-line analysis of motor unit activity could be performed. In addition, integrated EMG (iEMG) activity, airflow, arterial pressure, and end-tidal CO₂ and O₂ were monitored and recorded continuously on a polygraph (Grass Instruments, model RPS7C).

Experimental protocol. Motor unit activities in the external oblique muscle were identified while the cats expired against a positive end-expiratory pressure (PEEP) of 1-2.5 cmH₂O (12). PEEP was imposed by immersing a rubber tube, which was attached to the tracheal cannula, into a water-filled flask. PEEP was employed while we searched for motor units because it is known to increase abdominal muscle activity (12). Therefore, the probability of localizing a motor unit without eliciting any of the side effects that are often associated with prolonged exposure to other respiratory stimuli (i.e., hypoxia) was increased. The length of exposure to a given respiratory stimulus was a concern because we often required >2 h to identify a single motor unit. Subsequent to localization of unit activity, PEEP was removed, and recordings were made continuously for 3-4 min at successive FET_O2 levels of 1.0 (hyperoxia), 0.21 (normoxia), and 0.08-0.10 (hypoxia). During hypoxia, motor unit activities were recorded under both hypocapnic and isocapnic conditions. The isocapnic level was determined during the steady state of hyperoxia and was maintained throughout hypoxia by adding CO₂ to the inspired gas mixture. After unit activity at the three levels of inspired FET_O2 was recorded, the cats expired against 5, 10, 15, and 20 cmH₂O PEEP to evoke the maximal iEMG activity of the external oblique muscle. We assumed that maximal activity was achieved in all the animals when a further increase in PEEP was not accompanied by further changes in iEMG activity. In all of the animals, an increase in PEEP beyond 20 cmH₂O did not produce an increase in iEMG activity.

Data analysis. Six respiratory cycles of taped data were sampled (sampling frequency 1.8 kHz) from the last minute of each experimental intervention and stored on a microcomputer hard drive. A data-acquistion and analysis program (EGAA, RC Electronics) was then used to measure the inspiratory duration (TI), expiratory duration (TE), and tidal volume (VT) that corresponded to each breath. TI and TE values were used to calculate breathing frequency (f_b), which was used in conjunction with VT to calculate inspired minute ventilation (I).

Single motor units were discriminated by using either an amplitude threshold or a waveshape recognition pattern (RC Electronics). The digitized pulses derived from discrimination of the motor units were used to calculate the following variables: 1) the onset time of expiratory motor unit activity in relation to airflow; 2) the number of motor unit spikes per respiratory cycle (n/cycle); 3) the mean discharge frequency per respiratory cycle (f/cycle); 4) the number of motor unit spikes during each 10% epoch of TE; and 5) the discharge frequency during each 10% epoch of TE. The total number of motor units active and the number of motor units active during each 10% epoch of the expiratory cycle were also calculated.

Mean arterial blood pressure, peak iEMG activity (expressed as a percentage of the maximum EMG), FET_CO2, and FET_O2 were measured for the corresponding six respiratory cycles from the polygraph record. Arterial blood gas and pH values corresponding to each epoch were recorded. A mean value for each of these physiological variables was then calculated for each six-breath epoch. Subsequently, a group mean value was calculated for each of the experimental interventions.

A one-way analysis of variance with repeated measures was used to determine whether the mean values calculated for each experimental intervention were significantly different. If the analysis of variance revealed that a significant difference existed between the levels of the main factor, the means were compared by using the Student-Newman-Keuls post hoc test.

A 4 × 10 experimental design was used to compare the mean discharge frequencies that were calculated for the two factors, i.e., experimental intervention and TE epoch. The levels of the main factor experimental intervention were isocapnic hyperoxia, isocapnic normoxia, hypocapnic hypoxia, and isocapnic hypoxia. The levels of the main factor TE epoch were 0, 10, 20 ....... 100%. The design was also used to compare the mean number of motor unit spikes recorded during each 10% epoch of TE for each experimental intervention.

A two-way analysis of variance with repeated measures was used to determine whether a main effect (experimental intervention and TE epoch) or an interaction effect (experimental intervention × TE epoch) was significant. If a main or interaction effect was significant, the means were compared by using the post hoc test stated above. Linear regression analysis was used to determine the correlation between unit discharge frequency and peak iEMG activity. All values are presented as means ± SE, and the level of significance chosen was P < 0.05.

RESULTS

Ventilation and cardiovascular parameters. Table 1 shows the mean steady-state ventilation and cardiovascular parameters that were calculated for 31 trials completed under each experimental condition. The significant increases recorded during hypocapnic and isocapnic hypoxia were elicited solely by an increase in VT. The f_b recorded during hypocapnic hypoxia was significantly less than the values recorded during the other three experimental conditions. The decrease in f_b recorded during hypocapnic hypoxia was the result of a significant increase in TE; f_b did not change during isocapnic hypoxia because TI and TE decreased and increased, respectively, compared with the values recorded during hyperoxia and normoxia. There were no significant changes in mean arterial pressure.

Table 1. Cardiorespiratory variables recorded at each level of fractional concentration of end-tidal O2 Condition n Ventilatory and Cardiovascular Parameters Arterial Blood Gas Values VT, ml f, breaths/min  I, l/min TI, s TE, s MAP, mmHg pHa PaO2, Torr PaCO2, Torr 100% O2, isocapnic 31 42.78 ± 2.70  22.28 ± 0.91  0.94 ± 0.06  1.28 ± 0.07  1.53 ± 0.09  114.29 ± 5.54  7.27 ± 0.02  311 ± 6.7  36.0 ± 1.4  21% O2, isocapnic 31 45.28 ± 2.78  22.40 ± 0.76  1.00 ± 0.06  1.29 ± 0.06  1.47 ± 0.08  119.90 ± 5.70  7.31 ± 0.02  86.1 ± 2.3* 35.8 ± 1.4  8-10% O2 31 60.57 ± 3.74* 20.49 ± 0.67* 1.24 ± 0.08* 1.31 ± 0.05  1.71 ± 0.08* 118.97 ± 6.47  7.38 ± 0.02* 31.1 ± 0.9* 28.7 ± 0.9* 8-10% O2, isocapnic 31 80.67 ± 4.81*§ 21.61 ± 0.53  1.75 ± 0.11*§ 1.20 ± 0.04*§ 1.63 ± 0.06*§ 121.55 ± 6.52  7.31 ± 0.02  34.6 ± 1.1* 35.7 ± 1.3 Values are means ± SE; n, no. of trials that were performed at each level of fractional concentration of end-tidal O2. VT, tidal volume; f, breathing frequency; I, inspiratory minute ventilation; TI, inspiratory duration; TE, expiratory duration; MAP, mean arterial pressure; PaCO2, arterial partial pressure of CO2; PaO2, arterial partial pressure of O2; pHa, arterial pH. P < 0.05: * Significantly different from isocapnic hyperoxia; significantly different from isocapnic normoxia; significantly different from isocapnic hypoxia; § significantly different from hypocapnic hypoxia.

Influence of hypoxia on whole external oblique EMG activity. During hyperoxia, normoxia, and isocapnic hypoxia, phasic activity of the whole external oblique muscle was evident during 20/31, 21/31, and 31/31 trials, respectively. Figure 1, left, shows an example of an external oblique iEMG that was recorded from one cat during one trial of steady-state isocapnic normoxia. The iEMG was characterized by a steeply rising phase followed by either a plateau or a less steeply rising phase of neural activity that abruptly declined before the onset of inspiration.

Figures 1 and 2 show that isocapnic hypoxia caused the slope of the rising phase and the peak amplitude of the iEMG to increase significantly. The peak iEMG amplitude recorded during hypocapnic hypoxia was not significantly different from the values recorded during hyperoxia and normoxia (Fig. 2). This occurred because during the transition from isocapnic hypoxia to hypocapnic hypoxia iEMG activity decreased in 26 of 31 trials.

Influence of hypoxia on external oblique motor unit activities. A total of 35 external oblique motor units, with phasic expiratory activity, were recorded during 31 experimental trials in 10 cats. The average number and the range of motor units that were obtained from each cat were 3.5 ± 0.5 and 1-6, respectively. The activity of each unit was evident while the cats expired against a PEEP of 1-2 cmH₂O. However, only 16 of 35 and 20 of 35 of these units were active during the steady state of hyperoxia and normoxia, in the absence of PEEP. The number of active units increased to 29 in isocapnic hypoxia. Of the six units that did not discharge during isocapnic hypoxia, one unit was inactive at all levels of FET_O2, whereas the remaining five units were active only during hyperoxia and normoxia. A total of 18 units were active during hypocapnic hypoxia, because the activity of 11 of 29 units was abolished by the transition from isocapnic to hypocapnic hypoxia.

Figure 3 shows that the onset time of unit discharge during each experimental intervention varied; some motor units commenced firing during the first 10% of TE, whereas others did not discharge until 70% of expiration was completed. However, the greatest percentage of the total population of motor units active during each experimental intervention did not commence firing until ~30% of TE was completed (Fig. 3). In addition, the average onset times calculated for each experimental condition (isocapnic hyperoxia = 30.0 ± 3.8% of TE, isocapnic normoxia = 37.5 ± 4.0% of TE, hypocapnic hypoxia = 36.1 ± 3.3% of TE, isocapnic hypoxia = 29.6 ± 2.3% of TE) were not significantly different.

The group mean discharge frequency and the motor unit spike count during each 10% epoch of TE are presented in Fig. 4. During each experimental intervention, the discharge frequency and motor unit spike count increased abruptly at 30% of TE; this phase was followed by a gradual increase in the discharge frequency and unit spike count, which reached a maximum between 60 and 70% of TE. Between 30 and 90% TE, the mean frequency recorded during isocapnic hypoxia was greater than the values recorded during hyperoxia and normoxia. Diminutions of FET_CO2 to hypocapnic levels during hypoxia abolished this difference.

Figure 5 shows the average f/cycle and n/cycle values for each experimental intervention. A significant increase in both variables accompanied the transition from hyperoxia and normoxia to isocapnic hypoxia.

The discharge frequency of the motor units that were active during both normoxia and isocapnic hypoxia is plotted as a function of the peak iEMG amplitude in Fig. 6. There was a significant correlation between the two variables (r = 0.76, r² = 0.57). The r² value suggests that rate coding accounted for almost 60% of the increase in iEMG amplitude.

DISCUSSION

This investigation examined the behavior of external oblique motor units at three levels of inspired FET_O2 in 10 spontaneously breathing anesthetized cats. Our experiments lead to four important observations: 1) at each level of FET_O2, the onset of unit discharge did not commence until ~30% of expiration was completed; 2) after the onset of activity, the motor unit firing was characterized by an initial abrupt increase in discharge frequency, which then continued to increase gradually or remained constant before declining abruptly at 90-100% of TE; 3) isocapnic hypoxia caused an increase in f/cycle and n/cycle in units that were active during isocapnic hyperoxia and normoxia and lead to the recruitment of units that were inactive under the latter conditions. The increase in f/cycle accounted for ~60% of the increase in peak iEMG amplitude that occurred as a result of hypoxia; 4) abolition of motor unit activity or a decrease in discharge frequency accompanied the transition from isocapnic to hypocapnic hypoxia.

Critique of the methods. Tungsten microelectrodes were used to record from either single motor units or a burst of no more than three units during the conditions imposed throughout the experimental protocol. If a dense burst of motor units was recorded during any condition (usually during isocapnic hypoxia), the experimental trial was abolished because 1) the software package that we employed could not accurately discriminate the selected unit and 2) it is our experience that visual discrimination of a single unit from a dense burst of units is often subjective and not reproducible. Therefore, at the experimental stage of the investigation, we rigorously selected those trials that would be included in the subsequent data analysis.

It may be argued that the transversus abdominis or internal oblique muscle is a more representative model of expiratory muscle activity and that the selection of the external oblique muscle could account for the reduced expiratory activity recorded during hyperoxia and normoxia. However, an investigation completed by Fregosi (5) using an animal preparation identical to that employed in the present investigation showed that the transversus abdominis, internal oblique, and external oblique muscles behaved similarly during baseline conditions and chemoreceptor stimulation. Therefore, we believe that the external oblique muscle contributes significantly to expiration in the spontaneously breathing anesthetized cat.

One of the goals of the present investigation was to compare and contrast the results with the findings of Fregosi et al. (7). Therefore, it may be argued that to make these comparisons hypercapnia should have been used in the present investigation to elicit increases in iEMG activity, since this was the stimulus employed by Fregosi et al. (7). However, Fregosi (5) showed that hypoxia and hypercapnia often produce a similar change in the amplitude of the abdominal EMG of spontaneously breathing vagally intact animals. Therefore, because these two stimuli reportedly cause a similar increase in iEMG amplitude, and because we were interested in studying the interactive effects of O₂ and CO₂ (isocapnic hypoxia vs. hypocapnic hypoxia) in a spontaneously breathing preparation, we selected hypoxia as the primary respiratory stimulus.

External oblique whole muscle and motor unit discharge characteristics. The discharge pattern of the whole external oblique muscle that was recorded during the present investigation was characterized by a steeply rising phase followed by either a plateau or a less steeply rising phase of neural activity that abruptly declined before the onset of inspiratory flow. Although this pattern of activity was similar at each level of FET_O2, a subjective assessment of the rate of rise at the onset of discharge and a quantitative measure of peak iEMG activity revealed that these two variables were significantly greater during isocapnic hypoxia compared with hyperoxia and normoxia. This finding is similar to the results obtained from many investigations, which show that isocapnic hypoxia elicits an increase in abdominal muscle or nerve activity in vagally intact spontaneously breathing animals that are awake (16, 20) or anesthetized (5) and in decerebrate or anesthetized preparations that are paralyzed and ventilated (7, 8, 10).

The iEMG pattern of activity that was evident at all levels of FET_O2 was caused primarily by changes in motor unit discharge frequency, which increased abruptly at 30-40% of TE and then continued to increase gradually or remained constant before declining abruptly at 90-100% of TE. A similar onset-time and discharge pattern has been recorded from internal intercostal (14) and parasternal intercostal motor units (19) of spontaneously breathing cats and humans, respectively.

In contrast to the findings of the present investigation, Fregosi et al. (7) reported that the pattern of transversus abdominis motoneuronal activity is characterized by a marked increase in discharge frequency at the start of the expiratory period, which gradually declines during the remainder of the period. It is possible that the findings of Fregosi et al. may have been the result of disinhibition of transversus abdominis motoneurons, which occurred because inhibitory synaptic inputs were removed subsequent to decerebration and thoracotomy.

The increase in the rate of rise and peak amplitude of expiratory iEMG activity during hypoxia was mediated by both rate coding and recruitment of single motor units. During hyperoxia and normoxia, only 16 of 35 and 20 of 35 of the motor units tested were active, respectively. The transition from hyperoxia or normoxia to isocapnic hypoxia was accompanied by an increase in the discharge frequency of the units that were already active and by the recruitment of additional motor units. The correlation of motor unit discharge frequency and peak iEMG activity, which was calculated for the motor units that were active during both isocapnic normoxia and isocapnic hypoxia, suggests that motor unit rate coding was responsible for close to 60% of the increase in iEMG activity that was observed.

To our knowledge, the response of abdominal motor unit activities to hypoxia has not been described previously. However, studies of external oblique and internal intercostal motor units recorded from anesthetized spontaneously breathing cats (12, 14) during expiratory occlusion and hypoxia, respectively, yielded similar results. Fewer external oblique and internal intercostal motor units were active during quiet breathing in those experiments, but the discharge of these units increased, and additional units were recruited during expiratory airway occlusion and hypoxia. Fregosi et al. (7) suggested that only a few motor units were active during quiet breathing in Sears' (14) investigation because pentobarbital anesthesia selectively depresses internal intercostal nerve activity more than phrenic nerve activity in decerebrate unanesthetized cats (6). This suggestion might also account for the low number of motor units that were active during normoxia in the investigation by Russell et al. (12) and the present investigation, since the animals were anesthetized with allobarbital and urethan and with -chloralose and urethan, respectively. Alternatively, the lack of unit activity might have occurred because expiration during quiet breathing was passive.

In contrast to the results obtained from previous investigations (12, 14) and the present investigation, Fregosi et al. (7) reported that all transversus abdominis motoneurons that were recorded in their investigation were active during normocapnia and that the transition to hypercapnia was not accompanied by the recruitment of additional motoneurons. However, it is possible that the motoneurons studied in the Fregosi et al. investigation were recruited during normocapnia, because the arterial PCO₂ level that was selected and maintained for the decerebrate, paralyzed, and ventilated animals was significantly greater than each animal's apneic threshold, which was probably low because of the removal of inhibitory drives secondary to decerebration, thoracotomy, and vagotomy.

The increase in abdominal EMG activity recorded during isocapnic hypoxia in the present investigation was abolished by hypocapnia, despite the finding that VT under hypocapnic conditions was significantly greater than the values recorded during hyperoxia and normoxia. The reduction in external oblique EMG activity was caused by the complete abolition of unit activity (derecruitment) and a decrease in the discharge frequency of the motor units that remained active. This latter finding is similar to the findings of Russell et al. (12), which showed that external oblique motoneuron activity that was present at the onset of artificial hyperventilation was abolished within 3 min because of the reduction in PCO₂ that was induced by hyperventilation.

Regulation of expiratory motor unit activities during hypoxia. Abdominal motoneurons are subjected to various excitatory inputs in the spontaneously breathing animal. The changes in motor unit activity that were observed in the present investigation may have been generated directly by an increase in bulbospinal expiratory neuronal activity that was elicited by the hypoxia-induced activation of the peripheral chemoreceptors (3, 11, 17). In support of this hypothesis, Ledlie et al. (10) showed that hypoxia increases abdominal motor activity despite the elimination of vagal and muscle proprioceptive feedback. Alternatively, activation of the external oblique muscle may have been caused by inspiratory muscle activation, which, by producing an increase in VT, stimulated the activity of the external oblique motor units via either a vagal afferent feedback loop (1, 5, 12, 20) or through a proprioceptive mechanism that was present at the segmental or suprasegmental level (5, 15, 20). The earlier work of Bishop (2) as well as more recent investigations (5, 20) support a role for vagal mechanisms. Thus vagotomy will reduce or abolish the hypoxia-induced increase in abdominal activity in awake and anesthetized animals, even when the influence of muscle proprioceptive feedback is controlled by making pre- and postvagotomy comparisons at constant levels of VT.

Despite the presence of these segmental and/or suprasegmental excitatory drives, their effect on the respiratory-related activity of the abdominal motor units during isocapnic hypoxia was dependent on the presence of a facilitatory drive that was CO₂ dependent. The withdrawal of the CO₂ drive, which occurred as a result of the hyperventilatory response to hypoxia, resulted in the attenuation or total abolition of the central respiratory drive to the external oblique motor units. This response occurred, even though excitatory inputs to inspiratory spinal motoneurons still existed, as inspired VT was significantly greater during hypocapnic hypoxia compared with isocapnic hyperoxia and normoxia. This finding supports St. John's observation (17) that expiratory motoneurons have a weaker response to hypoxia than inspiratory neurons. Indeed, the activities of 5 of 35 motor units that we studied were abolished by hypoxia. This observation helps to explain the marked variability of abdominal muscle activities during hypoxia that has been reported by many investigators (8, 16, 18).

Conclusion. We conclude that external oblique motor units are activated during the latter two-thirds of expiration. Moreover, our estimates suggest that rate coding accounts for ~60% of the increase in iEMG amplitude in hypoxia and that the recruitment of motor units contributes to the remainder of this increase. In addition, we demonstrated that excitation of motor units during hypoxia is critically dependent on changes in CO₂ and/or VT.

ACKNOWLEDGEMENTS

We thank David D. Fuller for technical assistance.

FOOTNOTES

Address for reprint requests: R. F. Fregosi, Dept. of Physiology-Gittings Bldg., The University of Arizona, Tucson, AZ 85721-0093.

Received 5 June 1996; accepted in final form 5 August 1996.

REFERENCES

1.	Arita, H., and B. Bishop. Responses of cat's internal intercostal motor units to hypercapnia and lung inflation. J. Appl. Physiol. 54: 375-386, 1983.
2.	Bishop, B. Reflex control of abdominal muscles during positive-pressure breathing. J. Appl. Physiol. 19: 224-232, 1964.
3.	Davies, R. O., and M. Kalia. Carotid sinus nerve projections to the brainstem in the cat. Brain Res. Bull. 6: 531-541, 1981.
4.	Fregosi, R. F. Changes in the neural drive to abdominal expiratory muscles in hemorrhagic hypotension. Am. J. Physiol. 266 (Heart Circ. Physiol. 35): H2423-H2429, 1994.
5.	Fregosi, R. F. Influence of hypoxia and carotid sinus nerve stimulation on abdominal muscle activities in the cat. J. Appl. Physiol. 76: 602-609, 1994.
6.	Fregosi, R. F., and D. Barlett, Jr. Internal intercostal nerve discharges in the cat: influence of chemical stimuli. J. Appl. Physiol. 66: 687-694, 1989.
7.	Fregosi, R. F., J. C. Hwang, D. Barlett, Jr., and W. M. St. John. Activity of abdominal muscle motoneurons during hypercapnia. Respir. Physiol. 89: 179-194, 1992.
8.	Fregosi, R. F., S. L. Knuth, D. K. Ward, and D. Barlett, Jr. Hypoxia inhibits abdominal expiratory nerve activity. J. Appl. Physiol. 63: 211-220, 1987.
9.	Kelsen, S. G., M. D. Altose, and N. S. Cherniack. Interaction of lung volume and chemical drive on respiratory muscle EMG and respiratory timing. J. Appl. Physiol. 42: 287-294, 1977.
10.	Ledlie, J. F., A. I. Pack, and A. P. Fishman. Effects of hypercapnia and hypoxia on abdominal expiratory nerve activity. J. Appl. Physiol. 55: 1614-1622, 1983.
11.	Lipski, J., A. Trezebski, J. Chodobska, and P. Kruk. Effects of carotid chemoreceptor excitation on medullary expiratory neurons in cat. Respir. Physiol. 57: 279-291, 1984.
12.	Russell, J. A., B. P. Bishop, and R. E. Hyatt. Discharge of abdominal muscle - and -motoneurons during expiratory loading in cats. Exp. Neurol. 97: 179-192, 1987.
13.	Saupe, K. W., C. A. Smith, K. S. Henderson, and J. A. Dempsey. Respiratory muscle recruitment during selective central and peripheral chemoreceptor stimulation in awake dogs. J. Physiol. Lond. 448: 613-631, 1992.
14.	Sears, T. A. Efferent discharges in alpha and fusimotor fibers of intercostal nerves of the cat. J. Physiol. Lond. 174: 295-315, 1964.
15.	Sears, T. A. Some properties and reflex connexions of respiratory motoneurons of the cats thoracic spinal cord. J. Physiol. Lond. 175: 386-403, 1964.
16.	Smith, C. A., D. A. Ainsworth, K. S. Henderson, and J. A. Dempsey. Differential responses of expiratory muscles to chemical stimuli in awake dogs. J. Appl. Physiol. 66: 384-391, 1989.
17.	St. John, W. M. Respiratory neuron responses to hypercapnia and carotid chemoreceptor stimulation. J. Appl. Physiol. 51: 816-822, 1981.
18.	Takasaki, Y., D. Orr, J. Popkin, A. Xie, and T. D. Bradley. Effect of hypercapnia and hypoxia on respiratory muscle activation in humans. J. Appl. Physiol. 67: 1776-1784, 1989.
19.	Whitelaw, W. A., and T. W. J. Watson. Spike trains from single motor units in human parasternal intercostal muscles. Respir. Physiol. 88: 289-298, 1992.
20.	Yasuma, F., J. Kimoff, L. F. Kozar, S. J. England, T. D. Bradley, and E. A. Phillipson. Abdominal muscle activation by respiratory stimuli in conscious dogs. J. Appl. Physiol. 74: 16-23, 1993.