Neural drive to inspiratory pump muscles is increased under many pathological conditions. This study determined for the first time how neural drive is distributed to five different human inspiratory pump muscles during tidal breathing. The discharge of single motor units (n = 280) from five healthy subjects in the diaphragm, scalene, second parasternal intercostal, third dorsal external intercostal, and fifth dorsal external intercostal was recorded with needle electrodes. All units increased their discharge during inspiration, but 41 (15%) discharged tonically throughout expiration. Motor unit populations from each muscle differed in the timing of their activation and in the discharge rates of their motor units. Relative to the onset of inspiratory flow, the earliest recruited muscles were the diaphragm and third dorsal external intercostal (mean onset for the population after 26 and 29% of inspiratory time). The fifth dorsal external intercostal muscle was recruited later (43% of inspiratory time; P < 0.05). Compared with the other inspiratory muscles, units in the diaphragm and third dorsal external intercostal had the highest onset (7.7 and 7.1 Hz, respectively) and peak firing frequencies (12.6 and 11.9 Hz, respectively; both P < 0.05). There was a unimodal distribution of recruitment times of motor units in all muscles. Neural drive to human inspiratory pump muscles differs in timing, strength, and distribution, presumably to achieve efficient ventilation.
- motor unit
- respiratory control
- inspiratory muscles
human breathing is an essential rhythmic activity that does not need volition, but it does require shortening of many inspiratory muscles that act on the chest wall (diaphragm, intercostal, and scalene muscles, the so-called “pump” muscles) and those that act on the upper airway (“valve” muscles). The upper airway must be maintained patent while the muscles acting on the chest wall generate forces to overcome airway resistance and the elastic recoil of the chest wall and lungs to allow inspiratory flow. Neural drive from medullary respiratory centers (14, 29) acting via central respiratory drive potentials depolarizes and thus recruits inspiratory motoneurons innervating chest wall muscles and derecruits them by active hyperpolarization (e.g., Refs. 1, 36). In contrast, the recruitment of upper airway motoneurons is achieved with inspiratory depolarization but without expiratory inhibition (32). The timing of drive to the different inspiratory muscles is thought to be controlled by one or more pontomedullary oscillators (16, 24, 30, 37).
Recordings from single motor units have shown that human diaphragm motor units discharge phasically during inspiration in quiet breathing (e.g., Ref. 10), CO2-driven ventilation (17), and voluntary breathing with different inspiratory flows and volumes (3), with little tonic activity during expiration. There is evidence that the timing of inspiratory activity varies systematically among intercostal muscles in animals (for review, see Refs. 9, 26) and humans (8, 18). Thus, for a particular set of intercostal muscles, motor units in muscles with the largest inspiratory mechanical advantage, i.e., those muscles that produce the largest change in pleural pressure per unit muscle mass and unit muscle tension (12, 41), discharge earlier and reach higher discharge frequencies. However, the recruitment differences between the various pump muscles have not been investigated.
We hypothesized that there would be differences in the timing of firing of single motor units in the various inspiratory pump muscles and that the firing frequencies would differ between muscles. This information provides insight into the descending respiratory “commands” that drive inspiration. Hence, the present study was conducted to determine the firing behavior of single human diaphragm motor units during quiet breathing and to compare this with the motor unit behavior of the scalenes, parasternal intercostal, and external intercostal muscles, all of which also discharge during quiet inspiration. Second, we looked for evidence that there is a bimodal distribution of distinct “early” and “late” firing times during inspiration, as has been reported for inspiratory motoneurons in animals (e.g., Refs. 13, 23). The data allowed us to make a comparison between the behavior of the five inspiratory pump muscles recently reported for the genioglossus, a major dilator of the upper airway (34). A preliminary report of some findings has been presented (35).
The present investigation is based on a complete reanalysis of motor unit recordings obtained during normal quiet breathing in two previous experiments (8, 17). They investigated the mean peak firing rates of single motor units in the costal diaphragm, scalenes, and parasternal intercostal muscles with and without increased chemical drive (17) and the mean peak firing rates in the dorsal external intercostal muscles during quiet breathing (8). This reanalysis allowed us to compare the firing times and frequencies of single motor units sampled in five major inspiratory muscles: the costal diaphragm, the scalenes, the second parasternal intercostal muscle, and the third and fifth dorsal external intercostal muscles. Studies were performed on five healthy men (aged 33–51 yr). Three subjects were studied in both experiments. Each subject gave informed, written consent to the procedures approved by the human ethics committee of the University of New South Wales. Key methods are given below, and further details of the recordings are available (8, 17).
Each muscle was studied on a separate occasion. With the exception of the scalene, ultrasonography assisted placement of needle electrode in the other muscles. Subjects were seated comfortably, unrestrained, and breathed quietly through a mouth piece (Fig. 1). Tidal volume was obtained by integration of the flow signal. Inspiratory time (Ti) was measured from the airflow signal. To compare between subjects and between motor units, the times at which the units discharged were expressed as a percentage of the total Ti (%Ti). Respiratory movements of the chest and abdomen were recorded with inductance bands around the chest and abdomen. The gains of the two signals were calibrated using an isovolume maneuver (4). To assess the earliest increases in drive to the five muscles, we also measured the onset of inspiratory activity using the multiunit electromyographic (EMG) recordings. The multiunit activity was rectified and integrated (time constant, 50 ms), and the onset time (relative to inspiratory airflow) was measured for each breath at the increase in activity above tonic or baseline levels.
EMG recording techniques.
Motor unit EMG activity was recorded using a monopolar needle (Medelec DMG 5; Surrey, UK). A reference surface electrode was positioned ∼3 cm from the recording needle. A large flexible ground was positioned on the right shoulder. All recordings were made on the right side. Before recording from the diaphragm and parasternal intercostal muscles, a small injection of local anesthetic was made at the site of insertion (1–1.5 ml of lignocaine, 1–2% with epinephrine). The injection was made superficial to the muscle being studied. The injection is unlikely to have blocked any motor axons, but if it did, this would not alter the activity of any units that were recorded.
For the diaphragm, the needle was inserted perpendicular to the skin in the seventh or eighth intercostal space, in the anterior axillary line. For the scalenes, the needle was inserted ∼2 cm above the clavicle in the posterior triangle of the neck, with the location defined by palpation during voluntary sniffs. For the parasternal intercostal muscles, measurements were made in the second intercostal space (∼2 cm lateral to the sternal edge). The electrode was inserted at 45°, with the needle tip pointing toward the sternum. For the dorsal external intercostals, the electrode was inserted into the dorsal portion of the third interspace (∼2 cm cranially and 2 cm medial to the medial border of the spine of the scapula). The recording position for the fifth dorsal external intercostal was similarly located two spaces lower.
For all muscles, once a suitable recording site had been secured, at least 10 stable breaths were recorded during quiet breathing before the electrode was moved to a new site. Approximately 10 recording sites were studied for each muscle for each subject, with the needle moved away from clearly isolated units after each site. Subjects had minimal or no discomfort throughout the procedures, and no feedback of breathing was provided. Each site was selected for formal recording when there was clear multiunit EMG activity (Fig. 2). The activity of the motor units was monitored continually via an oscilloscope (5 ms/div), and signals were monitored online with the data acquisition system. The EMG signals were amplified (×1,000–10,000), filtered (53 Hz to 3 kHz), and stored on computer (Spike2 with 1401 interface, CED) for offline analysis. Spikes from motor units were selected based on a threshold crossing, and they were manually sorted using individual “templates” based on their size and morphology.
Measurements of discharge properties.
Measurements of the discharge properties of the single motor units were based on the instantaneous frequency plots for each motor unit over three consistent consecutive breaths. The three consecutive breaths chosen had consistent respiratory timing and changes in inspiratory flow and volume. All the units for each muscle were classified into tonic or phasic categories, depending on whether they discharged throughout both inspiration and expiration. To be classified as “tonic,” units had to maintain a minimal firing frequency that exceeded 2.0 Hz. The onset time, relative to the onset of inspiratory airflow, was taken as the time that the unit first began to discharge or the time the unit increased its discharge frequency above the background tonic firing rate determined visually. The end discharge time was taken as the time of the last discharge of the unit in each breath or when the discharge frequency of the unit returned to tonic values. The preinspiratory activity occurred when activity began before inspiratory flow (<0% Ti). Likewise, postinspiratory activity occurred when activity continued after 100% Ti. The onset discharge frequency was calculated from the frequency of the first interspike interval within each breath. For tonic units, it was measured at the first increase of the instantaneous frequency above the tonic level. The end discharge frequency was determined with a similar procedure. The peak discharge frequency was calculated with a running average of over 200 ms for each breath and then averaged across the three breaths. This process minimizes the effect of any motor unit firing missed in the spike discrimination analysis. All spike analyses were performed by a single observer.
In addition, to assess the initial recruitment time for the whole population of motor units sampled at each site, multiunit EMG was rectified and integrated (decay time constant, 50 ms), and the onset time was measured for each breath at the increase in activity above tonic or baseline levels and expressed relative to the onset of inspiratory flow (0% Ti). Figure 2 shows examples of the integrated selective multiunit EMG records for each muscle. The initial increase in multiunit EMG activity from each muscle was sorted based on its onset time, and then each data point was joined by an unbroken line (see Fig. 3). This gives the cumulative data for the entire population of onset times.
The time of discharge (onset, peak, and end) and discharge frequencies of single motor units, tidal volume, and Ti were compared between the five testing sessions using one-way ANOVA tests with Student-Newman-Keuls post hoc analysis. If data were not normally distributed, then a Kruskal-Wallis test was applied with Dunn's post hoc analysis. For units in each muscle, Pearson product-moment correlations were performed on the onset time (% Ti), the duration of firing, and the peak discharge frequencies. The proportion of phasic and tonic units and the presence of preinspiratory/postinspiratory activity in each muscle were analyzed with χ2 tests. In addition, cross correlations between the signal of lung volume (from the summed signal of the inductance bands) and instantaneous firing frequency (smoothed over 200 ms) were computed for all potential phase variations between the two signals. The strength of each correlation was calculated with a linear coefficient of determination (r2). The respiratory phase of the maximal r2 determined whether the unit was an inspiratory discharging unit (31, 34). Statistical significance was set at P < 0.05. Unless indicated, values are given as means ± SE.
Data were obtained from a total of 280 single motor units in five subjects with recordings from the diaphragm (40 U), scalene (56 U), second parasternal intercostal (69 U), third dorsal external intercostal (67 U), and fifth dorsal external intercostal (48 U). There was no significant difference for inspiratory flow, Ti, or tidal volume when recordings were made from different muscles (see Table 1).
Typical recordings for the five muscles are shown in Fig. 2. Overall, 239 (85%) of the motor units discharged phasically during inspiration, whereas 41 motor units (15%) discharged throughout inspiration and tonically during expiration. Figure 3 plots the timing of discharge of each motor unit in each muscle. All motor units that discharged tonically increased their discharge frequency during inspiration (Fig. 2F). The degree of tonic firing differed between the inspiratory muscles (P < 0.001). In the diaphragm, all units were phasically active with no tonic activity, and in the third dorsal external intercostal the units were predominately phasic, with only one motor unit tonically active. In contrast, 29% of scalenes (n = 16), 13% of parasternal intercostal units (n = 9), and 31% of fifth dorsal external intercostal units (n = 15) discharged tonically.
Timing of inspiratory discharge.
There were significant differences between the inspiratory muscles in the timing of inspiratory discharge. In addition to the discharge timing for each single motor unit, Fig. 3 illustrates the onset timing determined from the selective multiunit EMG recordings. The thin dotted horizontal lines indicate those units that discharged tonically, and the thick horizontal lines show the time that the discharge frequency increased phasically during inspiration. The solid ascending curves represent the cumulative onset times of the integrated multiunit EMG from each recording site. The progressive recruitment of the single units and the multiunit EMG shows that there was a unimodal distribution of recruitment times throughout inspiration for all muscles (see Fig. 4). Figure 4 illustrates the distribution of recruitment times in absolute time (ms), whereas the times quoted elsewhere are expressed as a %Ti.
The mean onset time for phasically active inspiratory units was earliest for the diaphragm units at 26 ± 3% (mean ± SE) of Ti, and latest for the fifth dorsal external intercostal units (43 ± 4% Ti). The diaphragm and the third dorsal external intercostal muscle (29 ± 2% Ti) were recruited significantly earlier than units from the fifth dorsal external intercostal (P < 0.05) but were not recruited significantly earlier than scalene (32 ± 4% Ti) and second parasternal intercostal (34 ± 2% Ti) units. This difference was confirmed by measurement of the timing of the peak correlation between discharge frequency and lung volume (see Table 2). The peak correlation for the phasic units in the diaphragm (−0.69 ± 0.14 s; r2 = 0.78), third dorsal external intercostal (−0.38 ± 0.03 s; r2 = 0.84), scalene (−0.17 ± 0.14 s; r2 = 0.78), and second parasternal intercostal (−0.44 ± 0.31 s; r2 = 0.82) inspiratory units occurred before the end of inspiration, whereas the mean peak correlation for fifth dorsal external intercostal inspiratory phasic units occurred just after the end of inspiration (0.59 ± 0.13 s; r2 = 0.82). These times differed significantly (P < 0.05).
The mean onset of activity derived from the integrated multiunit EMG provided an estimate of the earliest onset of activity for a population of motor units in each muscle. The diaphragm (−2.5 ± 0.6% Ti) was recruited first (earlier than scalenes and fifth dorsal external intercostal; P < 0.05), followed by the third dorsal external intercostal (−1.0 ± 2.6% Ti, earlier than fifth dorsal external intercostal; P < 0.05), second parasternal intercostal (4.4 ± 3.0% Ti), and scalene (7.1 ± 5.0% Ti) muscles (P= not significant). The fifth dorsal external intercostal was recruited last (14.5 ± 3.3% Ti; Fig. 3). Motor unit activity before inspiratory airflow was recorded from each muscle in the multiunit recordings. There were differences between muscles in the likelihood of “preinspiratory” activity (before inspiratory airflow, <0% Ti). The diaphragm multiunit recordings usually had preinspiratory activity (84% of sites; P < 0.001). Preinspiratory activity was less common in the scalene (35%), second parasternal intercostal (41%), and third dorsal external intercostal (46%), and it occurred least of all in the fifth dorsal external intercostal muscle (19%).
Postinspiratory activity (after inspiratory airflow, >100% Ti) also occurred in all muscles but was most obvious in the fifth dorsal external intercostal muscle. In early expiration (measured at 120% Ti), 77% of the fifth dorsal external intercostal motor units were still active, fewer units were active in the other muscles (P < 0.05): diaphragm (50%), third dorsal intercostal (49%), scalenes (45%), and second parasternal intercostal (45%). As a consequence, the mean end firing time for the fifth dorsal external intercostal units (146 ± 5% Ti) was significantly later than for the diaphragm (124 ± 4% Ti), third dorsal external intercostal (125 ± 4% TI), scalene (113 ± 4% Ti), and second parasternal intercostal (127 ± 4% Ti; P < 0.05). For each muscle, there was a significant negative linear correlation between onset time and duration of firing (Fig. 5). Motor units that were recruited early discharged for the longest duration (P < 0.001). The strength of the peak coefficient of determination (r2) between lung volume and each unit's discharge frequency during quiet breathing were assessed (see Table 2). Phasic motor unit had higher r2 values than the tonic units (P < 0.001).
As shown in Fig. 6A, the onset and peak firing frequencies for each phasic motor unit differed significantly between muscles. The mean onset discharge frequencies for the phasic units in the diaphragm (7.7 ± 0.3 Hz) and third dorsal external intercostal (7.2 ± 0.2 Hz) were significantly higher than for the scalene (6.0 ± 0.2 Hz), second parasternal intercostal (6.6 ± 0.2 Hz), and fifth dorsal external intercostal (6.0 ± 0.2 Hz; P < 0.05). The mean peak frequency for the diaphragm (12.6 ± 0.5 Hz) was also higher than for the scalene (9.3 ± 0.4 Hz), second parasternal intercostal (10.8 ± 0.3 Hz), and fifth dorsal external intercostal (10.1 ± 0.4 Hz; P < 0.05). The peak frequency for the third dorsal external intercostal (11.9 ± 0.4 Hz) was higher than the peak frequency for scalene and fifth dorsal external intercostal (P < 0.05). The change in frequency during inspiration from onset frequency to peak frequency was largest for the diaphragm (4.9 ± 0.4 Hz) and third dorsal external intercostal (4.8 ± 0.4 Hz) motor units compared with the scalene (3.3 ± 0.3 Hz), the second parasternal intercostals (4.2 ± 0.3 Hz), and the fifth dorsal external intercostal (4.1 ± 0.4 Hz; P < 0.05). For the tonically active motor units in scalene, parasternal intercostal, and fifth dorsal external intercostal, there was no difference in either the onset or peak firing frequencies (Fig. 6B). The change between the onset and peak firing frequencies for the tonic units (∼4.2 Hz) was not significantly different from that of the phasic units in the same muscle.
To determine whether the inspiratory drive peaked simultaneously for each muscle, the time at which single units reached their maximal firing frequency during inspiration was compared (Fig. 7). Peak firing occurred significantly later for the fifth dorsal external intercostal, at 90 ± 4% Ti, than for the diaphragm (74 ± 2% Ti), the third dorsal external intercostal (72 ± 2% Ti), and the scalenes (76 ± 3% Ti; all P < 0.05), with the peak for the second parasternal intercostal occurring at 81 ± 3% Ti.
For units in the diaphragm, third dorsal external intercostal and second parasternal intercostal, the duration of discharge was positively correlated with the peak firing rate (P < 0.05), and the onset time of firing was negatively correlated with the peak firing rate (P < 0.05). In these muscles, the earliest recruited motor units fired for the longest time and reached the highest peak frequencies. There was no relationship between recruitment time and firing frequency for scalene or fifth dorsal external intercostal muscles.
This study provides evidence for differential distribution of neural drive to the diaphragm and other human inspiratory pump muscles during quiet breathing in humans. The onset, peak, and end motor unit discharge characteristics are similar for the diaphragm and the third dorsal external intercostal muscle (Fig. 7). These motor units discharged earlier and at higher rates than those in the scalene, second parasternal intercostal, and fifth dorsal external intercostal muscles. Unlike the diaphragm and third dorsal external intercostal muscles, which had minimal or no tonic activity, a third of the units were tonically active in the fifth dorsal external intercostal. The fifth dorsal external intercostal motor units were activated significantly later and reached their peak firing later than units in the other muscles. Inspiratory motor units were recruited throughout inspiration with a unimodal distribution of onset times.
Although there was no systematic trend for inspiratory flow, Ti, or tidal volume to differ when recordings were made from different muscles, there was higher inspiratory flow during recordings from the diaphragm compared with the third dorsal external intercostal. However, inspiratory flow was the same during the experiments in which data were collected from the scalene and diaphragm muscles, whereas the onset times were later and firing frequencies were lower for the scalene units than for the diaphragm units. Thus these differences are not likely to have affected our major conclusions.
As for diaphragm motor units in volitional inspiratory tasks in humans (3), all five pump muscles progressively recruited new units throughout inspiration, with no evidence for the bimodal pattern of recruitment seen in animal studies (e.g., Refs. 13, 23). The lack of evidence for a large population of early recruited units is interesting. There are several possible explanations, including a sampling bias. However, the graded recruitment of motor units in all five inspiratory muscles is evident from the onset times of firing of the single motor units as well as the integrated multiunit EMG recordings. Therefore, we do not think that the differences are due to sampling bias.
Based on studies of central respiratory drive potentials in animals (15, 36), the prevailing view suggests that descending drive is temporally uniform across inspiratory motoneuron pools (see also Ref. 39) and that the timing of activation depends on the motoneuron resistance and related intrinsic properties (Ref. 1, for review, see Ref. 33). This view is incompatible with one of our findings for tonically active units. Many of the tonically active units increased their discharge late in inspiration (particularly in the fifth dorsal external intercostal; see Fig. 3). Thus these tonically active motoneurons had consistently responded to an effective suprathreshold net excitatory drive, and so the late increase in inspiratory firing modulation suggests late arrival of descending inspiratory drive at some motoneurons that were already active. We argue that the motoneurons receive inspiratory drive that is unevenly distributed within and between each muscle.
Gradients of inspiratory drive have been observed within the intercostal muscles in humans (8, 18) and animals (11, 26) and are functionally linked to local gradients of mechanical advantage across and along intercostal spaces (for review, see Ref. 9). This suggests that control of the timing and amount of inspiratory activity is programmed at a “pre-”motoneuronal level and may involve the activity of respiratory-related interneurons (6). These gradients of neural drive are centrally determined and persist after the removal of all afferent input in the dog (for review, see Ref. 9). Such a control mechanism when applied to all the inspiratory pump muscles could explain our findings. One potential contributor could be differences in the degree of monosynaptic excitation of inspiratory motoneuron pools from medullary respiratory centers (6, 20, 29, 38).
The firing rates of single motor units differed between the inspiratory pump muscles. They were ∼2 Hz higher in the diaphragm and third dorsal external intercostal muscles. This elevated rate may reflect increased descending drive to the motoneuron pools or differences in the intrinsic properties of the motoneurons (for review, see Ref. 33). There were also large differences in the incidence of tonically firing units across the muscles, with no tonic firing in the diaphragm. The presence of tonic firing may reflect a specific descending tonic drive or separate local tonic activity arising either via persistent inward currents in the specific inspiratory motoneuron dendrites or related excitatory interneurons driving the motoneuron (22). On the other hand, there may be a graded distribution of inhibitory central respiratory drive potentials, which suppresses tonic firing in expiration.
There are major differences between the behavior of motor units innervating the pump muscles and the upper airway muscles such as the genioglossus (34). First, the peak firing frequency of pump muscles (range 8–13 Hz across the five muscles) is well below that of the peak frequencies in genioglossus during quiet breathing in the supine posture (range 15–25 Hz). Second, there is a high population of genioglossus motor units that discharge tonically with no inspiratory modulation (34). Third, the mean onset of firing is earlier for phasic inspiratory motor units in genioglossus (∼20% Ti) than for the earliest recruited pump muscle (diaphragm, 26% Ti, P < 0.001). Such a difference in timing would prepare the airway for inspiratory flow.
Although the method of sampling motor units from humans in vivo to obtain data on samples of inspiratory motoneurons is established (2, 3, 17, 19, 34), it has limitations. First, only single units whose potentials were identifiable from the raw EMG were extracted. Therefore, activity in the lowest threshold units can be missed. To circumvent this problem, we used selective multiunit recordings to assess the earliest onset of inspiratory activity in the vicinity of the electrode. These selective recordings reveal that, in the diaphragm (rather than the other inspiratory muscles), there is early activity in most costal recording sites, with 84% of sites showing multiunit activity before inspiratory airflow (Fig. 3). Nevertheless, new units also continue to be recruited throughout inspiration. Second, an increase in force may be achieved by an increase in firing rate of motor units (rate coding) as well as recruitment of additional units. The diaphragm responds to an increased chemical drive with a larger increase in rates than for muscles such as the scalene (17). The population responses described here provide no information about the balance of rate coding and recruitment for the different muscles. Nevertheless, there were differences in the timing of firing and in the firing frequencies across the different muscles. Our comparison of motor unit behavior between the various inspiratory muscles is made on the assumption that the recordings were acquired during quiet breathing generated by pontomedullary respiratory centers. Electrophysiological and imaging studies in humans support this view (e.g., Refs. 5, 27, 28).
In conclusion, our results suggest that the neural drive is differentially timed to human inspiratory pump muscles. Functionally, this may be important so that the activation of breathing muscles is coordinated to achieve efficient ventilation (7). We propose that this is achieved via a premotoneuronal network that organizes the timing of recruitment of the inspiratory muscles. It is important to determine this because the operation of this system must change in pathological conditions (21, 25, 40, 42).
This work was supported by the National Health Medical Research Council of Australia.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society