Abdominal motor unit activity during respiratory
and nonrespiratory tasks
THREETHAMBAL PUCKREE,1 FRANK CERNY,1 AND BEVERLY BISHOP2
Departments of 1Physical Therapy and Exercise Science and of 2Physiology,
State University of New York at Buffalo, Buffalo, New York 14214
Puckree, Threethambal, Frank Cerny, and Beverly
Bishop. Abdominal motor unit activity during respiratory and
nonrespiratory tasks. J. Appl. Physiol. 84(5): 1707–1715,
1998.—Abdominal muscles serve multiple roles, but the func-
tional organization of their motoneurons remains unclear. To gain
insight, we recorded single motor unit potentials from the inter-
nal oblique (IO) and transversus abdominis (TA) muscles of three
standing subjects during quiet breathing, a leg lift, and an
expiratory threshold load. Inspiratory airflow, recorded from a
pneumotachometer, provided tidal volumes and respiratory cycle
timing. Fine wires, implanted under ultrasonic imaging, detected
single motor unit potentials that were visually distinguished by
their spike morphology. From the number of spikes, firing pro-
files, times of occurrence in the respiratory cycle, and their onset,
instantaneous, mean, and peak firing frequencies we deduced
that 1) breathing patterns varied across tasks, 2) different
motor units were recruited for each task with essentially no
overlap, 3) their firing displayed prominent expiratory activ-
ity during each task, and 4) the recruitment levels and
discharge patterns of IO and TA were different. We conclude
that the IO and TA motor pools receive a strong central
respiratory drive, yet each pool receives its own distinct,
task-dependent synaptic input.
single motor units; abdominal motor control; motor unit
recruitment; firing profiles; task groups; functional organiza-
tion; motor pools
THE ABDOMINAL MUSCLES serve many functions. Yet the
functional organization and neural control of their spinal
motoneurons is poorly understood. Many electromyo-
graphic (EMG) studies have analyzed the output of the
different muscle layers comprising the human abdominal
wall during a variety of respiratory and nonrespiratory
activities including reflex responses (7), speech (6, 10),
postural adjustments and trunk movements (20, 21), volun-
tary activation (4), vomiting (19), and other expulsive
functions (3, 16) (for a recent review see chapt. 4 in Ref. 3).
Results from studies using both surface (27) and indwell-
ing (5) electrodes have suggested that recruitment of the
abdominal muscles is task dependent. How specific abdomi-
nal motor units are activated during different tasks has not
been studied. Therefore, in this study, we have analyzed
and compared the firing patterns of motor units in the
internal oblique (IO) and transversus abdominis (TA)
muscles recorded simultaneously from implanted wires
during quiet breathing (QB), during a sustained leg lift
(LL), and during an expiratory threshold load (ETL) in
three standing adults.
METHODS
Subjects
Three healthy men (23–48 yr of age) with no history of
neuromuscular or cardiorespiratory disease performed each
http://www.jap.org 8750-7587/98 $5.00 Copyright r 1998 the American Physiological Society
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (177.099.005.033) on September 11, 2018.
Copyright © 1998 American Physiological Society. All rights reserved.
task described below at least twice. The study was approved
by the Human Subjects Review Committee of the School of
Health-Related Professions of the State University of New
York at Buffalo. The subjects signed an approved consent
form before participating in the study.
Tasks
The standing subjects were asked to 1) breathe quietly
(QB), 2) extend and hold the right leg 10 in. (,stride length)
forward in a sustained LL while keeping the knee straight,
and 3) expire against a column of water 7.5 cm deep. This
latter task, called ETL, was imposed by inserting a rigid tube
from the expiratory side of a two-way breathing valve into the
water column. These three tasks were chosen because each
resulted in low levels of abdominal motor unit activity,
making feasible recognition of single unit potentials and their
isolation from the muscle’s interference pattern. Each task
was performed twice for at least 2 min. Once the respiratory
pattern had stabilized after initiation of a task, data were
analyzed over five consecutive breaths.
Acquisition of Breathing Data
The subject wore a nose clip and breathed through a
mouthpiece attached to a low-resistance, two-way breathing
valve (Hans Rudolph no. 2700). Inspiratory flow was recorded
with a pneumotachometer (Fleisch, model 2) attached to the
inspiratory port. The durations of inspiration (TI ), expiration
(TE ), and total cycle duration (Ttot ) were measured from this
inspiratory flow signal. Tidal volume (VT ) and mean inspira-
tory flow (VT/ TI ) were determined on a breath-by-breath
basis from the calibrated recording of the integrated pneumo-
tachometer signal.
Detection of Single Motor Unit Activity
The upper right side of the abdomen was imaged ultrasoni-
cally (Scanner 400/450, Pie Medical) to identify the depths
from the skin’s surface and thicknesses of the underlying
external oblique (EO), IO, and TA muscles. The location of the
thickest part of each muscle, considered the optimal site for
implantation of the fine-wire recording electrodes, was marked
on the skin, and the depth of each muscle at these sites was
recorded in millimeters. Means for the three subjects are
shown in Table 1. The upper abdominal region used for
insertion for the IO and TA electrodes is indicated in the top
of Fig. 1.
Preparation and Insertion of Recording Electrodes
Four fine insulated wires (diameter 0.002 mm, Stablohn
800A-H-poly-nylon green, California Fine Wire) were threaded
through each 26-gauge hypodermic needle. About 2 mm of the
protruding tips of the four wires were folded over the needle’s
bevel, making ‘‘hooks’’ to secure the wires in the muscle, once
implanted. Each of the two recording electrodes was exam-
ined under the microscope to ensure that the only surfaces
devoid of insulation were the cross-sectional areas at the
perpendicular cut at each wire’s tip. The other ends of the
1707
1708 ABDOMINAL SINGLE MOTOR UNIT ACTIVITY
Table 1. Mean depth, muscle thickness, and
implantation site from costal angle for the three
subjects, as measured during ultrasonic imaging
of the surface of the abdominal wall
Distance From Thickness of Distance From
Surface, mm Muscle, mm Costal Angle, mm
Skin 0 6 45
EO 6 4 47
IO 10 6 43
TA 16 4 45
EO, external oblique; IO, internal oblique; TA, transversus abdominis.
wires were flamed to remove the insulation for the connection
to the amplifier. The needle-wire assemblies were dry auto-
claved at 120°C for 1 h. After sterilization, each electrode was
tested to guarantee that the wires were free to move freely
within its needle’s shaft.
The sterile needle was inserted to the appropriate depth
after the predetermined abdominal insertion site was cleansed
with 10% Clinidine. Before the needle was withdrawn, the
muscle was again ultrasonically imaged to determine that the
needle’s tip was in the intended muscle layer. Slight move-
ment of the needle produced sufficient distortions of the
muscle tissue to verify that the tip of the needle was in the
desired location. One pair of wires from each insertion site
was connected to a small-head amplifier (built in house), a
signal conditioner, an audioamplifier, and an oscilloscope
(Gould, model 1604). The subject was asked to contract his
abdominal muscles slightly. If single motor unit potentials
(SMUPs) were not distinguishable on the speaker and oscillo-
scope, a second pair of wires was selected and tested. This
procedure was repeated until ‘‘clean,’’ sharp SMUPs were
Fig. 1. Diagram showing area of electrode inser-
tion (top). Digitized single motor unit poten-
tials (SMUPs; SMUP no. indicated by a sub-
scripted no.) recorded from the same internal
oblique (IO4 ) (left) from subject (S) 1 and trans-
versus abdominis (TA2 ) (right) muscles in S3
over the course of $3 h during indicated experi-
mental conditions of quiet breathing (QB), a
sustained leg lift (LL), and a 7.5-cmH2O expira-
tory threshold load (ETL). Stability of SMUPs
over time, despite changes in posture or respira-
tory load, suggests an absence of electrode
movement. Grd, ground.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (177.099.005.033) on September 11, 2018.
Copyright © 1998 American Physiological Society. All rights reserved.
detected. The pair of wires that detected these clean signals
was used throughout the entire experiment while the other
two wires were just left dangling in space. The needle was
withdrawn, leaving the wires in place within the muscle. To
avoid their inadvertent dislodgement, sterile tape was placed
on the skin over the emergence site of the wires. For the
remainder of the experiment, all recordings from this muscle
were made from this same pair of wires. The same procedure
was followed for implanting the second recording electrode in
the other abdominal muscle. A surface electrode taped over
the right iliac crest served as ground.
Data Acquisition and Analysis
The SMUPs detected from the selected pair of implanted
wires and the pneumotach signal were recorded on FM tape
for off-line digitization and analysis. The analog EMGs and
the flow signal were played back at 2.38 in./s, sampled at 20
kHz, and digitized at 12-bit resolution.
Isolation of SMUPs. The Enhanced Graphics and Acquisi-
tion and Analysis System (EGAA, version 4.1, 1994, devel-
oped by RC Electronics) was applied to the first subject’s
digitized signals from the implanted wires, and Spike 2
software (Cambridge Electronic Design, Chicago, IL) was
applied to the recordings from the other two subjects. Both
programs generated a template of a spike based on its rise
time, peak amplitude, decay time, duration, number of phases,
and spike area. Neither software program automatically
sorted potentials generated by a single motor unit satisfacto-
rily. Therefore, final selection, isolation, and sorting of SMUPs
throughout an experiment was done by visually matching
each spike by overlaying it on the computer-generated tem-
plate. This procedure resulted in the isolation of many
individual spikes of small amplitude, which led to a large
total number of waveforms that could not be grouped with
 ABDOMINAL SINGLE MOTOR UNIT ACTIVITY 1709

others as a motor unit (Table 2). Many of these waveforms Table 3. Mean values obtained during QB,
fired infrequently and randomly, suggesting they were of LL, and ETL
little physiological significance. Analysis was performed only
on those units that fired more than ten times during the five VT, fb ,
consecutive breaths examined during a stable breathing liters breaths/min TI, s TE, s VT/ TI, l/s VT/ TE, l/s
pattern following the initiation of each task. QB 1.0 6 0.1 14.3 6 0.8 1.8 6 0.3 2.4 6 0.4 0.6 6 0.1 0.4 6 0.1
The selected waveforms of the potentials remained stable LL 1.5 6 0.9 12.1 6 1.0 2.2 6 0.4 2.8 6 0.5 0.9 6 0.3 0.6 6 0.4
over hours, as shown by the examples in Fig. 1. Variations in a ETL 1.8 6 0.4 9.6 6 0.8 1.6 6 0.5 4.7 6 0.3 1.1 6 0.4 0.4 6 0.1
waveform’s amplitude up to 28% were considered within an
acceptable limit (26). All SMUPs for a given unit were Values are means 6 SE (n 5 5 breaths) of tidal volume (VT),
assigned a number (e.g., IO1, IO2, etc.) based on that unit’s breathing frequency (fb ), inspiratory duration (TI), expiratory dura-
tion (TE), inspiratory flow (VT/ TI), and expiratory flow (VT/ TE) during
order of appearance in the experiment. QB, LL, and ETL.
Analysis of SMUPs. The time of occurrence of the dis-
charges of each motor unit during each task was examined in
relation to the TI and TE determined from the inspiratory flow
trace. The time at which each SMUP occurred was deter- lengthening of TE during ETL accounted for the slow
mined by placing a cursor at the peak of the SMUP and
breathing frequency. The large VT with no significant
manually recording the displayed time of occurrence. These
values were entered into a spreadsheet for subsequent calcu- change in TI during ETL resulted in a higher VT/ TI
lation of several variables. than that during either QB or LL.

Glossary
Muscle Activity
The following terminology is used throughout the remain-
ing text. During QB, the interference patterns of both IO and
TA displayed intermittent firing of several units with
Spike Compound action potential of a little or no respiratory rhythm (Fig. 2). However, dur-
single motor unit
Interspike interval (ISI) Time between 2 consecutive
ing LL and ETL, both IO and TA showed considerable
spikes fired by the same mo- motor unit recruitment with the greatest enhancement
tor unit, ms occurring in expiration. For ease of presentation, data
Instantaneous frequency ISI, impulses/s from one subject (S1 ) are presented in full detail; data
Doublets Two spikes, the ISI of which is from the other two subjects (S2 and S3 ) were qualita-
,20 ms (13) tively similar and, therefore, are only summarized as
Firing profile In a motor unit, plot of the time appropriate. The reported observations for individual
of occurrence of each spike units (e.g., IO4 and TA7 ) were repeatable in six of six
throughout a breathing cycle trials for S1. Observations in S2 and S3 were similarly
Mean frequency Total number of spikes occur- repeatable, where such data were available, in four of
ring during a breathing cycle,
impulses/s
six and five of six trials, respectively.
Peak frequency Shortest ISI within a respira- Recruitment in S1’s IO. During QB, 14 IO units were
tory cycle, excluding doublets, identified in S1’s records. Thirteen of these units dis-
impulses/s charged only occasionally, with spikes occurring sporadi-
Onset frequency ISI between the 1st and 2nd cally in either phase of respiration. Only one of the
spikes of a rhythmical burst detected units, namely IO4, fired in each of the five
(i.e., ISI at the reinitiation of breaths, and these discharges occurred predominantly
a unit’s firing after an inhibi- in expiration, with a short pause, total silence, or an
tion), impulses/s occasional spike occurring in inspiration.
During the sustained LL, SMUPs of ten IO units
RESULTS
were analyzed as shown in Fig. 3, left. Most units fired
Breathing Patterns quite sporadically at well-spaced intervals but predomi-
nantly during expiration. IO4 (bottom trace), the only
Data in Table 3 show that the breathing pattern unit showing any significant number of spikes, fired
differed among tasks. Breathing was significantly during each expiration and was essentially silenced in
deeper and slower during ETL than during QB or LL. A inspiration.
During ETL, when expiration was prolonged, 12 of
the 34 analyzable IO units discharged at least 10
spikes/breath (Fig. 3). This activity occurred predomi-
Table 2. Total no. of active units identified in IO
nantly during expiration, as shown by their firing
and TA during each task
patterns in Fig. 3, right. IO4 was among the units that
QB LL ETL fired heavily during expiration. In fact, Fig. 4 shows
Subject that the number of spikes per unit time occurring in
No. IO TA IO TA IO TA
expiration (TE ) was significantly greater than during
1 14 1 10 7 12 16 inspiration (TI ) in all three tasks.
2 10 8 Rate coding of IO4. Of the 34 SMUPs identified in IO,
3 4 4 12 9 13 16
only IO4 was recruited for each of the three tasks. The
QB, quiet breathing; LL, leg lift; ETL, expiratory threshold load. firing profiles and instantaneous frequencies of IO4

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (177.099.005.033) on September 11, 2018.

Copyright © 1998 American Physiological Society. All rights reserved.
1710 ABDOMINAL SINGLE MOTOR UNIT ACTIVITY
Fig. 2. Top traces: compressed recording of SMUPs from IO and TA during QB, LL, and ETL. Bottom trace:
inspiratory flow with inspiration down.
during each task are shown in Fig. 5, A and B,
respectively.
During QB and LL, the IO4 spikes occurred in bursts
during late inspiration and throughout expiration, with
marked suppression of firing during early inspiration.
During ETL, IO4 firing again accelerated during expira-
tion, with very short pauses during inspiration, a
discharge pattern markedly different from that during
QB or LL. Figure 5B shows the firing pattern of IO4 as
instantaneous frequencies during the three tasks. Expi-
ratory enhancement occurred whether the task served
a respiratory or nonrespiratory function. Rate coding in
IO4 during each task is shown by the data in Fig. 6.
During the expiratory phase of QB, the firing rate of IO4
increased from a mean onset frequency of 1 impulse/s to
a peak frequency of 12.5 impulses/s, during LL from 3.6
to 9 impulses/s, and during ETL from 3.5 to 14 im-
pulses/s. (Doublets were excluded when peak frequen-
cies were calculated.) Mean firing frequencies of IO4
(horizontal dotted lines in Fig. 5B and bars in Fig. 6)
were not different across tasks.
Figure 7 shows three frequency distributions, one for
each task, with the number of ISIs for IO4 on the y-axis
and the duration of the ISIs and the instantaneous
firing frequencies (impulses/s) on the x-axis. Regardless
of the task, the ranges of these three frequency distribu-
Fig. 3. A: mean tidal volume (VT; in liters) and
durations of inspiration (I) and expiration (E); B:
firing profiles of most active units detected by IO
recording electrode. Arbitary nos. identify motor
units and are displayed in an ascending sequence
in accordance with the time the unit became
active during the task. Left panel: during a
sustained leg lift (LL); right panel: during a
7.5-cmH2O ETL. (Activity during QB is omitted
because so few IO units were active, and these
fired only sporadically in occasional breaths.)
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (177.099.005.033) on September 11, 2018.
Copyright © 1998 American Physiological Society. All rights reserved.
tions are nearly coincident, and their means and modes
are identical and are locked at an interpulse interval
of 150 ms (or an instantaneous frequency of 7.5 im-
pulses/s). In other words, a change in task did not
change the rate coding of IO4. The differences in ISIs
among tasks, of course, reflect the changes in the timing
of events in the respiratory cycle. For example, compared
with the breathing pattern during QB and LL, the
pattern during ETL was deep and slow, with expiration
significantly prolonged (Table 2), accounting for the
greater number of ISIs during ETL (Fig. 6). Despite the
prolongation of TE and Ttot and the increased number of
ISIs during ETL, the mode of the durations of the ISIs
did not change from the 150 ms seen in the other tasks.
In summary, the firing profiles of IO units displayed a
respiratory rhythm. The number of IO units recruited
varied across tasks, with the greatest number of units
recruited units during ETL. Very few IO units (one in
S1 and S2 and two in S3 ) were recruited in all three
tasks, and in those few a change in task did not affect
rate coding.
Recruitment of TA in S1. During quiet breathing, only
one TA (TA7 ) unit was identified in S1’s records. This
unit discharged only occasionally, with spikes occurring
sporadically in either phase of respiration. TA7 also
fired during LL and ETL and was the only TA unit to be
 ABDOMINAL SINGLE MOTOR UNIT ACTIVITY
Fig. 4. Total no. of IO4 spikes (6SE) per inspiratory (TI ) or expiratory
duration (TE ) for 5 consecutive inspirations (open bars) and expira-
tions (solid bars) during each task. * Significant differences (P , 0.05)
between inspiration and expiration.
recruited for all three tasks. Its firing profile and that of
the other 22 TA units recruited during LL and ETL are
shown in Fig. 8. Each of these units displayed a
respiratory rhythm with acceleration of firing during
expiration and a slowing or pause during inspiration.
During LL, TA7 fired one doublet with an ISI of 4.4 ms
during the third expiration. During ETL, 19 doublets
with ISIs ranging from 1 to 18 ms occurred in the firing
profiles of three TA units, namely, TA7, TA48, and TA55.
Rate coding of TA7. Of the 23 SMUPs detected in TA
in S1, only TA7 was recruited for each of the three tasks.
Its firing profiles and instantaneous frequencies during
each task are shown in Fig. 9. The number of spikes
fired per second during TI and TE during each task is
shown in Fig. 10. During ETL, the number of spikes per
second during TE was six times higher than during LL
or QB, reflecting a very strong drive from central
expiratory neurons.
The firing rate of TA7 increased during QB from an
onset frequency of 1.7 impulses/s to a peak frequency of
5.6 impulses/s, during LL from 4.6 to 9.4 impulses/s,
and during ETL from 2.8 to 40.0 impulses/s (Fig. 11).
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (177.099.005.033) on September 11, 2018.
Copyright © 1998 American Physiological Society. All rights reserved.
1711
Fig. 6. Means of onset, mean, and peak firing frequencies of IO4
during QB, LL, and ETL over 5 consecutive breaths.
This steep rise in firing rate from onset frequency to
peak frequency during ETL shows that rate coding
within a breathing cycle was more dramatic than that
during LL or QB.
The three frequency distributions in Fig. 12 show the
number of TA7 spike intervals (y-axis) vs. the durations
of ISIs or instantaneous firing rates (x-axis). The modes
of these three frequency distributions for QB, LL, and
ETL were 240, 160, and 100 ms, respectively, showing
their task dependency. In other words, during ETL, the
TA7 unit was more likely to fire at 15 impulses/s than at
any other frequency, whereas during QB it was likely to
fire much more slowly at 3–5 impulses/s, or during LL
at 5–7.5 impulses/s.
In summary, the firing profiles of all TA units dis-
played a marked respiratory rhythm. The number of TA
units recruited was task dependent, with more units
recruited during ETL than during QB or LL. Very few
Fig. 5. A: IO4 firing profiles and B: instanta-
neous frequencies (Freq) during QB, LL, and
ETL. Vertical dotted lines on x-axis indicate
beginnings and terminations of inspiration
and expiration. Horizontal dotted lines indi-
cate mean firing frequencies for the unit
during each task. imps, Impulses.
1712 ABDOMINAL SINGLE MOTOR UNIT ACTIVITY
Fig. 7. Frequency distributions showing no. of IO4 interspike inter-
vals (ISIs) during each expiration as a function of durations of ISIs in
ms and of instantaneous frequencies (IF) in impulses/s. Modes of
distributions are indicated by vertical line.
TA units (e.g., TA7 ) discharged during all three tasks,
and in those few rate coding was task dependent.
DISCUSSION
The major findings of this study are that 1) different
sets of IO and TA motor units were used for the
different tasks, with overlap between sets in only one or
two motor units in each muscle in each subject; 2) even
though the breathing pattern varied from task to task
(Table 2), motor units in both muscles displayed a
respiratory rhythm in their firing profiles in both
respiratory (QB and ETL) and nonrespiratory (LL)
tasks; and 3) rate coding was similar across tasks in
some units (e.g., IO4 ), whereas rate coding was task-
dependent in others (e.g., TA7 ).
Recruitment Levels Are Task Specific
The number of motor units recruited varied between
muscles during the same task and differed dramati-
Fig. 8. TA data displayed in a way analogous to
that for the IO data of Fig. 3.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (177.099.005.033) on September 11, 2018.
Copyright © 1998 American Physiological Society. All rights reserved.
cally in the same muscle across tasks. During QB,
activity was detected in more IO than TA units (Table
2), a fact recently reported by Abe et al. (1). However,
most of the active units fired only sporadically. There-
fore, it is unlikely that either of these muscles was
making much of a physiological contribution to QB.
During the dynamic act of raising the leg, recruit-
ment is massive, making isolation of SMUPs from the
dense interference pattern impossible and revealing
that a significant fraction of the IO and TA motoneu-
rons are phasic for limb movements. Of course, the
behavior of these units is not revealed in the present
analysis. While the LL was sustained, however, the IO
and TA motor units recruited, with few exceptions (e.g.,
IO4 and TA7 in S1 ), were different from those units
recruited during QB.
During ETL, many motor units in both muscles were
recruited. In fact, the sum of units recruited in IO and
TA during ETL was greater than the number participat-
ing in QB and LL combined. This large-scale recruit-
ment during ETL was expected, since the abdominal
muscles are the major expiratory muscles and ETL is a
reliable stimulus for evoking the abdominal expiratory
reflex (2).
Of the total motor units isolated for analysis in the IO
and TA, only one unit in each of the two muscles in S1
and S2, and two units in S3, participated in all three
tasks. In other words, the recruitment of most units
was task specific. This finding supports the concept
that the abdominal motor pools are organized on the
basis of task groups (14). This fractionation does not
occur in most limb muscles, the motoneurons of which
are recruited in multiple motor tasks. The almost total
absence of overlap in the participation of IO and TA
motor units in the three tasks suggests that at least the
low-threshold IO and TA motoneurons are fractionated
into subpopulations (14). Whether these low-threshold
 ABDOMINAL SINGLE MOTOR UNIT ACTIVITY
motoneurons are differentially responsive to segmental
and descending inputs or whether the inputs are
inhomogeneously distributed to these motoneurons can-
not be discerned from our results.
Interpretation of our findings must be made recogniz-
ing that many factors affect recruitment order. Among
these factors are the intrinsic properties of the motoneu-
rons such as size, threshold, organization, and effi-
ciency of synaptic inputs. In addition, many anatomic
factors influence feedback from sensory structures in
the muscles, including the projections of their segmen-
tal and ascending fibers. Functional factors affecting
recruitment order are conditions such as force level,
changing muscle lengths due to changes in lung vol-
ume, and distribution of the descending signals. None
of these factors were controlled or measured in the
present study but they most likely varied across tasks.
Hence, the sources and mechanisms accounting for the
differences in the IO and TA recruitment levels during
the three tasks remain to be elucidated.
The absence of overlap in the behavior of most of the
IO and TA motor units supports the concept that their
motor pools are functionally organized into subpopula-
Fig. 10. Total no. of TA7 spikes (6SE) per TI or TE for 5 consecutive
inspirations (open bars) and expirations (solid bars) during each task.
* Significant differences (P , 0.05) between inspiration and expira-
tion.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (177.099.005.033) on September 11, 2018.
Copyright © 1998 American Physiological Society. All rights reserved.
1713
Fig. 9. A: firing profiles; B: instantaneous
frequencies during QB, LL, and ETL for TA7.
Dotted lines are as defined in Fig. 5.
tions or task groups. A ‘‘task group’’ comprises all the
motoneurons contributing to a specific muscle contrac-
tion. In some muscles, a task group may completely
overlap with an anatomically defined motor pool (e.g.,
extensor carpi radialis) or form a subpopulation of the
motor pool (e.g., extensor digitorum communis) (22).
Additional studies are required to determine the de-
tailed functional organization of the IO and TA motor
pools.
Influence of Respiration
Anatomically the IO and TA motor pools lie in close
positional relationship within the ventral horns (19, 25,
28), but anatomical proximity does not necessarily
imply functional uniformity. The various motor pools
innervating the abdominal muscles span multiple spi-
nal segments within the ventral horns. Descending
axons from central expiratory neurons provide presyn-
aptic inputs via massive terminal arborizations along
the length of the column of motoneurons (11, 18). On
Fig. 11. Means of onset, mean, and peak frequencies of TA7 during
QB, LL, and ETL over 5 consecutive breaths.
1714 ABDOMINAL SINGLE MOTOR UNIT ACTIVITY
Fig. 12. Frequency distributions for TA7 data derived and plotted in
a way analogous to that for IO4 in Fig. 7. Vertical dotted lines indicate
modes of each distribution.
the basis of this well-distributed common input, it
might be expected that neighboring motoneuron pools
would be concurrently activated to similar levels and
with similar firing patterns. In fact, most early surface
EMGs were interpreted as if the entire abdominal wall
behaved as a functional unit, particularly during respi-
ratory tasks (8). Subsequent reports, however, have
revealed that this is not the case (1, 2). Regionalization
of functions among the abdominal muscles is now a
well-accepted fact, with TA being recognized as a major
contributor to expiration and rectus abdominis (RA) as
a major hip and spine flexor (27).
The firing profiles of IO and TA units tended to
remain in phase with respiration, despite changes in
breathing pattern across both respiratory and nonrespi-
ratory tasks (Table 2). This observation reflects the
common input to both motor pools from central respira-
tory neurons. Any change in the depth and timing of a
breath alters feedback about the instanteous changes
in lung volume and trunk wall configuration. Such
pulmonary and somatic proprioceptive feedback power-
fully modulates the output of central inspiratory and
expiratory neurons (9) and the spinal motoneurons
(12). The resulting compensatory responses serve to
optimize the length of the diaphragm and the pattern of
breathing (15) and to minimize the work and cost of
breathing.
Spinal projections of premotor expiratory neurons in
the caudal ventral respiratory group (VRG-E) are the
major common source of expiratory control of the
abdominal motoneurons (24). In the cat, abdominal
motoneurons receive only sparse monosynaptic excita-
tion from these VRG-E premotoneurons (17, 18), with
spinal interneurons being the recipients and integra-
tors of the descending VRG-E signals destined for the
spinal abdominal motoneurons. Whether monosynap-
tic projections are more common in humans than in
cats is not known. Nonetheless, in this study, most of
the IO and TA units that discharged in each of the five
monitored breaths during each of the three tasks
displayed a rhythmic firing pattern, with its peak
frequency occurring in expiration and a minimum
frequency, or pause, occurring in inspiration. This
tendency for expiratory enhancement of firing and
inspiratory inhibition was prominent during both the
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (177.099.005.033) on September 11, 2018.
Copyright © 1998 American Physiological Society. All rights reserved.
respiratory (QB or ETL) and the nonrespiratory (LL)
tasks. The fact that the differences in breathing pattern
across tasks were reflected in IO and TA firing profiles
supports the concept that human abdominal motoneu-
rons are recipients of breath-by-breath inputs from
central respiratory neurons.
Rate Coding
Both IO4 and TA7 showed major increases in their
instantaneous firing rates from onset frequency at the
beginning of an expiration to the peak frequency reached
during the expiratory burst (Figs. 6 and 10). These
increases in firing frequency were qualitatively similar
for the two motor units. However, IO4 and TA7 differed
in their patterns of rate coding across tasks. The modes
of the frequency distributions for IO4 in Fig. 7 were
locked at 7.5 impulses/s (150 ms ISI), regardless of the
task or breathing pattern. This constancy of rate coding
across tasks in IO4 is a reflection of the strength of the
VRG-E excitatory input to the motoneuron innervating
the IO4 motor unit. In contrast, the modes of analogous
frequency distributions for TA7 (Fig. 12) differed with
each task. During ETL, the mode was at a relatively
high instantaneous firing frequency (20 impulses/s; or
an ISI of 50 ms), during LL at a lower frequency (6.25
impulses/s), and at an even lower frequency (4.25
impulses/s) during QB. Thus, during ETL when expira-
tory flow was impeded, the expiratory drive dominated.
Mean firing frequency, an index of the excitatory
drive to a motoneuron, of IO4 was very similar from
task to task (Figs. 5B and 7), suggesting that the
expiratory drive to IO4 remained similar across tasks.
In contrast, the mean firing frequency of TA7 varied
across tasks. The mean frequency during ETL (10.5
impulses/s) was considerably above that during QB (4.5
impulses/s) or LL (5.8 impulses/s), as shown by the
dashed horizontal lines in Figs. 9B and 11. Hence,
during ETL, the TA7 motoneuron presumably received
a significantly stronger presynaptic drive from the
VRG-E neurons than during QB or LL. This supports
the long-held belief that TA is the major contributor to
the expiratory function of the abdominal wall (5).
In the only other known study on the behavior of
single motor units in abdominal muscles, Sant’Ambrogio
et al. (23) reported on the behavior of units in the RA
and EO while a subject voluntarily increased his ab-
dominal pressure. When generating a 10-mmHg in-
crease in abdominal pressure, the RA and EO motor
units fired at 15 impulses/s, their peak firing frequency.
A small initial increase in discharge frequency from
onset to peak frequency always preceded the recruit-
ment of new units. Because the peak frequency was
only 50% of that during a maximal voluntary contrac-
tion, the authors concluded that recruitment was the
major mechanism used during this voluntary control of
RA and EO. Because of the differences in tasks, it is not
possible to compare our results with those of
Sant’Ambrogio et al. In fact, the present study is the
first report about recruitment and rate coding in IO and
TA motor units during the performance of respiratory
 ABDOMINAL SINGLE MOTOR UNIT ACTIVITY
as well as nonrespiratory tasks. The results suggest
that the pattern of rate coding may be muscle specific.
In summary, in this study, abdominal motor unit
behavior has been quantitatively analyzed during three
natural behaviors in the absence of any artificial con-
straints, other than a necessary restraint on the level of
activity. The results have confirmed that most IO and
TA motoneurons receive strong drives from central
expiratory neurons during both respiratory and nonres-
piratory tasks. The results have revealed that different
groups of IO and TA motor units are recruited for
different tasks, with little overlap among groups. In
those few IO and TA units that participated in each
task, distinct recruitment and discharge patterns were
seen.
Address for reprint requests: F. Cerny, Dept. of Physical Therapy,
Exercise, and Nutrition Sciences, 410 Kimball Tower, SUNY at
Buffalo, Buffalo, NY 14214 (E-mail: cerny@acsu.buffalo.edu).
Received 30 July 1996; accepted in final form 23 December 1997.
REFERENCES
1. Abe, T., N. Kusuhara, N. Yoshimura, T. Tomita, and P. A.
Easton. Differential respiratory activity of four abdominal
muscles in humans. J. Appl. Physiol. 80: 1379–1389, 1996.
2. Barrett, J., F. Cerny, J. A. Hirsch, and B. Bishop. Control of
breathing patterns and abdominal muscles during graded loads
and tilt. J. Appl. Physiol. 76: 2473–2480, 1994.
3. Bishop, B. The abdominal muscles. In: Neural Control of the
Respiratory Muscles, edited by A. D. Miller, A. L. Bianchi, and
B. P. Bishop. Boca Raton, FL: CRC Press, 1996, p. 35–46.
4. Dawson, K., M. D. Hourihan, C. M. Wiles, and J. C. Chawla.
Separation of voluntary and limbic activation of facial and
respiratory muscles in ventral pontine infarction. J. Neurol.
Neurosurg. Psychiatry 57: 1281–1282, 1994.
5. De Troyer, A., M. Estenne, V. Ninane, D. Van Gansbeke, and
M. Gorini. Transversus abdominis muscle function in humans.
J. Appl. Physiol. 68: 1010–1016, 1990.
6. Estenne, M., L. Zocchi, M. Ward, and P. T. Macklem. Chest
wall motion and expiratory muscle use during phonation in
normal humans. J. Appl. Physiol. 68: 2075–2082, 1990.
7. Floyd, W. F., and P. H. S. Silver. Electromyographic study of
patterns of activity of the anterior abdominal wall muscles in
man. J. Anat. 84: 132–145, 1950.
8. Goldman, J. M., R. P. Lehr, A. B. Millar, and J. R. Silver. An
electromyographic study of the abdominal muscles during pos-
tural and respiratory manoeuvres. J. Neurol. Neurosurg. Psychia-
try 50: 866–869, 1987.
9. Hernandez, Y. M., B. G. Lindsey, and R. Shannon. Intercos-
tal and abdominal muscle afferent influence on caudal medullary
expiratory neurons that drive abdominal muscles. Exp. Brain
Res. 78: 219–222, 1989.
10. Hoit, J. D., B. L. Plassman, R. W. Lansing, and T. J. Hixon.
Abdominal muscle activity during speech production. J. Appl.
Physiol. 65: 2656–2664, 1988.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (177.099.005.033) on September 11, 2018.
Copyright © 1998 American Physiological Society. All rights reserved.
1715
11. Kirkwood, P. A., and T. A. Sears. Interaction between the
monosynaptic EPSP and the central respiratory drive potential
of expiratory motoneurones in the cat. J. Physiol. (Lond.) 232:
38P–40P, 1973.
12. Kondo, T., and B. Bishop. Electrically evoked responses of the
human external oblique abdominal muscle. Exp. Neurol. 95:
336–356, 1987.
13. Kudina, L. P., and N. L. Alexeeva. Repetitive doublets of
human motoneurones: analysis of interspike intervals and recruit-
ment pattern. Electroencephalogr. Clin. Neurophysiol. 85: 243–
247, 1992.
14. Loeb, G. E. Hard lessons in motor control from the mammalian
spinal cord. Trends Neurosci. 10: 108–113, 1987.
15. McKenzie, D. K., S. C. Gandevia, R. B. Gorman, and F. C. G.
Southon. Dynamic changes in the zone of apposition and
diaphragm length during maximal respiratory efforts. Thorax
49: 634–638, 1994.
16. Mead, J., K. Yoshino, Y. Kikuchi, G. M. Barnas, and S. H.
Loring. Abdominal pressure transmission in humans during
slow breathing maneuvers. J. Appl. Physiol. 68: 1850–1853,
1990.
17. Merrill, E. G. The descending pathways from the lateral respira-
tory neurones in cats. J. Physiol. (Lond.) 218: 82P–83P, 1971.
18. Miller, A. D., L. K. Tan, and S. F. Lakos. Brainstem projections
to cats’ upper lumbar spinal cord: implications for abdominal
muscle control. Brain Res. 493: 348–356, 1989.
19. Miller, A. D., L. K. Tan, and I. Suzuki. Control of abdominal
and expiratory intercostal muscle activity during vomiting: role
of ventral respiratory group expiratory neurons. J. Neurophysiol.
57: 1854–1866, 1987.
20. Oddsson, L. I. E. Control of voluntary trunk movements in man.
Mechanisms for postural equilibrium during standing. Acta
Physiol. Scand. 140, Suppl. 595: 1–60, 1990.
21. Partridge, M. J., and C. E. Walters. Participation of the
abdominal muscles in various movements of the trunk in man:
an electromyographic study. Phys. Ther. Rev. 39: 791–800, 1959.
22. Riek, S., and P. Bawa. Recruitment of motor units in human
forearm extensors. J. Neurophysiol. 68: 100–108, 1992.
23. Sant’Ambrogio, G., F. Saibene, H. Saiki, and M. Decandia.
Intrabdominal pressure and rate of firing of abdominal muscle
motor units. Arch. Fisiol. 65: 313–322, 1967.
24. Shannon, R. Reflexes from respiratory muscles and costoverte-
bral joints. In: Handbook of Physiology. The Respiratory System.
Control of Breathing. Bethesda, MD: Am. Physiol. Soc., 1986,
sect. 3, vol. II, pt. 1, chapt. 13, p. 431–447.
25. Smith, C. L., and M. Hollyday. The development and postnatal
organization of motor nuclei in the rat thoracic spinal cord. J.
Comp. Neurol. 220: 16–28, 1983.
26. Stalberg, E. V., and M. Soonoo. Assessment of variability in
the shape of the motor unit action potential, the ‘‘jiggle’’ at
consecutive discharges. Muscle Nerve 17: 1135–1144, 1994.
27. Strohl, K. P., J. Mead, R. B. Banzett, S. H. Loring, and P. C.
Kosch. Regional differences in abdominal muscle activity during
various maneuvers in humans. J. Appl. Physiol. 51: 1471–1476,
1981.
28. Tani, M., M. Y. Kida, and K. Akita. Relationship between the
arrangement of motoneuron pools in the ventral horn and rami-
fication pattern of the spinal nerve innervating trunk muscles in
the cat (Felis domestica). Exp. Neurol. 128: 290–300, 1994.