Vestibular contribution to balance control in the medial gastrocnemius and soleus

Abstract

The soleus (Sol) and medial gastrocnemius (mGas) muscles have different patterns of activity during standing balance and may have distinct functional roles. Using surface electromyography we previously observed larger responses to galvanic vestibular stimulation (GVS) in the mGas compared with the Sol muscle. However, it is unclear whether this difference is an artifact that reflects limitations associated with surface electromyography recordings or whether a compensatory balance response to a vestibular error signal activates the mGas to a greater extent than the Sol. In the present study, we compared the effect of GVS on the discharge behavior of 9 Sol and 21 mGas motor units from freely standing subjects. In both Sol and mGas motor units, vestibular stimulation induced biphasic responses in measures of discharge timing [11 ± 5.0 (mGas) and 5.6 ± 3.8 (Sol) counts relative to the sham (mean ± SD)], and frequency [0.86 ± 0.6 Hz (mGas), 0.34 ± 0.2 Hz (Sol) change relative to the sham]. Peak-to-trough response amplitudes were significantly larger in the mGas (62% in the probability-based measure and 160% in the frequency-based measure) compared with the Sol (multiple P < 0.05). Our results provide direct evidence that vestibular signals have a larger influence on the discharge activity of motor units in the mGas compared with the Sol. More tentatively, these results indicate the mGas plays a greater role in vestibular-driven balance corrections during standing balance.

the control of standing balance is a multisensory (vision, vestibular, auditory, and somatosensory) process which requires the coordination of postural muscles throughout the body. This coordination is particularly important at the ankles where the soleus (Sol) and gastrocnemius muscles generate the plantar flexor moment that counteracts the forward toppling of the body to maintain our vertical orientation. Despite a common attachment to the calcaneus, the Sol and gastrocnemius differ in their pattern of activity during standing balance, which may reflect distinct functional roles (Elias et al. 2014; Herman and Bragin 1967; Héroux et al. 2014; Joseph et al. 1955; Loram et al. 2005; Mori 1973; Vieira et al. 2012). The Sol, which is almost continuously active in standing balance and has twice the physiological cross-sectional area of medial gastrocnemius (mGas), provides much of the torque required for standing (Héroux et al. 2014; Joseph et al. 1955; Mori 1973; Ward et al. 2009). In contrast, the mGas is more intermittently active during standing balance and may function to correct for transient disturbances to postural orientation on top of the pedestal of torque provided by the more continuously active Sol (Héroux et al. 2014; Vieira et al. 2012; Ward et al. 2009). Functional differences between muscles are typically attributed to their intrinsic physiology (fiber type composition, cross-sectional area; Edgerton et al. 1975; Fukunaga et al. 1992) and muscle-joint geometry (Hoy et al. 1990); however, these differences are presumably also accompanied by a selective or contextual bias in sensory feedback. Indeed, the Sol has much higher muscle spindle density than the mGas and is thought to receive greater muscle spindle feedback than the gastrocnemius (Banks 2006; Tucker and Turker 2004).

The vestibular system, which contributes to postural control of standing by encoding motion of the head, is easily probed using galvanic vestibular stimulation (GVS) (for review: Fitzpatrick and Day 2004). During GVS, a small electrical current is applied on the skin over the mastoid processes which induces a vestibular error signal, primarily indicating a roll rotation of the head toward the cathode electrode (Day and Fitzpatrick 2005; Fitzpatrick and Day 2004; St. George et al. 2011). This vestibular error signal is then mapped to a compensatory muscle response that helps maintain standing balance in the presence of perceived rotation. This compensatory balance behavior derives from a stereotypical pattern of biphasic electromyographic (EMG) activity and ground reaction shear forces (Britton et al. 1993; Day and Fitzpatrick 2005; Fitzpatrick et al. 1994; Fitzpatrick and Day 2004; Nashner and Wolfson 1974). Previously, we observed that vestibular-evoked muscle responses are larger in mGas than Sol in standing subjects (Dakin et al. 2007; Dakin et al. 2010a), suggesting that during standing balance mGas muscle activity is influenced more by vestibular input than is Sol muscle activity. However, there is a limitation with this interpretation: the differences in response size were obtained from surface EMG recordings (Dakin et al. 2007; Dakin et al. 2010a). Several factors other than motor unit recruitment and firing frequency (e.g., electrode placement, muscle volume beneath the electrodes, signal propagation through the tissue and motor unit entrainment) can influence the amplitude of the surface EMG response to a sensory stimulus, and one or more of these factors could explain our previous results. To overcome these limitations, we compared the effect of GVS on the discharge activity of human Sol and mGas motor units (MUs) in freely standing subjects. By recording the activity of single MUs, the influence of an isolated vestibular error signal on MU firing behavior can be directly explored. We hypothesized that the vestibular stimulus would modify measures of mGas MU discharge probability and firing frequency to a larger degree than those of the Sol.

METHODS

Subjects.

Six healthy male subjects [mass 80.5 ± 5 kg and height 1.77 ± 0.08 m (mean ± SD)] between the ages of 23 and 48 yr with no known history of neurological disease or injury participated in this study. The experimental protocol was explained, and written informed consent was obtained. All procedures used in this study conformed to the standards of the Declaration of Helsinki and were approved by the University of British Columbia's Clinical Research Ethics Board.

Single motor unit recordings.

Single MU activity was recorded with custom-made fine-wire intramuscular electrodes. The electrodes were made with two 0.05-mm-diameter insulated stainless steel wires (California Fine Wire, Grover Beach, CA) twisted together. The tips of the wires were cut to expose the cross section of uninsulated wire and were then folded back to create two ∼1-mm barbs to anchor the wire in the muscle. The wound wires were threaded into a 1.5-in. 25-gauge hypodermic needle (EXEL International Medical Products, St Petersburg, FL) and steam sterilized (PVdry2 Barnstead|Harvey, Dubuque, IA). Wire electrodes were inserted into the mGas (6 bipolar intramuscular electrodes) and Sol (4 bipolar intramuscular electrodes) muscles (c.f. figure 1 in Héroux et al. 2014) of the right leg under ultrasound guidance (SonoSite MicroMaxx, Bothell, WA). The mGas electrode locations were distributed evenly over the bottom half of the muscle belly while the Sol electrodes were located on both the medial and lateral margins of the Sol, below the distal edges of the more superficial medial and lateral heads of the gastrocnemius. Recent evidence suggests a large percentage of mGas MUs span the length (32 of 69 recorded MUs) or full width (24 of 36 recorded MUs) of the muscle belly (Héroux et al. 2015). Therefore, we believe our sample to be broadly reflective of the muscle as a whole. Signals were amplified (×5,000) and filtered (30-6,000 Hz) with a high-impedance amplifier (GRASS 15CT, Astromed, West Warwick, RI) and digitized at 15,152 Hz with a 16-bit CED DAQ board and Spike2 software (Cambridge Electronics Design, Cambridge, UK). This specific sampling rate was a product of the CED DAQ clock management and is a function of the number channels being recorded and the requested sampling rate.

Fig. 1.

Raw data for medial gastrocnemius (mGas) and soleus (Sol) motor units (MUs) during quiet standing. A: a single MU recorded from the mGas. On the top left is the result of the template match for the prominent MU in the trace. Below is the instantaneous discharge rate for the same MU (mGas 8.5 ± 2.6 imp/s; Sol 6.9 ± 1.5 imp/s). Note the higher variability in the instantaneous frequency relative to the Sol. B: template matching and raw trace for a Sol MU. Note the lower variability in the instantaneous frequency. Imp/s, impulses per second. Differences in the discharge behavior between motor units in these two muscles are discussed further in Héroux et al. 2014.

Vestibular stimulation.

Vestibular stimuli consisted of a series of monophasic ± 4 mA square-wave pulses lasting 250 ms with variable interstimulus intervals (1.000–1.750 s). Sham stimuli were also included to serve as a control (no stimulus, but trigger signal generated for data analysis). The stimulus waveform was created in a custom LabVIEW program (National Instruments, Austin, TX) and delivered by a DAQ board (PXI 6289; National Instruments) to an isolated constant-current stimulator (model 2200 Analog Stimulus Isolator; A-M Systems, Carlsborg, WA). The binaural bipolar vestibular stimuli were delivered through carbon rubber electrodes (9 cm²) coated in conductive gel (Spectra 360, Parker Laboratories, Fairfield, NJ) secured to the mastoid processes. Both bipolar electrode configurations were used to control for potential response asymmetry: anode right/cathode left (anode right) and anode left/cathode right (anode left).

Testing protocol.

Participants stood freely, their head turned over their left shoulder with Reid's plane pitched up ∼18° relative to horizontal and their eyes fixated on a point on the wall (Fitzpatrick and Day 2004). The trunk was also slightly rotated to the left to align the interaural axis with the sagittal plane causing balance responses evoked by the GVS to be in the anteroposterior direction (Fitzpatrick and Day 2004), thus orienting the direction of the perturbation with the line of action of the Sol and mGas. Participants were also encouraged to lean slightly forward to ensure ongoing plantar flexor muscle activity. Each trial took between 5 and 9 min and consisted of 250 randomly presented anode right, 250 anode left stimuli, and 250 random sham stimuli. Eight trials were completed for a total of 2,000 anode right stimuli, 2,000 anode left stimuli, and 2,000 sham stimuli. A rest period of at least 2 min was provided between trials.

Due to the requirement that participants were actively engaged in standing balance, it was not possible for us to control for the discharge rate of the recorded lower limb MUs (Héroux et al. 2014; Mochizuki et al. 2006; Vieira et al. 2012). Doing so would have led to unacceptably long testing sessions. Also, to ensure sufficient spikes were collected to observe the effect of the vestibular stimulus on single MU activity, each MU needed to be identified reliably over several trials (each 5–9 min). These constraints limited the number of suitable MUs to only those that could be tracked for at least four trials.

Data and statistical analysis.

MU action potentials were analyzed with Spike2 software (Cambridge Electronics Design, Cambridge, UK). A template matching algorithm identified MU action potentials based on their size, shape, and timing (Fig. 1). Identified MUs were manually reviewed to include unmatched potentials and instances where two MU action potentials were superimposed. Once a MU was identified and sorted, its discharge times were exported to Matlab for further analysis (Mathworks, Natick, MA). Baseline firing rates were compared between MUs of the two muscles using an independent samples t-test assuming unequal variance.

MU discharge times were analyzed with a probability-based measure, the peristimulus time histogram (PSTH), to characterize the vestibular-induced MU response in a manner analogous to stimulus-triggered averaged surface EMG. This analysis addressed whether vestibular-evoked responses recorded from single MUs reflect the biphasic responses reported with surface EMG recordings. To create the PSTH, the 250-ms time period on either side of stimulus onset was divided into 2-ms bins and the total number of times a MU discharged within each bin was tallied across all 2,000 stimuli. A Matlab script based on the local regression and likelihood method, described by Mitra and Bokil (2008; http://chronux.org/), was used to provide a smoothed PSTH estimate from which the amplitude and latency of the primary peak and trough of the vestibular-induced response could be identified (see Fig. 2C). The minimum or maximum of the local regression was taken as the amplitude of the evoked response. In general, responses to electrical vestibular stimulation are conventionally described in terms of a short-latency EMG response (SL: 50–70 ms) and a medium-latency EMG response (ML: 100–120 ms) (Britton et al. 1993, Dakin et al. 2007, 2010b; Fitzpatrick and Day 2004). However, to simplify comparison, we used the peak-to-trough amplitude of the SL and ML responses, respectively, as our measure of response size. To determine if a response was present in a given muscle, the peak-to-trough amplitude of the SL and ML response was compared with the peak-to-trough amplitude of the sham trial, at times corresponding to the average SL and ML responses across the two electrode configurations, using a repeated-measures ANOVA (with electrode configurations and sham as the levels). Post hoc comparisons were carried out using Tukey's HSD test. To compare responses between the mGas and Sol, each PSTH was converted to a probability mass function by dividing the bin count by the total number of spikes in the peristimulus window (−500 to 1,000 ms). This transformation equates bin height to the probability of spike firing at each 2-ms time interval within the peristimulus window. Then the peak-to-trough amplitude between the SL and ML responses was determined using local regression. Since there was no statistical difference between the two electrode configurations in the repeated-measures ANOVA (see results and Table 1), peak-to-trough values were averaged across electrode configurations, and Sol and mGas response amplitude were compared using an unpaired t-test.

Fig. 2.

Vestibular-evoked responses. An example of each of the three stimuli (A), and the accompanying raw spike train (B) from a single mGas motor unit. C: peristimulus time histogram (PSTH) of a single motor unit's response to the vestibular stimulus in each electrode configuration and the sham trial. This motor unit had a mean firing rate of 7.8 ± 2.3 imp/s. Superimposed in black is the curve fitted with the local regression and likelihood method.

Table 1.

Results of the repeated-measures ANOVA for the PSTH and PSF compared with sham (main effects decomposed using Tukey's post hoc)

				Main Effect		Post hoc
	AR	AL	Sham	F	P	AR-AL	AR-Sham	AL-Sham
PSTH
mGas	22 ± 8.6	25 ± 6.9	2 ± 2.0	75.9	P < 0.001	P = 0.35	P < 0.001	P < 0.001
Sol	11 ± 5.4	12 ± 5.8	1 ± 0.9	28.7	P < 0.001	P = 0.92	P < 0.001	P < 0.001
PSF
mGas	1.6 ± 0.5	1.8 ± 0.6	0.3 ± 0.2	59.1	P < 0.001	P = 0.28	P < 0.001	P < 0.002
Sol	0.6 ± 0.2	0.7 ± 0.2	0.2 ± 0.1	17.0	P < 0.01	P = 0.71	P < 0.01	P < 0.01

Each row displays the compared means (±SD), the ANOVA comparison (F ratio and P value), as well as the P values from Tukey's HSD test for each muscle.

AR, anode right; AL, anode left; mGas, medial gastrocnemius; Sol, soleus; PSTH, peristimulus time histogram; PSF, peristimulus frequencygram.

Stimulus-induced responses were statistically different from sham in both the PSTH and PSF; however, there was no difference between electrode configurations.

A potential limitation of using probabilistic measures such as the PSTH is that peaks and troughs can be dominated by temporary phase locking of MUs to the stimulus onset masking changes in motor unit firing frequency (Moore et al. 1970; Turker and Powers 2005). In some instances, complementary measures based on MU discharge rate are better suited to characterize changes in MU discharge behavior resulting from a stimulus (Turker and Powers 2005). Thus we calculated the peristimulus frequencygram (PSF), which reflects the change in instantaneous discharge rate relative to the onset of the stimulus. The PSF provides a more accurate representation of the time profile of the net synaptic current to the motor neuron and will provide an alternative indication of the relative influence of the vestibular stimulus on MUs of the Sol and mGas (Bessou et al. 1968; Powers and Turker 2010; Turker and Powers 2005). To calculate the PSF, instantaneous discharge rates were calculated as follows:

$$1/(t_n-t_{n-1})$$

where t is the discharge time relative to stimulus onset. Instantaneous rates were plotted at discharge time t_n relative to stimulus onset and averaged across all 2,000 stimuli. Peaks and troughs in the PSF were estimated from the curve derived from the local regression and likelihood method. To determine the presence of vestibular-evoked responses in the PSF, the peak-to-trough amplitude of the SL and ML response for both electrode configurations were compared with the peak-to-trough amplitude of the sham trial, at times corresponding to the SL and ML responses averaged across the two electrode configurations, using a repeated-measures ANOVA. Post hoc comparisons were carried out using Tukey's HSD test. As there was no statistical difference between the two electrode configurations PSF peak-to-trough amplitudes were averaged across electrode configurations. Comparison between the two muscles was then performed using an unpaired t-test. All statistical analyses were performed in Statistica (Dell Statistica, Statsoft, Tulsa, OK) with the level of significance set to P < 0.05. Results are presented as means and SD.

RESULTS

General observations.

A total of 21 MUs were analyzed, 12 from the mGas [21,708 ± 8,331 spikes (mean ± SD)] and 9 from the Sol (16,851 ± 9,526 spikes). Across MUs, mean discharge rate was greater for mGas MUs compared with those from the Sol [8.4 ± 1.6 vs. 6.9 ± 1.3 Hz (t₁₉ = 2.49, P = 0.022)] (Fig. 1).

Biphasic motor unit responses to a vestibular stimulus.

Galvanic vestibular stimulation evoked biphasic responses in the PSTH of mGas and Sol MUs (Fig. 3) and, as expected, the sham stimulus had no effect on MU discharge behavior. The vestibular stimulus evoked responses in the PSTH of all 12 mGas MUs and 8 of 9 Sol MUs. On average, the MU count at the peak or trough changed relative to the sham by 11 ± 5 counts in the mGas and 6 ± 4 counts per 2-ms bin in the Sol representing a 50 ± 22 (mGas) and 41 ± 39 (Sol) percent change relative to the sham. Anode right stimulation caused an initial increase in bin count followed by a decrease in bin count below baseline levels. The overall pattern was similar but reversed for the anode left condition. The SL response peaked at 78 ± 8 ms (mGas) and 75 ± 8 ms (Sol) while the ML response peaked at 138 ± 25 ms (mGas) and 128 ± 24 ms (Sol).

Fig. 3.

Local regression and likelihood functions for the mGas and Sol PSTH. In gray are the mean-removed fitted curves for the PSTH of each motor unit. In black are the means of the fitted curves for each condition. The top row are responses from the 12 mGas motor units to the anode right, anode left, and sham conditions. Below are the corresponding curves for the 9 Sol motor units. Time zero indicates stimulus onset. The horizontal dotted line indicates the prestimulus mean. The calibration bars indicate the size of the change relative to baseline.

The peak-to-trough amplitude of the mGas MU PSTH responses evoked by GVS were significantly larger than the sham condition (Anode Right, 22 ± 8 vs. 2 ± 2 spikes; Anode Left, 25 ± 7 vs. 2 ± 2 spikes, F_2,22 = 75.9, P < 0.001; see Table 1), but were not statistically different from each other (P = 0.35). Similarly, in the Sol muscle, the PSTH responses evoked by GVS were also significantly greater than in the sham condition for both electrode configurations (Anode Right, 11 ± 5 vs. 1 ± 1 spikes; Anode left, 12 ± 6 vs. 1 ± 1 spikes, F_2,16 = 28.7, P < 0.001; see Table 1), and not statistically different from each other (P = 0.92).

Responses in the PSF were generally less prominent than in the PSTH and thus were present in only a subset of all the collected MUs (10 of 12 mGas MUs and 4 of 9 Sol MUs). An additional MU from both the mGas and Sol exhibited small responses to only one stimulus polarity. Similar to the responses observed in the PSTH, GVS induced a biphasic change in discharge frequency in both muscles, with the SL response peaking at 119 ± 21 ms (mGas) and 95 ± 18 ms (Sol) and the ML response peaking at 206 ± 20 ms (mGas) and 167 ± 32 ms (Sol). The peak-to-trough amplitude of the PSF response evoked by GVS was significantly greater than the sham for both muscles [mGas: Anode Right, 1.6 ± 0.5 vs. 0.3 ± 0.2 imp/s; Anode left, 1.8 ± 0.6 vs. 0.3 ± 0.2 imp/s (F_2,18 = 59.1, P < 0.001; see Table 1); Sol: Anode Right, 0.6 ± 0.2 vs. 0.2 ± 0.1 imp/s; Anode left, 0.7 ± 0.2 vs. 0.2 ± 0.1 imps/s (F_2,6 = 17.0, P < 0.003; see Table 1)]. In general, the polarity of the PSF response was comparable to those in the PSTH. Anode Right currents resulted in an initial increase in MU discharge rate relative to the sham trial followed by a decrease in discharge rate relative to the sham (Fig. 4A) whereas Anode Left currents evoked inverted responses relative to the Anode Right condition (Figs. 4A and 5). Neither muscle exhibited a statistically significant effect of electrode configuration (mGas: P = 0.28; Sol: P = 0.71).

Fig. 4.

Average mGas and Sol peristimulus frequencygram (PSF) for both electrode configurations. A: the left column displays the mean PSF for both the mGas and Sol. For illustrative purposes, only motor units from the mGas (n = 8) and Sol (n = 4) that exhibited the most prominent responses were used. mGas responses are on the top and Sol responses on the bottom. B: instantaneous frequency for a single subject showing spike clustering following the stimulus. In the top plot (Anode Right) there is spike clustering at ∼100 ms and reduction in spikes just before 200 ms. The behavior is reversed in the Anode Left condition (bottom). C: change in interspike interval due to the electrical vestibular stimulus. The oblique black line indicates the series of bins at each 1-ms interval (Spike) over which a spike has occurred. Plotted to the right of each bin (spike + 1) is a point indicating the average spike time for subsequent motor unit discharge following the occurrence of a spike in the left adjacent bin (spike). The illustration indicates that for Anode Right currents, the short-latency (SL) response results in a shortening of interspike intervals, whereas the medium-latency (ML) response results from a lengthening of interspike intervals. The opposite was true for Anode Left currents. For all subplots, gray dots illustrate responses to Anode Left electrical vestibular stimuli, and black dots responses to Anode Right stimuli.

Fig. 5.

Comparison between the PSTH and the PSF for the Anode Left/Cathode Right condition in the mGas. Compared with the PSTH, which provides a probabilistic measure of changes in motor unit discharge due to the stimulus, the PSF indicates the change in motor unit discharge rate due to the stimulus. The black arrows indicate the timing of the SL and ML peaks of the PSTH and the timing of the first and second peaks of the PSF. Time zero is stimulus onset.

Larger vestibular-evoked motor unit responses in mGas than Sol.

The peak-to-trough amplitude of the SL and ML responses were significantly larger in mGas MUs compared with the Sol (Fig. 6). Normalized PSTH peak-to-trough amplitudes, averaged across electrode configurations, were 63% (1.5 ± 0.4 vs. 0.9 ± 0.4 × 10⁻³) larger in the mGas than in Sol (t₁₉ = 3.56, P = 0.002). Similarly, PSF mGas responses were 153% (1.7 ± 0.4 vs. 0.7 ± 0.2 imp/s) larger than PSF Sol responses (t₁₄ = 5.5, P < 0.001).

Fig. 6.

Comparison of response amplitudes between the mGas and Sol. A: the peak-to-trough difference in SL-ML response amplitude observed in the PSTH as a change in the probability of spike occurrence. B: comparison of the stimulus-induced peak-to-trough change in discharge frequency between the Sol and mGas. Responses to Anode Right and Left vestibular stimuli have been averaged to provide a single mean value for each muscle. In light gray are individual subject responses with the mean and standard deviations in black. *Significant difference between the mGas and Sol muscles (see results).

DISCUSSION

The purpose of this study was to determine how a vestibular stimulus influences the discharge behavior of MUs in two lower leg muscles that have distinct patterns of activity in standing balance. Based on prior observation that the mGas muscle exhibits larger responses to vestibular stimuli than the Sol during standing balance, we hypothesized that a vestibular error signal would exert a larger influence over the discharge activity of mGas MUs compared with those in the Sol. This hypothesis was confirmed: MU responses to a vestibular stimulus were significantly larger in mGas than Sol.

The effect of a vestibular stimulus on motor unit discharge behavior.

Electrical vestibular stimulation, delivered in a bipolar binaural configuration to standing subjects, evoked distinct biphasic responses in the PSTH of triceps surae MUs. The polarity and timing of these responses mirrored those previously observed with surface EMG recordings under similar experimental conditions (SL 50–70 ms; ML 100–120 ms; Britton et al. 1993; Fitzpatrick and Day 2004; Nashner and Wolfson 1974). These biphasic responses have been described as the muscular counterpart of the biphasic force response, as recorded with a force plate, induced by electrical vestibular stimulation (Fitzpatrick et al. 1994). In the present study 1,000–2,000 electrical vestibular stimuli were required to observe a response in single MU recordings (in both the PSTH and PSF). The overall effect of these stimuli was an average change of 11 ± 5 counts per 2-ms bin in the mGas and 6 ± 4 counts per 2-ms bin in the Sol, relative to the sham. The vestibular-induced changes in MU discharge behavior were small in amplitude and occurred over multiple 2-ms bins following the vestibular stimuli (see Figs. 2 and 3). These small and prolonged (i.e., not limited to one or two 2-ms bins) vestibular-evoked responses contrast with the MU responses elicited by other forms of neural stimulation (e.g., magnetic or electrical transcranial stimulation, peripheral nerve stimulation, or tendon percussion; Burke et al. 1984; Day et al. 1989; Priori et al. 1993; Thompson et al. 1991). Our analyses revealed that the vestibular stimulus induced mostly a small, but consistent, shortening or lengthening of the MUs interspike intervals (see Fig. 4). We propose that the vestibular-evoked responses observed in lower limb MUs represent organized responses to a signal of head movement. The electrical vestibular stimulus induces a vestibular error signal that is interpreted by the central nervous system as head rotation which is compensated for by the balance system in a task-dependent manner. This contrasts with a simple reflex pathway which would be expected to elicit a MU response at a shorter latency, with narrow peaks, and larger amplitudes. Because of the need for task-dependent compensation, vestibular stimuli may require greater neural integration and processing than other forms of nerve stimulation to produce the diffuse response in axial and appendicular muscles necessary to maintain posture (Forbes et al. 2015). Indeed, there is a longer than expected delay between vestibular stimuli and their evoked responses in the appendicular muscles (Britton et al. 1993). Assuming that vestibular signals travel via fast conducting vestibulo- or reticulospinal pathways, and that these pathways have conduction velocities comparable to corticospinal pathways (which is the case in cats; see Peterson et al. 1975; Takahashi 1965; Wilson et al. 1967), the latency of vestibular induced responses should be similar to those obtained with magnetic simulation of the motor cortex. In appendicular muscles, however, vestibular-induced responses are delayed by ∼30 ms compared with responses evoked by magnetic stimuli applied over the motor cortex (Britton et al. 1993). These delays likely reflect the central and peripheral processing required to transform the vestibular error signals (Britton et al. 1993; Yakusheva et al. 2013) and generate an appropriate compensatory balance response.

Larger vestibular-evoked responses in the medial gastrocnemius than the soleus.

The vestibular stimulus generated larger responses in mGas MUs compared with those of the Sol muscle for both probability-based and frequency-based measures. These results confirm our earlier findings using surface EMG recordings (Dakin et al. 2007; Dakin et al. 2010b) and provide strong evidence that descending inputs of vestibular origin have a greater influence on mGas motor neurons compared with Sol motor neurons during standing balance. We suggest the greater influence of vestibular inputs on mGas MUs compared with those of the Sol is due to the different roles these muscles play in active balance control.

It is well established that vestibular stimulation elicits responses only in muscles actively involved in balance control (Britton et al. 1993; Fitzpatrick et al. 1994; Luu et al. 2012). For example, vestibular-evoked responses are present in leg muscles active in standing balance but responses are absent when comparable levels of activity are generated in the same muscles while standing with the trunk supported (Fitzpatrick et al. 1994). The amplitude of vestibular-induced muscle responses also appear to scale with a muscle's involvement in maintaining standing balance (Britton et al. 1993; Fitzpatrick et al. 1994; Forbes et al. 2015; Luu et al. 2012). This was most recently demonstrated in standing balance by Mian and Day (2014) who observed that the compensatory response to vestibular stimulation can be biased to the sagittal plane by changing mediolateral stance width. Standing with a wide stance increases stability in the frontal plane with little effect on stability in the sagittal plane and results in a reduction in the size of the frontal plane component of the vestibular-evoked response components with little effect on the sagittal plane component. A similar functional dependency can also be observed in walking where the influence of a vestibular error signal on lower limb muscle activity waxes and wanes contingent on the muscle from which the activity is being recorded and the phase of the gait cycle (Blouin et al. 2011; Dakin et al. 2013). We believe a balance-related gradient of vestibular influence on lower limb motor neurons may also explain the difference we observed between mGas and Sol MU response size to a vestibular stimulus. Our findings suggest that, compared with the Sol, the mGas plays a greater role in the stabilization of balance to transient disturbances of vertical posture.

Both the biochemistry and anatomy of these two muscles support this proposition. Biochemically, the Sol possesses a higher proportion of oxidative muscle fibers suited for prolonged activity and control of low-frequency sway, whereas the gastrocnemius possesses a higher proportion of glycolytic muscle fibers, which are better suited for intermittent compensation (Edgarton et al. 1975; Fukunaga et al. 1992). Anatomically, the gastrocnemius has longer muscle fibers (Friederich and Brand 1990), which are related to higher shortening velocity (Wickiewicz et al. 1983). Thus the mGas is better suited to generate the rapid contractions necessary to correct unexpected vestibular-evoked balance perturbations. Correspondingly, the vestibular system detects transient as well as continuously varying stimuli but it may not be sufficiently sensitive (threshold of 0.5–3°/s) to encode low-frequency, low-amplitude sway (Benson et al. 1989; Collins and De Luca 1993; Grabherr et al. 2008; Peterka and Benolken 1995; Peters et al. 2015; Sadeghi et al. 2007; Valko et al. 2012). The greater effect of vestibular error signals on mGas MU activity indicates that mGas is preferentially activated, and may be better suited, to counter unpredicted accelerations of the head, and by extension, of the whole body. Our results therefore support the idea that mGas (see Fig. 1) is biased toward generating the corrective balance torque used to maintain the vertical position of the body in the event of transient disturbances to posture (Elias et al. 2014; Héroux et al. 2014; Woollacott et al. 1984).

Limitations of this study.

Both the PSTH and PSF are sensitive to baseline firing rate, consequently providing poor estimates of synaptic behavior when firing rates are too low. Under ideal conditions MU firing would be controlled and vestibular stimuli provided, or included in analysis, only when MU firing rate was in a predetermined range. Unfortunately, controlling the mean firing rate for a sufficient number of trials was unfeasible given the number of stimuli required to elicit a response in single MU, the dependency of vestibular responses on free sway (Lee Son et al. 2008), and the demands of the experiment on participants. Last, the proportion of input from the various presynaptic sources driving the mGas and Sol may have influenced the observed differences in vestibular-evoked responses between these muscles. Unfortunately, matching the relative strength of presynaptic inputs driving the mGas and Sol muscles during standing balance is currently beyond control.

Conclusions.

Here, we examined the relative influence of vestibular stimulation on the discharge behavior of MUs in two lower leg muscles that are believed to play distinct functional roles in the maintenance of a stable upright posture. Single MU responses to a vestibular stimulus were significantly larger in mGas compared with Sol, indicating that mGas plays a greater role in vestibular-driven balance corrections during standing balance.

GRANTS

This work was supported by the Natural Sciences and Engineering Research Council (NSERC) of Canada (J.-S. Blouin and J. T. Inglis). C. J. Dakin was funded by both NSERC and Canadian Institutes of Health Research (CIHR). M. E. Héroux was funded by the CIHR and the Michael Smith Foundation for Health Research (MSFHR). J.-S. Blouin received salary support from the MSFHR and Canadian Chiropractic Research Foundation. B. L. Luu and M. E. Héroux received salary support from a NSERC Discovery Accelerator Supplements (J. T. Inglis). The ultrasound equipment was purchased with a NSERC Research Tool and Instrumentation grant (J. T. Inglis, J.-S. Blouin, and M. G. Carpenter).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: C.J.D., M.E.H., B.L.L., J.T.I., and J.-S.B. conception and design of research; C.J.D., M.E.H., B.L.L., J.T.I., and J.-S.B. performed experiments; C.J.D., M.E.H., and B.L.L. analyzed data; C.J.D., M.E.H., B.L.L., and J.-S.B. interpreted results of experiments; C.J.D. and M.E.H. prepared figures; C.J.D., M.E.H., and J.-S.B. drafted manuscript; C.J.D., M.E.H., B.L.L., J.T.I., and J.-S.B. edited and revised manuscript; C.J.D., M.E.H., B.L.L., J.T.I., and J.-S.B. approved final version of manuscript.