1 To be submitted to: Journal of Neurophysiology 1
Reciprocal inhibition between Motor Neurons of the
2Tibialis Anterior and Triceps Surae in humans
3Utku S. Yavuz1, Francesco Negro2, Robin Diedrichs3, Dario Farina4
4
1Department of Anesthesiology, Pain Medicine, University Medical Center Göttingen,
Georg-5
August University, Göttingen, Germany 6
2Department of Clinical and Experimental Sciences, University of Brescia, Brescia, Italy
7
3Clinic for Trauma Surgery, Orthopedics and Plastic Surgery, University Medical Center
8
Göttingen, Göttingen, Germany 9
4Department of Bioengineering, Imperial College London, London, United Kingdom
10 11 12
Words in abstract: 248 13
Words in text: 3154 excluding figure cations 14 Figures: 6 15 Tables: 0 16 17 18
Address for correspondence: 19
Dario Farina 20
Department of Bioengineering 21
Imperial College London 22 SW7 2AZ London, UK 23 E-mail: d.farina@imperial@ac.uk 24 25 26 27
2 Abstract
1
Motor neurons innervating antagonist muscles receive reciprocal inhibitory afferent inputs in 2
order to facilitate the joint movement in the two directions. In this study, we investigate the 3
influence of the distribution of reciprocal Ia input [Is this really the main hypothesis ? We do not 4
measure the distribution of spindles …]on the strength of the mutual reciprocal inhibition 5
between the tibialis anterior (TA) and triceps surae (soleus and medial gastrocnemius) motor 6
units. We asses this mutual mechanism in large populations of motor units for building statistical 7
distributions and during standardized input to the motor neuron pools in order to minimize the 8
effect of modulatory pathways. Single motor unit activities were identified using high-density 9
surface electromyography (HDsEMG) recorded from the tibialis anterior (TA), soleus (Sol) and 10
medial gastrocnemius (GM) muscles during isometric dorsi- and plantar-flexion. Reciprocal 11
inhibition was elicited by electrical stimulation of the tibial (TN) or common peroneal nerve 12
(CPN). The probability density distributions of reflex strength for each muscle were estimated. 13
The strength of reciprocal inhibition in the medial gastrocnemius and the soleus motor units 14
showed a 4-fold narrower range with respect to the tibialis anterior motor units. This result 15
suggests an asymmetric transmission of reciprocal inhibition between ankle extensor and flexor 16
muscles. The most likely mechanism regulating the strength of reciprocal inhibition between 17
analyzed lower limbs is a differential synaptic distribution of reciprocal spindle afferents (Ia). 18
[Abstract needs further improvements.] 19
New & Noteworthy 20
We demonstrated that the bisynaptic reciprocal inhibition between triceps surae and tibialis 21
anterior muscles is asymmetrical. We investigated this mutual transmission of reciprocal 22
3 inhibition in large samples of motor units and with standardized input to the motor neurons. The 1
functional meaning of asymmetric transmission may be associated to the neural control strategies 2
of posture. 3
Keywords: reciprocal inhibition, single motor unit, high density EMG, synaptic distribution 4
Introduction 5
The proprioceptive muscle spindle system is a fundamental component of the sensory–motor 6
organization in human motor control. When a muscle fiber stretches, muscle spindle afferents (Ia) 7
cause monosynaptic activation of the alpha motor neurons of the homonymous muscle. In a 8
simplified model, the excitatory activity requires a corresponding relaxation of the antagonist 9
muscle for functional accord (Sherrington, 1913). It has been generally accepted that alpha motor 10
neurons are reciprocally activated via Ia afferents in voluntary movement (Knikou, 2008). The 11
reciprocal inhibitory pathway has been described in detail through intracellular recordings in 12
animal (Eccles et al., 1956; Eccles and Lundberg, 1958) and intramuscular recordings in human 13
studies (Kudina, 1980). It inhibits antagonist motor neurons through at least one interneuron 14
(Crone et al., 1987; Katz et al., 1991). The functional role of this mutual inhibitory pathway in 15
motor control, especially during bipedal walk (Lavoie et al., 1997; Petersen et al., 1999) and 16
posture (Kasai et al., 1998), has also been extensively investigated. However, these results 17
focused on modulatory pathway and not on the mutual distribution of reciprocal inhibitory input. 18
The distribution of synaptic inputs and the intrinsic properties of alpha motor neurons determine 19
the excitability of the motor neuron pool (Burke, 1968, 1970; Buller et al., 1980; Miles and 20
Türker, 1986; Semmler and Türker, 1994; Heckman et al., 2009). Therefore, we hypothesized 21
that different intrinsic characteristics of the motor neuron pools and synaptic input distributions 22
4 can induce an asymmetric reciprocal inhibition between ankle extensor and flexor muscles. This 1
asymmetry was observed in few early studies using global surface EMG and single motor unit 2
analysis in humans (Crone et al., 1987; Nielsen and Kagamiharat, 1992), and cats (Eccles and 3
Lundberg, 1958; Nichols, 1989). However, in those studies, it was not possible to establish 4
whether the asymmetric reciprocal inhibition was due to the different mutual distribution of 5
reciprocal inhibitory input, mainly because of methodological limitations. The only way to 6
directly investigate the effect of reciprocal inhibition on the output of the motoneuron pool is to 7
measure the influence of standardized input on the timings of discharge of individual motor units 8
and sampling a large number of motor units for building statistical distributions (Nielsen and 9
Kagamiharat, 1992; Yavuz et al., 2015). In this study, we applied this approach to examine the 10
distribution of Ia reciprocal inhibition between motor units of the tibialis anterior (TA) and 11
Triceps surae (soleus and medial gastrocnemius) muscles. 12
Methods 13
Seven healthy subjects (males, age: 35 ± 4 years) took part to the experiment. The experimental 14
protocol was approved by the Human Ethics Committee of the University Medical Center, 15
Georg-August-University of Göttingen (approval number: 1/10/12), and was in accordance with 16
the Declaration of Helsinki. Each subject provided informed written consent prior to the 17
experiments. 18
Experimental protocol 19
Subjects were seated on the chair of a Biodex System 3 (Biodex Medical Systems Inc., NY, 20
USA) with the right leg and foot firmly fastened. Three 64-channel high density surface EMG 21
(HDsEMG) grids (5×13 electrode grid, 8-mm interelectrode distance, ELSCH064NM2, OT 22
5 Bioelettronica, Torino, Italy) were placed on the belly of the TA, soleus (Sol) and medial
1
gastrocnemius (GM) muscles. The ankle joint was fixed in the anatomical resting position and 2
the knee joint angle was at 120°. For the recordings on the GM muscle, the leg was extended to 3
170° of the knee angle. An ankle strap electrode was used as reference (WS2, OT Bioelettronica, 4
Torino, Italy). The tibial (TN) and the common peroneal nerve (CPN) were stimulated in 5
different sessions in order to elicit reciprocal inhibition on TA and triceps sure (GM and Sol) 6
muscles, respectively (Figure 1). All sessions were randomized. The metal pin anode and cathode 7
were placed on the posterior and anterior parts of the fibula’s head for stimulating the CPN and 8
on the popliteal fossa for stimulating the TN. The cavities of metal pins were filled with 9
conductive gel. 10
Figure 1 11
The HDsEMG, the force and stimulation trigger signals were recorded at 10240 sample/s with 12
resolution of 12 bit, using a multichannel EMG data acquisition system (EMG-USB2, OT 13
Bioelettronica, Torino, Italy). The EMG data were band-pass filtered (20-500 Hz) and amplified 14
by a factor of 2000 in all sessions. The stimulation currents were delivered using a constant 15
current stimulator (Digitimer DS-5) that was controlled by a custom made Matlab (The 16
MathWorks, Inc.) script program. 17
The subjects were asked to perform three isometric maximum voluntary ankle dorsi- and 18
plantarflexion contractions, with 2 min of rest between each contraction. The maximum peak of 19
the three consecutive maximal contractions was chosen as maximum voluntary contraction MVC 20
force for each corresponding direction. The subject then performed sustained dorsi- or 21
plantarflexion contractions at 10% and 20% MVC in random order. During each contraction, a 22
minimum of 150 electrical stimuli were delivered to the TN or CPN nerves. To identify the 23
6 stimulation intensity, the Hoffman reflex (H-reflex) and direct motor response (M-wave) were 1
recorded on the homonymous muscle before each trial. The intensity of the stimulation (300 – 2
500 µs pulse duration) was set as the intensity that elicited an H-reflex with size of 10% of the 3
maximum M-wave (Figure 2A) and delivered to the nerve with 1-2 s random inter pulse 4
intervals. 5
Analysis 6
Single motor units were identified from the HDsEMG signals using a blind source separation 7
technique (Holobar and Zazula, 2007; Negro et al., 2016) that provides fully automatic 8
decomposition. Only units with a silhouette value (SIL) greater than 0.9 were used (Negro et al., 9
2016). The accuracy of HDsEMG decomposition in conditions similar to the current study has 10
been previously validated (Yavuz et al., 2015). The EMG signals were digitally band-pass 11
filtered with 20-500 Hz cutoff frequencies, down-sampled at 2048 Hz, and divided in intervals of 12
60 s with 10 s overlap and decomposed individually. The spike trains were then merged for the 13
same motor units across the decomposed intervals, as previously described (Yavuz et al., 2015). 14
The reciprocal inhibitory responses of the single motor units were measured using the peri-15
stimulus frequencygram (PSF) that shows the instantaneous discharge rates of a single motor unit 16
against the time instant of the stimulus (Awiszus et al., 1991; Türker and Cheng, 1994). The 17
cumulative sum (CUSUM) of the PSF was calculated in a 400-ms window (200 ms before and 18
after stimulation) in order to magnify the small reductions in motor unit discharges during the 19
reflex responses. The strength of reciprocal inhibition was quantified as the difference between 20
the minimum and the onset of the PSF-CUSUM v-alue of the reflex response and normalized to 21
the number of stimulations that were delivered to the nerve. Such normalization enables the 22
estimation of the extra discharge reduction per stimulation during the reciprocal inhibitory reflex 23
7 responses. The mean discharge rate of each motor unit was calculated in the pre-stimulus time 1
interval of 500ms. The onset of the inhibition was confirmed using PSTH-CUSUM. The 2
significance of inhibition strength was determined as the maximum PSF-CUSUM deviation from 3
the pre-stimulus time window. Only those troughs that exceeded the significance threshold were 4
accepted as genuine inhibitory responses (Brinkworth and Türker, 2003). 5
Figure 2 6
Statistical Analysis 7
The test H-reflex amplitudes that were normalized to the maximum M-wave were compared 8
across subjects using t-tests since the data were normally distributed. These tests verified whether 9
the amount of reflex input was similar across subjects. The motor units with low decomposition 10
accuracy (SIL < 0.9) and coefficient of variation of the inter spike interval greater that 35% were 11
excluded from the statistical analyses. The differences between reciprocal inhibition strengths 12
and between recruitment thresholds across motor unit populations were tested using the Mann-13
Whitney U test which compares the median and the interquartile ranges, since the data were not 14
normally distributed. The probability density distribution of single motor unit responses to 15
stimuli was estimated using the data from all trials (10% and 20% of MVC at dorsi- and plantar-16
flexion). The Freedman-Diaconis’ rule, which is based on the interquartile range, was used to 17
estimate the bin size of the probability density histogram. For each muscle and contraction force, 18
the best fitting distribution function was selected among normal, gamma and exponential, 19
according to the highest goodness of fit value. The goodness of fit value denotes the discrepancy 20
between the distribution of the empirical data and the theoretical distribution model (Breeze et 21
al., 1988). The median, interquartile (IQR), and the absolute range of the selected distribution 22
were compared between muscles. The association between the motor unit recruitment threshold 23
8 and the reciprocal inhibition amplitude was examined using the bivariate Spearman correlation 1
analysis. T-test was used to investigate whether the portion of responsive motor units was 2
different between muscles since the data were normally distributed. In all tests, the significance 3
level of p<0.05 was chosen, since no repetitive test was performed in this study. 4
Results 5
The normalized test H-reflex amplitudes computed from bipolar surface EMG and used for 6
determining the stimulation intensity, were not significantly different between subjects at the 7
same force level (p > 0.05). The test reflex amplitude was greater at 20% MVC than at 10% 8
MVC (p < 0.05). The mean stimulation intensities were 10.3±1.9 mA and 8.8±2.9 mA for 9
eliciting the test H-reflex in the TA and Sol muscles, respectively. Trials where the direct motor 10
response (M-wave) was observed were excluded from the analysis. 11
In total, 788 motor units were identified from the three muscles of the seven subjects at the two 12
contraction forces. The reciprocal inhibition reflex amplitude was significant for 376 of the 13
identified motor units. These motor units will be referred to as “responsive motor units” in the 14
following. The average number of responsive motor units across subjects was 83±7% and 15
74±14% of the total for TA, 37±9% and 38±18% of the total for Sol, and 30±15% and 23±9% of 16
the total for the medial GM, at 10% and 20% MVC respectively. The t-test statistics confirmed 17
that the ratio of the responsive motor units was significantly greater in the TA than Sol and GM 18
at both contraction levels (p < 0.05). 19
Figure 3 depicts the empirical probability density histogram and estimated fitting curve for 20
individual muscles at the two contraction levels. The gamma distribution function was the best 21
fitting model to the empirical data (goodness of fit values: 0.75 and 0.82 for TA, 0.83 and 0.88 22
9 for Sol, 0.77 and 0.80 for GM, at 10% and 20% MVC, respectively). The absolute range of 1
probability density histograms was approximately four times greater for TA (0.13 - 2.33 2
Hz/N.Stim and 0.09 - 2.75 Hz/N.Stim, at 10% and 20% MVC, respectively) than Sol (0.03 - 0.59 3
Hz/N.Stim, 0.03 - 0.28 Hz/N.Stim) and GM (0.06 - 0.48 Hz/N.Stim, 0.4 - 0.38 Hz/N.Stim) 4
(Figure 3). Figure 4 shows the cumulative probability curve for the three muscles. The Mann-5
Whitney U test quantitatively confirmed that the median and the range of reciprocal inhibition 6
amplitude distribution was greater for TA than the other muscles at both contraction levels 7
(p<0.05, Figure 4). No significant difference was found with regard to contraction force level 8 (p>0.05). 9 Figure 3 10 Figure 4 11
The motor unit recruitment thresholds were estimated from the low-pass filtered rectified EMG, 12
after normalization. The average recruitment threshold was similar across the muscles at 10% 13
MVC (6.3±1.6%MVC for GM, 7.6±1.1%MVC for Sol and 7.1±2.1%MVC for TA, p>0.05), but 14
TA motor units had greater thresholds than in the other muscles for 20% MVC (13.5±4.4%MVC 15
for GM, 14.9±1.1%MVC for Sol and 18.8±7.2%MVC for). The distribution of reflex amplitude 16
was calculated as a function of the recruitment threshold. The bivariate Pearson correlation test 17
revealed that the reflex amplitude was negatively correlated with recruitment threshold for TA 18
motor units only for the 20%MVC (Pearson correlation value ρ = -0.27, significance p = 0.03). 19
The correlation strength for TA at 10%MVC, and for GM and Sol at 10 and 20%MVC, were not 20
significant (ρ: 0.14 for GM, 0.09 for Sol and -0.08 for TA at 10%MVC; 0.06 for GM, -0.11 for 21
Sol at 20%MVC; p>0.05 for all correlations) (Figure 5). 22
10 To separate the effect of discharge rate, we further compared the reciprocal inhibitory response 1
amplitudes for motor units with similar baseline discharge rates (in the range 5-11 Hz at 10% 2
MVC and 5-13Hz at 20%MVC) across the three muscles (n=100 for TA, n=68 for Sol and n=37 3
for GM at 10%MVC; n=66 for TA, n=48 for Sol and n=22 for GM at 20%MVC) (Figure 6). The 4
differences between these subgroups were investigated using the Mann-Whitney U test, since the 5
normality test failed. The statistical test revealed that TA motor units showed a significantly 6
greater response (0.45±0.24 Hz/N.Stim, 0.49±0.31 Hz/N.Stim at 10 and 20%MVC) than GM 7
(0.16±0.25 Hz/N.Stim, 0.14±0.08 Hz/N.Stim at 10 and 20%MVC) and Sol (0.12±0.09 8
Hz/N.Stim, 0.12±0.05 Hz/N.Stim at 10 and 20%MVC) (p < 0.05). 9 Figure 5 10 Figure 6 11 Discussion 12
For large populations of motor units, we showed that the reciprocal inhibitory reflex of the triceps 13
surae muscles on the TA muscle is greater than that from the TA to the triceps surae. The 14
reciprocal interaction between ankle extensor and flexor muscles is a crucial sensory drive in 15
humans for adjusting the ankle joint torque during gait as well as maintaining upright posture. It 16
has been shown that the strength of reciprocal inhibition is modulated in different phases of 17
walking in the cat (Pratt and Jordan, 1987)and human (Capaday et al., 1990; Lavoie et al., 1997; 18
Petersen et al., 1999). Moreover, the predominant transmission of reciprocal inhibition changes 19
direction between plantar flexor and dorsiflexor during stance and swing phases of the walking 20
cycle (Petersen et al., 1999). These studies revealed the differential reciprocal inhibition during 21
functional tasks where the modulatory effects, including supraspinal and peripheral pathways, 22
11 cannot be excluded (Nielsen, 2016). However, it is still not clear how the actual distribution of 1
the reciprocal Ia input effects the strength of the inhibition between antagonist muscles. Previous 2
studies showed that synaptic currents generated by most of the sensory and descending inputs are 3
non-uniformly distributed among motor units (Heckman and Binder, 1988; Semmler and Türker, 4
1994; Powers and Binder, 2001). This non-uniform synaptic distribution may exist for an 5
effective control of movement. In this study, we investigated the mutual reciprocal Ia input 6
between the antagonist ankle extensor and flexor muscles. 7
The main result of the study is that the Ia reciprocal inhibition between triceps surae and tibialis 8
anterior is not symmetrically distributed. The medial gastrocnemius and the soleus motor units 9
have less pronounced reciprocal inhibition response compared to the tibialis anterior motor units. 10
Quantitatively we found that the distribution of TA motor unit inhibition amplitudes had 11
significantly larger median value and more than 4-fold larger range with respect to the Sol and 12
GM motor units. This result is in agreement with an earlier surface EMG study where an 13
asymmetry between reciprocal inhibition of ankle flexors and extensors was reported (Crone et 14
al., 1987). However, the most accurate method for estimating the distribution of the synaptic 15
input is measuring the reflex response amplitude of single motor units which shows its net post 16
synaptic potential (Türker and Powers, 2005). To the best of our knowledge, the only motor unit 17
study mutually comparing the reciprocal inhibition of lower limb antagonist muscles is by 18
Nielsen and Kagamihara (1992). These authors reported that the transmission in the bisynaptic 19
reciprocal inhibition is depressed during co-contraction for both antagonistic motor unit groups 20
(Nielsen and Kagamiharat, 1992). Nonetheless, several other mechanisms can change the 21
transmission gain of reciprocal inhibition during co-contraction tasks, due to the modulatory 22
presynaptic effects from descending and sensory pathways (see review from Nielsen, 2016). In 23
the present study, we used a standard input provided by electrical stimulation. The intensity of 24
12 the stimulation was small as to selectively excite the Ia axons of the antagonist muscle.
1
Moreover, we tested isometric contractions to minimize the co-activation of the antagonist 2
muscles. Thus, we limited the contribution of modulatory pathways to predict the actual 3
contribution of reciprocal Ia input. Finally, we were able to identify large samples of motor units 4
(a total of 788 motor units) based on a novel methodology (Yavuz et al., 2015) that provides a 5
stronger statistical basis for estimating probability densities. 6
We examined whether there was any association between the recruitment threshold and the reflex 7
response of individual motor units. Because of the relation between recruitment threshold and 8
motor unit size (Henneman, 1957; Heckman and Enoka, 2012), we hypothesized that the 9
different motor unit composition of TA, Sol and GM muscles might be a potential factor 10
contributing to the disparity between Ia reciprocal inhibition strength of these muscles. Assuming 11
a uniform Ia reciprocal input distribution on the motor neuron pool of a muscle, early recruited 12
small motor units should present greater reflex responses (Henneman et al., 1964; Burke, 1970; 13
Garnett and Stephens, 1980). Sol and GM muscles have a more homogenous distribution of 14
motor unit sizes than TA and Sol specifically contains predominantly slow twitch motor units 15
(Gollnick et al., 1974; Edgerton et al., 1975; Garnett et al., 1979). TA showed a significant 16
negative correlation between reflex amplitude and recruitment threshold while Sol and GM did 17
not (Figure 5). Despite these differences in motor unit composition, it is unlikely that the motor 18
units sampled at 10% and 20% MVC from the three muscles were substantially different. For all 19
muscles, these contraction levels indeed correspond exclusively to slow twitch motor units. 20
Accordingly, the results of asymmetric inhibition were fully confirmed also when we limited the 21
analysis only to the motor units with a similar background discharge rate for the three muscles. 22
The most likely explanation for the results is that of a differential synaptic transmission of 23
reciprocal inhibition between analyzed lower limb muscles. 24
13 An asymmetric strength in inhibitory pathways between TA and the triceps surae is presumably 1
associated to the functional interplay between these muscles. During quite standing the triceps 2
surae muscles counteract the micro-falls forward of the human body and stabilize posture (Loram 3
et al., 2005). Whereas the Sol is mainly tonically active, the GM shows bursts of activations 4
corresponding to each micro-fall (Vieira et al., 2012). During these activations, a strong 5
inhibitory pathway to the TA may be needed in order to maximize the effect of the bursting 6
activity, as in ballistic tasks. Indeed, motor unit recruitment in the GM during these bursts of 7
activations is extremely fast. Conversely, the TA has very limited activation during quite 8
standing since the forward sway is determined by gravity and by the position of the center of 9
pressure ahead of the feet. 10
In conclusion, we demonstrated that the strength of reciprocal inhibition between TA and triceps 11
surae muscle is asymmetrical, with a substantially stronger inhibition of the TA than of the 12
triceps surae. This mechanism has been observed in large samples of motor units and in 13
standardized synaptic input to the motor neurons. This disparity in the mutual inhibition may be 14
associated to the specific neural control strategies of posture. 15
Acknowledgement 16
We would like to express our special thanks to Kemal S. Türker who provided us helpful advises 17
within the present study. 18
Grants 19
Francesco Negro has received funding from the European Union’s Horizon 2020 research and 20
innovation programme under the Marie Skłodowska-Curie grant agreement No 702491 21
(NeuralCon). 22
14 1
References 2
Awiszus F, Feistner H, Schäfer SS. On a method to detect long-latency excitations and
3
inhibitions of single hand muscle motoneurons in man. Exp Brain Res 86: 440–446, 1991. 4
Binboĝa E, Türker KS. Compound group I excitatory input is differentially distributed to
5
human soleus motoneurons. Clin Neurophysiol 123: 2192–2199, 2012. 6
Breeze P, D’Agostino RB, Stephens MA. Goodness-of-Fit Techniques. Appl Stat 37: 115, 1988.
7
Brinkworth RSA, Türker KS. A method for quantifying reflex responses from intra-muscular
8
and surface electromyogram. J Neurosci Methods 122: 179–193, 2003. 9
Buller N, Garnett R, Stephens J. The reflex responses of single motor units in human hand
10
muscles following muscle afferent stimulation. J Physiol 303: 337–349, 1980. 11
Burke R. Group Ia Synaptic input to fast and slow twitch motor units of cat triceps surae. J
12
Physiol 196: 605–630, 1968. 13
Burke R. A comparison of peripheral and rubrospinal synaptic input to slow and fast twitch
14
motor units of triceps surae. J Physiol 207: 709–732, 1970. 15
Capaday C, Cody FW, Stein RB. Reciprocal inhibition of soleus motor output in humans
16
during walking and voluntary tonic activity. J Neurophysiol 64: 607–616, 1990. 17
Crone C, Hultborn H, Jespersen B, Nielsen J. Reciprocal Ia inhibition between ankle flexors
18
and extensors in man. J Physiol 389: 163–85, 1987. 19
Eccles J, Fatt P, Landgren S. The central pathway for direct inhibitory action of impulses in
20
largest afferent nerve fibres to muscle. J Neurophysiol 19: 75–98, 1956. 21
Eccles RM, Lundberg A. Integrative pattern of Ia synaptic actions on motoneurones of hip and
22
knee muscles. J Physiol 144: 271–298, 1958. 23
15
Edgerton VR, Smith JL, Simpson DR. Muscle fibre type populations of human leg muscles.
1
Histochem J 7: 259–266, 1975. 2
Garnett RA, O’Donovan MJ, Stephens JA, Taylor A. Motor unit organization of human
3
medial gastrocnemius. JPhysiol 287:33-43: 33–43, 1979. 4
Garnett R, Stephens JA. The reflex responses of single motor units in human first dorsal
5
interosseous muscle following cutaneous afferent stimulation. J Physiol 303: 351–364, 6
1980. 7
Gollnick PD, Sjödin B, Karlsson J, Jansson E, Saltin B. Human soleus muscle: A comparison
8
of fiber composition and enzyme activities with other leg muscles. Pflügers Arch Eur J 9
Physiol 348: 247–255, 1974. 10
Heckman CJ, Binder MD. Analysis of effective synaptic currents generated by homonymous Ia
11
afferent fibers in motoneurons of the cat. J Neurophysiol 60: 1946–66, 1988. 12
Heckman CJ, Enoka RM. Motor Unit. In: Comprehensive Physiology. John Wiley & Sons, Inc.,
13
2012 14
Heckman CJ, Mottram C, Quinlan K, Theiss R, Schuster J. Motoneuron excitability: The
15
importance of neuromodulatory inputs. Clin Neurophysiol 120: 2040–2054, 2009. 16
Henneman E. Relation between size of neurons and their susceptibility to discharge. Science
17
(80- ) 126: 1345–1347, 1957. 18
Henneman E, Somjen G, Carpenter DO. Excitability and Inhibitibility of Motoneurons of
19
Different Sizes. J. Neurophysiol 28(3): 599-620, 1965 20
Holobar A, Zazula D. Multichannel Blind Source Separation Using Convolution Kernel
21
Compensation. IEEE Trans Signal Process 55, 2007. 22
Kasai T, Kawanishi M, Yahagi S. Posture-dependent modulation of reciprocal inhibition upon
23
initiation of ankle dorsiflexion in man. Brain Res 792: 159–163, 1998. 24
16
Katz R, Penicaud A, Rossi A. Reciprocal Ia inhibition between elbow flexors and extensors in
1
the human. J Physiol 437: 269, 1991. 2
Knikou M. The H-reflex as a probe: Pathways and pitfalls. J Neurosci Methods 171: 1–12, 2008.
3
Kudina L. Reflex effects of muscle afferents on antagonist studied on single firing motor units in
4
man. Electroencephalogr Clin Neurophysiol 50: 214–221, 1980. 5
Lavoie BA, Devanne H, Capaday C. Differential control of reciprocal inhibition during walking
6
versus postural and voluntary motor tasks in humans. J Neurophysiol 78: 429–438, 1997. 7
Loram ID, Maganaris CN, Lakie M. Human postural sway results from frequent, ballistic bias
8
impulses by soleus and gastrocnemius. J Physiol 564: 295–311, 2005. 9
Miles TS, Türker KS. Does reflex inhibition of motor units follow the “size principle”? Exp
10
Brain Res 62: 443–445, 1986. 11
Miles TS, Türker KS. Decomposition of the human electromyogramme in an inhibitory reflex.
12
Exp Brain Res 65: 337–342, 1987. 13
Negro F, Muceli S, Castronovo AM, Holobar A, Farina D. Multi-channel intramuscular and
14
surface EMG decomposition by convolutive blind source separation. J Neural Eng 13: 1– 15
17, 2016. 16
Nichols TR. The organization of heterogenic reflexes among muscles crossing the ankle joint in
17
the decerebrate cat. J Physiol 410: 463–477, 1989. 18
Nielsen BYJ, Kagamiharat Y. The regulation of disynaptic reciprocal ia inhibition during
co-19
contraction. J Physiol Paris : 373–391, 1992. 20
Nielsen J, Kagamihara Y. The regulation of presynaptic inhibition during co-contraction of
21
antagonistic muscles in man. J Physiol 464: 575–593, 1993. 22
Nielsen JB. Human Spinal Motor Control. Annu Rev Neurosci 39: 81–101, 2016.
23
Petersen N, Morita H, Nielsen J. Modulation of reciprocal inhibition between ankle extensors
17 and flexors during walking in man. J Physiol 520: 605–619, 1999.
1
Powers RK, Binder MD. Input-output functions of mammalian motoneurons. In: Reviews of
2
physiology, biochemistry and pharmacology. Springer, 2001, p. 137–263. 3
Pratt C a, Jordan LM. Ia inhibitory interneurons and Renshaw cells as contributors to the spinal
4
mechanisms of fictive locomotion. J Neurophysiol 57: 56–71, 1987. 5
Semmler JG, Türker KS. Compound group I excitatory input is differentially distributed to
6
motoneurons of human tibialis anterior. Neurosci Lett 178: 206–210, 1994. 7
Sherrington CS. Reflex inhibition as a factor in the coordination of the movements and postures.
8
J Nerv Ment Dis 6: 251–310, 1913. 9
Türker KS, Cheng HB. Motor-unit firing frequency can be used for the estimation of synaptic
10
potentials in human motoneurones. J Neurosci Methods 53: 225–234, 1994. 11
Türker KS, Powers RK. Black box revisited: A technique for estimating postsynaptic potentials
12
in neurons. Trends Neurosci 28: 379–386, 2005. 13
Vieira TMM, Loram ID, Muceli S, Merletti R, Farina D. Recruitment of motor units in the
14
medial gastrocnemius muscle during human quiet standing: is recruitment intermittent? 15
What triggers recruitment? J Neurophysiol 107: 666–676, 2012. 16
Yavuz UŞ, Negro F, Sebik O, Holobar A, Frömmel C, Türker KS, Farina D. Estimating
17
reflex responses in large populations of motor units by decomposition of the high-density 18
surface electromyogram. J Physiol 593: 4305–4318, 2015. 19
20
Figure Captions: 21
22
Figure 1: Experimental setup. Three high density surface EMG electrodes (HDsEMG) were 23
placed on tibialis anterior (TA), soleus (Sol) and medial gastrocnemius (GM). The metal pin 24
18 anode (red) and cathode (black) were placed on the posterior and anterior parts of the fibula’s 1
head for stimulating the common peroneal nerve (CPN) and on the popliteal fossa for stimulating 2
the tibial nerve (TN). Dynamometer was used to elicit ankle torque during plantar and dorsi-3
flexion individually and displayed on the feedback screen. 4
Figure 2: Sequence of signal processing from data acquisition to offline estimation of single 5
motor unit reflex response amplitude. A) The test H-reflex on homonymous muscle was 6
monitored during the experiment in order to tune the stimulation intensity as to elicit an H-reflex 7
with size equal to 10% of the maximum M-wave. B) The high density surface EMG (HDsEMG) 8
was recorded from the antagonistic muscle. C) Single motor units were decomposed from the 9
HDsEMG signals. Each row of vertical lines refers to the discharge spike trains of a single motor 10
unit. D) From bottom to top, the graphs depict peri-stimulus frequencygram (PSF), cumulative 11
summation of PSF (PSFcusum), peri-stimulus time histogram (PSTH), and PSTHcusum, 12
respectively. The horizontal red lines show the significance thresholds. The vertical dashed lines 13
indicate the stimulation time (time zero) and the onset of the inhibition. The arrow shows the 14
distance between minimum and the onset values where the inhibition amplitude was measured. 15
Figure 3: The probability density distribution of reflex amplitudes for tibialis anterior (TA), 16
medial gastrocnemius (GM) and soleus (Sol) at each contraction level (10% and 20% of MVC). 17
Histograms show the probability density of empirical data. Solid red curves depict the estimated 18
best fitting functions. 19
Figure 4: Cumulative probability (upper row) and significance between distributions through the 20
median and range values (lower row). The dashed lines in the cumulative probability graphs 21
(upper row) and the vertical red lines in the box plots show the median reflex amplitude values. 22
Asterisk (*) indicates the significant differences between groups (P<0.05). 23
19 Figure 5: Ia reciprocal inhibition amplitude as a function of recruitment threshold. The dashed 1
lines depict the linear regressions. 2
Figure 6: The reflex amplitude of the motor unit populations of GM (n=37 for 10%MVC and 3
n=22 for 20%MVC), Sol (n=68 for 10%MVC and n=48 for 20%MVC) and TA (n=100 for 4
10%MVC and n=66 for 20%MVC), with similar baseline discharge rate (5-11 Hz for 10% MVC 5
and 5-13Hz for 20%MVC). The asterisk indicates that TA had significantly greater reflex 6
amplitude (P<0.05). The red horizontal lines indicate the median reflex amplitudes. The cross (+) 7
indicates the average reflex amplitudes. 8
9 10 11
20 Figures:
21 Figure 2
22 Figure 3
23 Figure 5
24 Figure 6
