Ten men (age 26.3 ± 2.8 years, height 175.1 ± 5.0 cm, and weight 67.0 ± 7.5 kg) participated in the present study. The procedure, purpose, risks, and benefits associated with the present study were explained to the subjects and written, informed consent was obtained from all of them. The ethics review committee on experimental research with human subjects of the Graduate School of Arts and Sciences at The University of Tokyo approved the experimental protocols, which were conducted in accordance with the guidelines of the Declaration of Helsinki.
Isometric contraction tasks
Experiments were conducted while subjects were in the supine position on the bench of a dynamometer (CON-TREX, CMV AG, Dübendorf, Switzerland). This was because preferential recruitment of the sensory fibers was induced by tSCS in the supine position compared to the prone and standing positions (Danner et al. 2016). In the supine position, the hip joint was extended to the anatomical position, and the right knee joint was flexed to 30°. The right ankle joint was plantar-flexed to 10° to relax the dorsiflexor muscles and fixed to the attachment of the dynamometer with non-elastic straps.
Following several warm-up trials, the subjects performed an isometric maximal voluntary contraction (MVC) for approximately 3 s. Isometric voluntary contraction types were plantar-flexion, dorsi-flexion, knee extension, and knee flexion. Two MVC trials were performed to obtain the maximal torque. The inter-trial interval was set to 1 min. If the peak torques between the two trials were different by more than 10%, an additional trial was performed. The trial with the highest peak torque among two or three MVCs was used for the analysis. Following a 2-min rest interval, voluntary contraction at 10% of the MVC level was performed. During isometric contraction tasks, the subjects tried to exert the joint torques by contracting the agonist muscles only (e.g., major agonist action of triceps surae muscles, plantar-flexion). The order of plantar-flexion, dorsi-flexion, knee extension, and knee flexion at 10% of the MVC level was randomized. The subjects were provided visual feedback of 10-Hz low pass-filtered torque signals and the target torque level via a computer monitor using specific software (LabChart 7, ADInstruments, Melbourne, Australia).
Surface electromyographic recording
Surface electromyographic (EMG) signals were recorded from SOL, TA, vastus lateralis (VL), and the long head of the biceps femoris (BF) in the right leg. Ag–AgCl electrodes (Vitrode F-150S, Nihon Kohden, Tokyo, Japan) with an inter-electrode distance of 20 mm were used for EMG acquisition from each muscle. The amplifier was set to a gain of 1000-fold with a bandpass filter between 5 Hz and 1 kHz (AB-611 J, Nihon Kohden). The EMG signals and torque signals were simultaneously sampled at 4 kHz using an AD converter (PowerLab, ADInstruments, Melbourne, Australia) and stored on a personal computer using software (LabChart 7, ADInstruments).
A double-cone coil (outside diameter of 110 mm) was placed over the leg area of the left motor cortex to obtain MEPs from the right SOL, TA, VL, and BF by TMS (Magstim 200 stimulator, Magstim, Dyfed, UK). At the beginning of the measurements, the optimal stimulating site (i.e., “hot spot”) providing the largest amplitude for the SOL evoked response was identified. The head of each subject was secured on a head rest. The TMS coil position was marked on the non-elastic cap to ensure that the same area of the cortex was stimulated throughout the experiment. Next, the resting motor threshold (RMT) was determined while subjects rested quietly. The RMT was defined as the lowest stimulation intensity for which peak-to-peak amplitudes of MEP were larger than 50 μV for at least three of five stimuli. The TMS intensity was set to 130% of the RMT of SOL (Abdelmoula et al. 2016). The stimulation intensity in the present study was 76.4 ± 14.7% of maximal stimulator output. MEPs were evoked during the resting and weak voluntary contraction conditions, and ten stimuli with an inter-pulse interval of 6 s were delivered in each condition.
PRM reflexes in the lower-limb were evoked using a constant current electrical stimulator with a single rectangular pulse of 1-ms duration (DS7A, Digitimer, Hertfordshire, UK). An anode (100 × 75 mm) was placed over the midline of the abdomen between the xiphoid process of the sternum and the umbilicus, and a cathode (50 × 50 mm) was placed on the midline of the back between spinous processes of upper-lumbar vertebrae. Since the stimulation electrode position of tSCS at different spinal levels affects recruitment of the evoked responses (Danner et al. 2011; Roy et al. 2012), in the present study, the cathodes were positioned where larger responses were evoked from all recorded muscles at any stimulation intensity, based on the visual determination of the response magnitude (Masugi et al. 2016; Nakagawa et al. 2018; Saito et al. 2019). The cathode was placed between L1 and L2 in all subjects. To confirm that the responses were initiated in the sensory fiber based on suppression of the second response owing to post-activation depression (Andrews et al. 2015), a double-pulse stimulation with a 50-ms inter-pulse interval and with various stimulation intensities were delivered while the subject was resting (Courtine et al. 2007; Minassian et al. 2007) (Fig. 1). Test stimulation intensity of tSCS was visually determined based on the procedure below. Stimulation intensity was increased from 10 to 20 mA below test stimulation intensity until the appearance of a second response in some muscles by 2-mA stimulation intensity increments. Then, the strongest stimulation intensity that could produce a large first response without a second response for SOL, TA, VL, and BF was chosen. The mean stimulation intensity for tSCS was 61.4 ± 12.2 mA. Under the resting and voluntary contraction conditions, ten stimuli with an inter-pulse interval of 6 s were delivered.
The root-mean-square (RMS) values of EMG signals of agonist muscles during MVC were calculated over 1000-ms during the same period that the peak torque was obtained. This was used for normalization of the EMG signals during agonist muscle contraction at the 10% of MVC level. As the background EMG, the RMS in a 500-ms window just before the stimulation by TMS and tSCS was calculated (Duclay et al. 2011; Saito et al. 2021). This was because a window of at least 200-ms is needed for the analysis given the resolution of the high-pass filtered EMG signals in the present study (i.e., cut-off frequency: 5 Hz). Peak-to-peak amplitudes of the MEPs and PRM reflexes during rest, knee extension, and flexion were calculated for SOL and TA, and peak-to-peak amplitudes during rest, plantar-flexion, and dorsi-flexion were calculated for VL and BF (Fig. 2). The amplitudes of MEPs and PRM reflexes across 10 trials were averaged for each condition.
The normality of the data distribution was investigated using the Kolmogorov–Smirnov test, and since the distribution of the data was partly non-Gaussian, non-parametric statistical tests were used. The amplitudes between the first and second responses evoked by tSCS with a double-pulse were compared using the Wilcoxon signed-rank test. Background EMG, MEP, and PRM reflex amplitudes of SOL and TA were analyzed among the three conditions (i.e., rest, knee extension, and flexion) by the Friedman test. Background EMG, MEP, and PRM reflex amplitudes of VL and BF were analyzed among the three conditions (i.e., rest, plantar-flexion, and dorsi-flexion) by the Friedman test. When a significant effect was found by the Friedman test, Scheffé’s test was performed as a post-hoc test for the pairs between resting and voluntary contraction conditions (i.e., rest and knee extension or flexion for SOL and TA; rest and plantar-flexion or dorsi-flexion for VL and BF). Data are expressed as means ± SD in the text.