**Exercise-Induced Fatigue in One Leg Does Not Impair the Neuromuscular Performance in the Contralateral Leg but Improves the Excitability of the Ipsilateral Corticospinal Pathway**

#### **Saied Jalal Aboodarda 1, Cindy Xin Yu Zhang 1, Ruva Sharara 1, Madeleine Cline <sup>1</sup> and Guillaume Y Millet 1,2,\***


Received: 5 September 2019; Accepted: 20 September 2019; Published: 25 September 2019

**Abstract:** To investigate the influence of pre-induced fatigue in one leg on neuromuscular performance and corticospinal responses of the contralateral homologous muscles, three experiments were conducted with different exercise protocols; A *(n* = 12): a 60 s rest vs. time-matched sustained left leg knee extension maximum voluntary contraction (MVC), B (*n* = 12): a 60 s rest vs. time-matched left leg MVC immediately followed by 60 s right leg MVC, and C (*n* = 9): a similar protocol to experiment B, but with blood flow occluded in the left leg while the right leg was performing the 60 s MVC. The neuromuscular assessment included 5 s knee extensions at 100%, 75%, and 50% of MVC. At each force level, transcranial magnetic and peripheral nerve stimuli were elicited to investigate the influence of different protocols on the right (tested) knee extensors' maximal force output, voluntary activation, corticospinal excitability, and inhibition. The pre-induced fatigue in the left leg did not alter the performance nor the neuromuscular responses recorded from the right leg in the three experiments (all *p* > 0.3). However, enhanced corticospinal pathway excitability was evident in the tested knee extensors (*p* = 0.002). These results suggest that the pre-induced fatigue and muscle ischemia in one leg did not compromise the central and peripheral components of the neuromuscular function in the tested contralateral leg.

**Keywords:** cross-over fatigue; isometric contraction; force; voluntary activation; transcranial magnetic stimulation

#### **1. Introduction**

Non-local muscle fatigue (NLMF) refers to a condition in which exercise-induced fatigue in a muscle group transiently impairs neuromuscular function in another limb [1]. In the absence of change in the contractile machinery of the non-exercised muscles (i.e., the index of peripheral fatigue), any attenuation in the neuromuscular function of these muscles is attributed to central mechanisms [2–4]. Presently, these central modulations are not well understood. However, it has been postulated that the fatiguing exercise-induced activation of group III and IV metabo- and mechano-receptors could compromise the motor control and execution of the non-fatigued muscles [5,6].

Transcranial magnetic stimulation (TMS) of the motor cortex in conjunction with electrical stimulation of the peripheral nerve (PNS) are non-invasive techniques that could be employed to investigate the central and peripheral mechanisms of fatigue [7]. Although several investigations have

explored the influence of lower limb exercise-induced fatigue on responsiveness of the corticospinal pathway innervating upper limb muscles (heterologous muscle groups) [3,5,8], currently, there is little empirical evidence pertaining to the influence of a fatiguing exercise performed by a lower limb muscle group on central voluntary activation and corticospinal responses of the contralateral homologous muscles.

A conventional research paradigm used to study the concept of NLMF includes a fatiguing exercise in one limb followed by the neuromuscular function assessment of the contralateral (non-fatigued) limb quantified during a single (4–6 s) maximum voluntary contraction (MVC) [9–13]. Given that in this approach the post-fatigue MVC is very brief, it has been postulated that participants might be able to overcome the sense of the pre-induced fatigue and maintain the MVC force output and central voluntary activation (VA) in the tested muscle group [4,10]. In an alternative paradigm, some investigators have recently used long-term sustained or intermittent MVCs performed by the tested muscle groups following the already fatigued contralateral limb [6,14,15]. The pre-induced fatigue in this approach has been shown to promote an augmented sense of fatigue and consequently, to attenuate the cycling performance in the tested contralateral limb [6]. However, there is little documented research that has studied this approach using sustained maximal isometric contractions. Halperin et al. [15] demonstrated a significant decline in the force output and VA (using the twitch interpolated technique) recorded during 12 (5 s) knee extension (KE) MVCs following a fatiguing exercise in the contralateral leg. On the contrary, Kennedy et al. [9] did not find any impairment in the MVC force, and VA recorded during 8 (2–3 s) KE MVCs. The controversy regarding these paradoxical findings was further fuelled by the fact that Kennedy et al. [9] maintained muscle ischemia in the fatigued leg to further activate the group III and IV muscle afferents, yet they did not find any change in MVC force and VA of the contralateral tested muscles. These findings argue against the idea that activation of group III and IV muscle afferents in one lower limb muscle could compromise the isometric performance in the contralateral limb. In addition to these controversies, none of the above-mentioned studies have explored the voluntary cortical activation (using TMS) nor the excitability and inhibition of the corticospinal pathway innervating the tested muscle groups following fatigue induced in the contralateral leg.

Thus, the purpose of the present study was threefold. In experiment A, we investigated whether fatiguing exercise in one leg (e.g., left) could alter the neuromuscular performance, corticospinal excitability, and inhibition of the tested (right) knee extensors recorded during brief maximal contraction performed immediately after the pre-induced fatigue. In experiment B, we explored whether pre-induced fatigue (in the left leg) combined with the subsequent sustained contraction (in the right leg) could alter the muscle performance and corticospinal responses in the tested (right) leg. In these two experiments, the group III and IV muscle afferents originating from the left knee extensors would stop firing after the termination of the sustained MVC in the leg [16]. Therefore, we hypothesized that the pre-induced left leg fatigue would not alter the neuromuscular performance recorded in the tested contralateral leg. In order to extend the input from group III and IV muscle afferents, we chose to maintain muscle ischemia in the fatigued (left) leg [16,17]. Thus, in experiment C, we investigated whether the prolonged activation of afferent receptors could modulate the neuromuscular performance and corticospinal responses in the tested (right) knee extensors. We hypothesized that maintaining blood ischemia in the fatigued leg could modulate the central motor drive and, consequently, the neuromuscular responses in the tested muscles.

#### **2. Materials and Methods**

#### *2.1. Participants*

Twelve healthy, recreationally active participants (six females: 22 ± 3 years, 168 ± 7 cm, 65 ± 1 kg; and six males: 25 ± 1 years, 175 ± 4 cm, 74 ± 9 kg) provided informed consent to participate in experiments A and B. Five participants from these two experiments and four additional participants

volunteered to complete experiment C (three females: 26 ± 3 years, 169 ± 9 cm, 66 ± 9 kg; and six males: 30 ± 4 years, 173 ± 6 cm, 76 ± 8 kg). Participants did not have any history of neurological, cardiovascular, or musculoskeletal injuries and were determined as right-leg dominant based on the preferred leg used to kick a ball [18]. They were instructed to refrain from rigorous physical activity, as well as ingesting caffeine and alcohol at least 24 hours prior to each testing session. Participants also completed the TMS safety checklist [19] and Physical Activity Readiness Questionnaire for Everyone PAR-Q+ form [20]. The procedures were conducted in accordance with the declaration of Helsinki and approved by the Health Research Ethics Authority of the University of Calgary (REB17-0147).

#### *2.2. Experimental Setup and Procedures*

The experimental protocol was performed on an isometric chair (Kin–Com, Chattecx Corporation, TN, USA) while the hips and knees were fixed at 90◦. To ensure that the upper body did not contribute to the KE force, straps along the trunk and waist were secured, and participants crossed arms to shoulders during contractions. Left and right legs were each secured inside a padded ankle cuff attached to a strain gauge (LC101-2K; Omegadyne, Sunbury, OH, USA) with a non-extensible strap. Force data was sampled at 2000 Hz, digitally converted using the Power Lab acquisition hardware and Lab Chart software (ADInstruments, Bella Vista, Australia), and monitored on a computer. The electromyography (EMG) was recorded from the left and right knee-extensors, and both TMS and PNS were employed during the neuromuscular evaluations [21].

Bipolar self-adhesive Ag/AgCl surface EMG electrodes (Kendall MediTrace, MA) were positioned on the muscle belly of the right and left vastus lateralis (VL), rectus femoris (RF), and biceps femoris (BF) [22]. A reference electrode was placed on the patella of the right leg. Before the placement of electrodes, the area of skin was shaved and abraded to remove dead skin with sandpaper and cleaned with an isopropyl alcohol swab to decrease skin resistance. An inter-electrode impedance of <5 kΩ was obtained prior to recording to ensure an adequate signal-to-noise ratio. All EMG signals were digitized at a sampling rate of 2000 Hz by the PowerLab system (16/30-ML880/P, ADInstruments) and amplified with an octal bio-amplifier (ML138, ADInstruments). EMG signals were bandpass filtered (5–500 Hz) and all data were analyzed offline using the Labchart 8 software (ADInstruments).

Single-pulse TMS was manually delivered to the left motor cortex via a 110 mm concave double-cone coil (maximum output of 1.4 T) connected to a magnetic stimulator (Magstim 2002 Company Ltd., Whitland, UK). The optimal location of the coil was determined on the scalp for every centimeter from 2 cm anterior and 1 cm to the left of the vertex (i.e., the midpoint from nasal–inion and tragi). Single pulses of TMS at 50% of maximum stimulator output (MSO) were delivered during brief (2–3 s) knee extensions at 20% of MVC and the location which evoked the highest twitch force and motor evoked potential (MEP) amplitudes for VL and RF (without an increase in BF MEP amplitude) was marked on a latex swim cap worn by participants. The optimal stimulator intensity was determined by delivering four stimuli at each intensity between 20 to 80% of the MSO (with 10% increments, in random order) and the intensity that elicited the highest twitch force, as well as VL and RF MEP amplitudes, was used for the rest of the session. The size of superimposed twitch (SIT), VL, or RF amplitude showed a plateau up to 80% of MSO for all participants, thus further increase in TMS intensities (90% or 100% MSO) was not necessary. The average stimulating intensity was 61 ± 8% (range: 40%–80%) of the maximal stimulator output for the six testing sessions.

Square-wave percutaneous electrical stimuli were delivered to the femoral nerve using a constant current stimulator (DS7A, Digitimer, Welwyn Garden City, UK). The cathode electrode (Kendall MediTrace, MA, USA) was secured on top of the inguinal triangle, and the anode (50 × 90 mm electrode, Durastick Plus, CA, USA) was placed in between greater trochanter and suprailiac projections. The stimuli intensity (pulse duration: 1000 μs, maximal voltage: 400 V) was increased in incremental steps until a maximal twitch force and compound muscle action potential (i.e., VL and RF Mmax) were observed. The supramaximal stimulation (i.e., 130% of stimulator output that achieved maximal twitch

and Mmax responses) was held constant throughout the protocol. The average stimulating intensity was 126 ± 52 mA (range: 65–185 mA) for the six testing sessions.

Experiment A (*n* = 12). Neurophysiological responses recorded immediately after a contralateral fatiguing exercise. In two randomly selected testing sessions, the fatigued (left) leg underwent either 60 s sustained KE MVC (Fatigue; Ftg-0) or 60 s rest (Rest-0) (Figure 1A). Before and immediately after these two protocols, the tested (right) leg performed a neuromuscular evaluation including a 5 s KE MVC superimposed with a PNS delivered on the force plateau and another PNS evoked 2 s after the MVC on the relaxed muscle (i.e., potentiated twitch, Pt). Three seconds after the Pt, subjects performed another 5 s MVC followed by contractions at 75% and 50% of MVC. Single-pulse TMS and PNS were delivered at 100%, 75%, and 50% of MVC with ~2 s interstimulus intervals, during which participants would reobtain the target force following each TMS interference to the voluntary force output. The resting interval between 100%, 75%, and 50% MVC was ~5 s. Two sets of neuromuscular evaluations were performed before and one set was performed immediately after each intervention. Additionally, immediately before and after each left leg protocol (either rest or 60 s sustained MVC), a PNS was evoked to the right leg at rest to account for any potential influence of contralateral protocols on contractile property and excitability of the tested muscles.

**Figure 1.** Details of experimental protocols. In experiment A (panel **A**) (*n* = 12), a 60 s rest or 60 s sustained left leg knee extension (KE) maximum voluntary contraction (MVC) were followed by a neuromuscular evaluation (NME) including 5 s sustained right leg KEs at 100%, 75%, and 50% of MVC. In experiment B (panel **B**) (*n* = 12), a 60 s rest or 60 s left leg MVC was immediately followed by a 60 s right leg MVC. In experiment C (panel **C**) (*n* = 9), a similar protocol as experiment B was performed while blood flow was occluded in the left leg while the right leg was performing a 60 s MVC.

Experiment B (*n* = 12). Neurophysiological responses recorded after a sustained MVC superimposed upon the pre-induced contralateral fatigue. In two randomly selected testing sessions (Figure 1B), participants undertook 60 s rest (Rest-60) or 60 s left leg sustained MVC (Ftg-60), followed immediately by 60 s sustained right leg MVC. Neuromuscular evaluations, as described in experiment A, were performed before and immediately after each experimental protocol.

Experiment C (*n* = 9). Influence of maintaining muscle ischemia in the fatigued contralateral leg on the neurophysiological responses of the tested muscles. In two randomly selected testing sessions, participants underwent the Ftg-60 protocol (see experiment B) with the exception of the "cuff session (Ftg-60-Occl), where blood flow to the left (fatigued) leg was occluded 2 s prior to the end of 60 s MVC (Figure 1C). A standard single-bladder adult thigh cuff connected to an automatic rapid inflation system (Hokanson E20 AG101, Bellevue, WA, USA) was located at the distal part of the thigh. The cuff was inflated to 300 mmHg using compressed air within 2 s.

#### *2.3. Data Analysis*

*Performance.* A force-time integral including the sum of the data points, multiplied by the sample interval during the 60 s sustained MVCs (performed by the tested leg) was calculated in experiments B and C.

*Neuromuscular function.* The maximal force output and the corresponding root mean square EMG [rmsEMG of the VL and RF muscles] was quantified during the first and last 2 s intervals of the sustained 60 s left and right leg MVCs, as well as over 500 ms prior to each TMS evoked during brief right leg MVCs. The VL and RF rmsEMG were normalized to the amplitude of the corresponding muscle compound action potential (Mmax) recorded during the same contraction to calculate the rmsEMG·Mmax−<sup>1</sup> ratio (rmsEMG100).

The central voluntary activation (VAPNS) was measured using the twitch interpolation technique [23] with the following formula, where Fb is the voluntary force output just before the PNS, D is the difference between Fb and the maximum SIT force evoked by the stimulus, FMAX is the highest force output during MVC, and FPt is the size of Pt force evoked at rest [24].

$$\text{VA (\%)} = 100 - \text{D} \times (\text{Fb/F}\_{\text{MAX}}) / \text{FPt} \times 100 \tag{1}$$

The voluntary cortical activation (VATMS) was assessed using the amplitude of SITs evoked by TMS during 5 s contractions at 100%, 75%, and 50% of MVC [25,26]. The *y*-intercept of the linear regression (*r*<sup>2</sup> > 0.9) between the SITs was computed to identify the estimated resting twitch (ERT). VATMS was then calculated using the following equation:

$$\text{VA}\_{\text{TMS}} = \left[1 - \text{(SIT/ERT)}\right] \times 100\tag{2}$$

In order to explore the excitability of the corticospinal pathway, the areas under the MEP and the Mmax signals were measured for both VL and RF muscles. The onset of MEP and Mmax were defined as the point at which the voltage trace became tangential to baseline in either the positive or negative direction. The responses to VL and RF MEP were normalized to the subsequent Mmax recorded during the same contraction to calculate MEP·Mmax-1 ratios (MEP100, MEP75, MEP50). The duration (ms) of the silent period (i.e., indicative of corticospinal inhibition) was assessed for VL and RF MEPs as the interval from the stimulus artifact to the return of the continuous EMG by visual inspection during 100%, 75%, and 50% of MVC contractions (SP100, SP75, SP50). The end of the SP for two participants in experiments A and B was not distinguishable, so the values were removed from the data pool.

#### *2.4. Statistical Analysis*

Statistical analyses were computed using SPSS software (version 23.0; SPSS, Inc., Chicago, IL, USA). Shapiro–Wilk and Mauchly tests were used to ensure the assumptions of normality and sphericity for all dependent variables, respectively. All variables were normally distributed. Greenhouse–Geisser

correction factor was applied when the assumption of sphericity was violated. Unless otherwise notified, sphericity assumed data were reported. For each experiment separately, a paired sample *t*-test was used to compare baseline measures and then a two-way repeated measure analysis of variances (ANOVA) was run to explore the effect of 2times (baseline vs. post-intervention values) × 2 conditions for all outcome variables. When ANOVAs showed significant main effects of times or conditions, Bonferroni post hoc test was used to compare values. When an interaction effect was observed, paired *t*-tests with Holm–Bonferroni corrections were applied. In one case where baseline measures revealed a significant difference between conditions (i.e., MEP100 in experiment A), the post-intervention values were normalized to the baseline (to account for the day to day variations), and then the paired *t*-test was used to compare the relative values between different conditions. Paired *t*-test was used to compare the force-time integral values (calculated during the 60 s MVCs performed by the tested leg) between the two conditions in each experiment. The effect size was calculated for repeated measure ANOVAs by converting partial eta-squared to Cohen's d and for paired comparisons by using the mean ± SD values [27]. According to Cohen [27], the magnitude of effect size was classified as small (0.2 ≤ d < 0.5), medium (0.5 ≤ d < 0.8), and large (d ≥ 0.8). The mean ± SD of variables are presented in the text, Figures 1–5, and the Table.

**Figure 2.** Representative traces from a single subject for the force output and VL and RF EMG signals at 100%, 75%, and 50% of MVC (panel **A**) and the MEPs and Mmaxs recorded from VL and RF at 100% of MVC at baseline and post-intervention levels following a 60 s rest (Rest-0) or 60 s knee extension MVC with the contralateral limb (Ftg-0) (panel **B**). The top and the bottom line of the shaded grey box (in panel **A**) represents 75% and 50% of MVC, respectively. In panel **B**, four MEPs and four Mmaxs recorded at baseline and two MEPs and two Mmaxs recorded at the post-intervention level are superimposed (see Figure 1 for more details).

**Figure 3.** Mean and SD of knee extension (right leg) maximum voluntary contraction (MVC), voluntary activation (VA) using transcranial magnetic stimulation (VATMS), and peripheral nerve stimulation (VAPNS) values normalized to the baseline. In experiments A (panels **A**,**D**,**G**), a 60 s rest (Rest-0) vs. a 60 s left leg MVC (Ftg-0) were compared. In experiment B (panels **B**,**E**,**H**), a 60 s rest (Rest-60) or a 60 s left leg MVC (Ftg-60) followed immediately by a 60 s right leg MVC were compared. In experiment C (panels **C**,**F**,**I**), Ftg-60 was performed at the absence or presence of a blood flow occlusion in the left leg (Ftg-60-Occl). \* Significantly different from baseline (time effect).

**Figure 4.** Mean and SD of potentiated twitch (Pt) and rectus femoris (RF) root mean square EMG (rmsEMG) values normalized to the baseline. In experiments A (panels **A**,**D**), a 60 s rest (Rest-0) or a 60 s left leg MVC (Ftg-0) were compared. In experiment B (panels **B**,**E**), a 60 s rest (Rest-60) or a 60 s left leg MVC (Ftg-60) followed immediately by a 60 s right leg MVC were compared. In experiment C (panels **C**,**F**), Ftg-60 was performed at the absence or presence of a blood flow occlusion in the left leg (Ftg-60-Occl). \* Significantly different from baseline (time effect).

**Figure 5.** Mean and SD of motor evoked potential normalized to the subsequent muscle compound action potential (MEP·Mmax−<sup>1</sup> [MEP]) and the duration of silent period (SP) recorded from the rectus femoris (RF) during MVC. The values are presented as a percentage of baseline. In experiments A (panels **A**,**D**), a 60 s rest (Rest-0) or a 60 s left leg MVC (Ftg-0) were compared. In experiment B (panels **B**,**E**), a 60s rest (Rest-60) or a 60 s left leg MVC (Ftg-60) followed immediately by a 60 s right leg MVC were compared. In experiment C (panels **C**,**F**), Ftg-60 was performed in the absence or presence of a blood flow occlusion in the left leg (Ftg-60-Occl). \* Significantly different from baseline (time effect), # significantly different from baseline (interaction effect).

#### **3. Results**

*Force changes during left (fatigued) leg sustained MVCs.* The decline in the maximal force output recorded during the 60 s left KE MVC in three experiments was: −55% (mean ± SD: 471 ± 137 to 209 ± 67 N) following Ftg-0 in experiment A, −54% (477 ± 133 to 218 ± 54 N) following Ftg-60 in experiment B, and −51% (529 ± 139 to 260 ± 35 N), and −56% (571 ± 170 to 251 ± 93 N) after Ftg-60 and Ftg-60-Occl in experiment C, respectively.

#### *3.1. Experiment A*

The pre-induced fatigue in the contralateral (left) leg did not alter MVC, rmsEMG100, Pt, Mmax, VAPNS, and VATMS recorded from the right (tested) leg in the Ftg-0 compared to Rest-0 condition (Figures 3A,D,G, and 4A,D). The resting twitch and corresponding Mmax evoked before and after 60 s left leg MVC (Figures 1A and 2B) also did not show any difference between the two conditions (data not shown). However, a significant interaction effect of time × group was observed for the corticospinal excitability (MEP·Mmax−<sup>1</sup> ratio) recorded from VL (F1,11 <sup>=</sup> 6.29, *<sup>p</sup>* <sup>=</sup> 0.031, d <sup>=</sup> 1.58), and RF (F1,11 = 13.33, *p* = 0.004, d = 0.84) at 100% MVC (MEP100). While, higher values were observed for VL (*p* = 0.032, d = 0.39) and RF (*p* = 0.020, d = 0.43) following the contralateral leg fatigue (in Ftg-0), a decline was observed for the resting condition (in Rest-0) (Figure 5A).

#### *3.2. Experiment B*

As expected, the MVC force (F1,11 = 50.9, *p* < 0.001, d = 4.29), Pt (F1,11 = 109.3, *p* < 0.001, d = 6.32), VAPNS (F1,11 = 7.21, *p* = 0.023, d = 1.69) and VATMS (F1,11 = 27.7, *p* = 0.001, d = 3.71) significantly declined (all time effects) following 60 s right leg MVC. However, the pre-induced fatigue in the left leg did not result in any significant difference in the neuromuscular performance and corticospinal excitability and inhibition measures between Rest-60 and Ftg-60 conditions (Figures 3B,E,H, 4B,E and 5B,E).

#### *3.3. Experiment C*

The VATMS values from one participant were removed from the data pool because the linear regression line exhibited *r<sup>2</sup>* < 0.9. Similar to experiment B, the MVC force (F1,8 = 84.3, *p* < 0.001, d = 6.13), Pt (F1,8 = 47.4, *p* < 0.001, d = 4.87), VAPNS (F1,8 = 16.8, *p* = 0.005, d = 3.09) and VATMS (F1,7 = 16.8, *p* = 0.003, d = 2.90) significantly declined and MEP100 (F1,8 = 29.9, *p* = 0.001, d =1.49) increased (all time effects) following 60 s right leg MVC. However, the pre-induced fatigue and the subsequent blood occlusion in the left leg (while the right leg was performing the 60 s MVC) did not result in any significant difference in the neuromuscular performance and corticospinal excitability and inhibition between Ftg-60 and Ftg-60-Occl conditions (Figures 3C,F,I, 4C,F and 5C,F).

The left leg pre-induced fatigue in experiments B and C did not alter the force-time integral values recorded across 60 s sustained right leg MVCs between Rest-60 (19,737 ± 6847 N·s) vs. Ftg-60 (18,264 ± 7413 Ns) as well as the Ftg-60 (22,422 ± 3645 Ns) vs. Ftg-60-Occl (22,744 ± 4197 Ns). In addition, the corticospinal excitability (MEP) and inhibition (SP) recorded from VL and RF during contractions at 100%, 75%, and 50% of MVC did not demonstrate any significance in the three experiments unless otherwise stated above. Since VL and RF rmsEMG, MEP, and SP demonstrated similar patterns, only the results of RF were presented in Figures 3–5.

#### **4. Discussion**

The most important findings of the present study are that (i) in line with our first hypothesis, the pre-induced left leg fatigue did not alter the right knee extensors neuromuscular function, while (ii) the VL and RF corticospinal excitability (MEP100) significantly increased following the 60 s contralateral MVC (in Ftg-0) compared to a time-matched rest (in Rest-0), and (iii) contrary to our second hypothesis, the pre-induced fatigue, in combination with the subsequent left leg blood occlusion did not alter the neuromuscular function or performance (i.e., force-time integral during 60 s MVC) nor corticospinal responses between Ftg-60-Occl vs. Ftg-60 condition. These results confirm that 60 s sustained MVC-induced fatigue and subsequent muscle ischemia in one lower limb muscle, activating the group III and IV muscle afferents, does not compromise the neuromuscular performance in the contralateral limb when performed in isometric conditions.

#### *4.1. MVC Force and Central Motor Drive*

Multiple studies have found that exercise-induced fatigue in one leg may impair the neuromuscular function in the contralateral leg (for review see [1]). However, the results of the current study, in line with several other investigations [9–13], have failed to show the NLMF effect in the lower limb muscles. This lack of observable NLMF is apparent in a research paradigm where the post-intervention neuromuscular function is assessed during a single, brief (4–6 s) MVC [6,11,28,29]. It was hypothesized that in this paradigm, despite a general sensation of fatigue created by the pre-induced fatiguing task, participants would be able to maintain the level of central motor commands to the contralateral tested muscles to prevent a deterioration in maximal performance. In line with this hypothesis, the indices of central fatigue, including rmsEMG100 as well as the central neural drive (measured by VAPNS) and neural drive associated with circuits at or above the cortical motor cells (measured by VATMS), did not change following Ftg-0 and Rest-0 conditions. These findings confirm that a 60 s MVC in one leg does not result in any central constraint to the performance of the tested contralateral muscles when the post-fatigue assessment includes a brief MVC [4,30].

In an alternative research paradigm, Halperin et al. [4,15] demonstrated that MVC force and VAPNS were compromised when 12 × 5 s MVCs (as opposed to a single MVC) were performed following the fatiguing contralateral limb. Contrary to their observation, however, Kennedy et al. [9] did not find any impairment in the MVC force and VAPNS recorded during 8 × 3 s MVCs, despite the fact that these investigators maintained muscle ischemia in the fatigued leg to further activate the group III and IV muscle afferents. The results of our study (in experiments B and C) support those reported by Kennedy et al. [9] indicating that adding 60 s sustained MVC in combination with subsequent muscle ischemia to the contralateral leg does not alter the central (rmsEMG100, VAPNS, and VATMS) and peripheral (Pt and Mmax) indices of fatigue in the tested limb. Of note, in experiments B and C, while VATMS demonstrated 18 to 25% reduction (Table 1, Figure 3E,F), only 3% to 10% decrement was observed for VAPNS (Table 1, Figure 3H,I). Although one may take these results to suggest that different sites within the central nervous system would play distinct roles in impaired maximal force output after the fatiguing task, limitations associated with the interpretation of the voluntary activation data [31] make it difficult to draw an explicit conclusion about the results. Nonetheless, pre-induced fatigue and muscle ischemia in the contralateral limb did not alter the contribution of cortical and motoneuronal circuits in the end-exercise central fatigue level.



*Brain Sci.* **2019** , *9*, 250

Note: The shaded boxes indicate that there was a significant time effect between baseline and post-test measures; # means significantly different from baseline (interaction effect).

As opposed to previous experiments that used intermittent brief MVCs as the fatiguing task for the tested leg [9,15], we used 60 s sustained MVC to generate high intramuscular pressure and decreased oxygen delivery due to muscle ischemia induced by sustained MVC [32]. The 60 s MVC was chosen to match the total duration of intermittent MVCs (12 × 5 s MVCs) used by Halperin et al. [15]. However, the left leg pre-induced fatigue in combination with the blood ischemia did not alter the force-time integral values recorded during the 60 s right leg sustained MVCs. These results suggest that the activation of group III and IV muscle afferents induced by a 60 s sustained MVC in one leg does not modulate the motor performance in the contralateral leg during either sustained or brief MVCs.

A point of deliberation in the present study may be that the magnitude of the pre-induced fatigue generated by the left leg was not enough to constitute an NLMF. Indeed, two 100 s sustained KE MVCs used by Halperin et al. [15] resulted in ~70% force drop in the fatigue limb whereas one 60 s MVC resulted in ~55% reduction in MVC in our study. However, the neuromuscular performance was not altered in the study by Kennedy et al. [9] despite a 77% decline in MVC force. We also minimized the time delay between the fatiguing tasks and the post-intervention neuromuscular assessments to avoid recovery of central and peripheral fatigue, yet no evidence of NLMF was found. This time delay in previous studies was approximately 15–60 s [9,15]. It is also worth clarifying that the failure in observing NLMF phenomenon in the current study might be associated with the size of muscle mass involved (i.e., knee extensors in one leg) and the mode of exercise (i.e., sustained MVC) performed during the fatiguing task. Prior investigations that used bilateral dynamic knee extensions [8], leg cycling [5], and arm cranking exercises [32] found NLMF in the rested/remote muscle groups. The accelerated neuromuscular fatigue within the tested muscle groups in these studies have been attributed to central inhibitory factors mediated by afferent feedback from pre-fatigued muscles. Nonetheless, further studies using a longer duration of sustained MVC and larger muscle mass during whole-body dynamic exercises are required to explore the influence of group III and IV muscle afferents on the NLMF phenomenon.

#### *4.2. Corticospinal Excitability and Inhibition*

The corticospinal pathway excitability recorded from RF and VL muscles demonstrated relatively greater values (MEP100) following 60 s contralateral MVC (in Ftg-0) compared to a time-matched rest (in Rest-0). Although the current study showed a small effect size, this so-called "cross-over facilitation" of the corticospinal pathway had previously been shown in the upper limb musculature [33] and was attributed to (i) transfer of excitatory signals from one hemisphere to another via the callosal commissure [34,35], (ii) increase in the excitability of motoneurons innervating the tested muscles via release of monoaminergic neuromodulators [36], and (iii) activation of additional brain regions including ipsilateral and contralateral prefrontal and sensorimotor areas as a compensatory mechanism to prevent reduction in maximal force output [37]. With the single-pulse TMS used in the present study, it is indeed difficult to confirm the contribution of the above-mentioned mechanisms; particularly because the MEP75 and MEP50 recorded from both VL and RF did not show an increase following contralateral fatiguing contraction. However, in support of these notions contributing to cross-over facilitation of MEP recorded during MVC (MEP100), Tanaka and Watanabe [37] have suggested that additional brain regions are involved in a task if participants choose to mobilize all of their mental effort to execute a "maximal" rather than a "submaximal" contraction.

The failure in observing cross-over facilitation of MEP100 following Ftg-60 (experiment B) and Ftg-60-Occl (experiment C) is not clear; however, this could be attributed to the 60 s right leg MVC. More specifically, the corticospinal excitability (i.e., MEP100) demonstrated a significant increase following the 60 s sustained right leg MVC (time effect) (Figure 5B,C). This MEP facilitation was observed regardless of the preceding conditions undertook by the left leg (i.e., 60 s MVC vs. 60 s MVC + occlusion in experiment C). Thus, our data suggest that the 60 s sustained MVC performed by the right (tested) knee extensors facilitated a strong excitatory effect to the corticospinal system while the contralateral fatiguing contraction did not enhance this facilitation any further.

Our data in all three experiments demonstrated no statistical difference in the duration of SP compared within or between conditions. This finding indicates that the pre-induced fatigue and the subsequent blood ischemia in one leg do not alter the rate of corticospinal inhibition in the contralateral homologous muscles recorded during 100%, 75%, and 50% of MVC contractions.

#### *4.3. Limitations*

Methodological considerations of the present study include: (i) The sequence of contractions in the neuromuscular assessment (i.e., 100%, 75%, and 50% of MVC) was kept consistent throughout the experiment whereas an early recruitment of high threshold motoneurons during the MVC could have offset the influence of contralateral contraction on MEP75 and MEP50, (ii) the size of MEP monitors the excitability of the pyramidal tract neurons as well as the spinal motoneurons and spinal interneurons [38]; thus, further research is required to elucidate the influence of pre-induced fatiguing exercises on the cortical and spinal excitability responses, (iii) some may argue that the antagonist muscle excitability (M-wave amplitude) and contractile property (peak twitch) might have affected the slope of the regression line used to calculate the estimated resting twitch in VATMS measurement, however, the hamstring Mmax and peak twitch were not measured due to complexity of the sciatic nerve electrical stimulation; therefore, caution should be taken in interpretation of the VATMS data, as it is difficult to determine the contribution of hamstring coactivation on TMS-evoked knee extensors SITs, (iv) pre-induced fatiguing exercise in one limb may modulate central and peripheral homodynamic associates such as cardiorespiratory responses, neurotransmitters, neuromodulators, hormonal factors as well as autonomic responses, however, these parameters were not directly measured in the current study, thus they were not discussed, and finally, (v) more investigation should be directed towards the measurement of intracortical facilitation and inhibition using the paired-pulse TMS paradigm.

#### **5. Conclusions**

In conclusion, the results of the present study suggest that the pre-induced fatigue evoked by the 60 s sustained KE MVC, with or without subsequent blood ischemia, in the fatigued leg does not compromise the performance and neuromuscular function of the tested contralateral leg. However, the pre-induced fatigue may enhance the responsiveness of the ipsilateral corticospinal pathway innervating the tested limb, provided that the neuromuscular evaluation in the tested muscles is performed immediately after the pre-induced fatigue.

**Author Contributions:** Conceptualization, S.J.A., C.X.Y.Z., M.C. and G.Y.M.; Data curation, S.J.A., C.X.Y.Z., R.S., M.C. and G.Y.M.; Formal analysis, S.J.A., C.X.Y.Z., R.S., M.C. and G.Y.M.; Funding acquisition, M.C. and G.Y.M.; Investigation, S.J.A., C.X.Y.Z., R.S., M.C. and G.Y.M.; Methodology, S.J.A. and G.Y.M.; Project administration, S.J.A., C.X.Y.Z. and M.C.; Resources, G.Y.M.; Supervision, S.J.A., R.S. and G.Y.M.; Validation, S.J.A.; Visualization, S.J.A.; Writing—original draft, S.J.A. and G.Y.M.; Writing—review & editing, S.J.A., C.X.Y.Z., R.S., M.C. and G.Y.M.

**Funding:** This research was partially funded by the Eyes High Postdoctoral Scholars, Program for Undergraduate Research Experience (PURE) and Markin Undergraduate Student Research Program (USRP) in Health & Wellness awards, University of Calgary, Alberta, Canada.

**Acknowledgments:** We would like to thank Elaheh Mousavian for working on raw data figure (Figure 2).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Corticospinal-Evoked Responses from the Biceps Brachii during Arm Cycling across Multiple Power Outputs**

**Evan J. Lockyer 1,2, Katarina Hosel 1,3, Anna P. Nippard 1,2, Duane C. Button 1,2 and Kevin E. Power 1,2,\***


Received: 14 July 2019; Accepted: 15 August 2019; Published: 19 August 2019

**Abstract:** *Background*: We examined corticospinal and spinal excitability across multiple power outputs during arm cycling using a weak and strong stimulus intensity. *Methods*: We elicited motor evoked potentials (MEPs) and cervicomedullary motor evoked potentials (CMEPs) in the biceps brachii using magnetic stimulation over the motor cortex and electrical stimulation of corticospinal axons during arm cycling at six different power outputs (i.e., 25, 50, 100, 150, 200 and 250 W) and two stimulation intensities (i.e., weak vs. strong). *Results*: In general, biceps brachii MEP and CMEP amplitudes (normalized to maximal M-wave (Mmax)) followed a similar pattern of modulation with increases in cycling intensity at both stimulation strengths. Specifically, MEP and CMEP amplitudes increased up until ~150 W and ~100 W when the weak and strong stimulations were used, respectively. Further increases in cycling intensity revealed no changes on MEP or CMEP amplitudes for either stimulation strength. *Conclusions*: In general, MEPs and CMEPs changed in a similar manner, suggesting that increases and subsequent plateaus in overall excitability are likely mediated by spinal factors. Interestingly, however, MEP amplitudes were disproportionately larger than CMEP amplitudes as power output increased, despite being initially matched in amplitude, particularly with strong stimulation. This suggests that supraspinal excitability is enhanced to a larger degree than spinal excitability as the power output of arm cycling increases.

**Keywords:** MEP; CMEP; arm cranking; motoneurone; exercise

#### **1. Introduction**

The influence of muscle contraction intensity on the excitability of the corticospinal pathway in humans has been well-studied during isometric contractions. Most of this research has involved the use of non-invasive stimulation techniques to assess corticospinal and/or spinal excitability to muscles of the upper [1–3] and, to a lesser extent, the lower limb [4] across a wide range of isometric contraction intensities. In general, the findings from these studies indicate that motor evoked potentials (MEPs) elicited by transcranial magnetic stimulation (TMS) increase in size as the strength of isometric muscle contractions increase up until a peak, after which they plateau and subsequently decrease as contraction strength approaches maximal ((i.e., 100% maximal voluntary contraction (MVC)) [1–4]. This modulation in MEP is accompanied by a similar change in the cervicomedullary MEP (CMEP) elicited by transmastoid electrical stimulation (TMES) of corticospinal axons, suggesting that the change in corticospinal excitability is largely mediated by spinal factors [1,4].

Using a strong stimulus intensity (set to evoke responses equal to 65–80% the maximal compound muscle action potential (Mmax)), Martin et al. (2006) showed that MEP and CMEP areas increased linearly in size during weak isometric contractions (i.e., <50% MVC) of the biceps brachii as muscle contraction intensity increased, whereas during strong contractions (i.e., >50% MVC) MEP and CMEP areas plateaued at ~75% MVC, and subsequently decreased as the contraction intensity approached 100% MVC [1]. When a lower stimulus intensity (set to evoke responses equal to 30–50% Mmax) was used, MEP and CMEP areas followed a similar pattern of modulation with contraction intensity, however, peak responses were not observed until ~90% MVC, after which MEP and CMEP areas decreased. Moreover, the decline in MEP and CMEP area with the lower stimulus intensity was less marked than that observed when the stronger stimulus intensity was used [1]. Thus, the intensity of stimulation is an important factor to consider in assessing corticospinal excitability given how it can influence the primary measurement(s), and the associated interpretation of the data.

Substantially less information, however, is available regarding the influence of muscle contraction intensity on the modulation of corticospinal excitability during rhythmic motor outputs such as those observed during cycling [5–7]. This is an important topic to consider given that rhythmic motor outputs, such as arm cycling, are partially generated by spinally located networks of interneurons referred to as central pattern generators [8,9], and that corticospinal excitability is modulated differently during rhythmic locomotor outputs than during isometric contractions, indicating task-specificity [5,10,11]. In two separate studies from our lab, we have investigated changes in corticospinal and spinal excitability as arm cycling intensity (i.e., power output) was increased [6,7]. However, changes in excitability were assessed across a small range of power outputs, and thus may not have observed potential changes in excitability that occurred at higher cycling intensities. Thus, it remains unknown whether a similar peak, plateau and subsequent decline in corticospinal and spinal excitability are observed with increasing arm cycling intensity, as observed in isometric contractions.

Accordingly, the purpose of the present study was to: (1) characterize the influence of muscle contraction intensity on changes in corticospinal and spinal excitability projecting to the biceps brachii over a wide range of arm cycling intensities, and (2) assess the influence of stimulation intensity on corticospinal and spinal outputs as cycling intensity increased. Specifically, we sought to examine the effects of using a weak and a strong stimulus intensity on corticospinal and spinal excitability as power output was increased during cycling. We hypothesized that: (1) using the weak stimulus, corticospinal and spinal excitability would increase similarly across all arm cycling power outputs, and (2) using the *strong* stimulus, corticospinal and spinal excitability would increase but experience a plateau, and subsequent decrease as cycling intensity increased towards the maximum power output examined.

#### **2. Materials and Methods**

#### *2.1. Participants*

This study consisted of a familiarization session and two experimental sessions; (1) a transcranial magnetic stimulation (TMS) session and (2) a transmastoid electrical stimulation (TMES) session (see *Protocol* below). A total of nine healthy, male volunteers (24.2 ± 5.9 years, 180.7 ± 7.8 cm, 82.2 ± 8.3 kg, 1 left-hand dominant) with no known neurological impairment participated in session one, and eight of those volunteers (1 left-hand dominant) returned on a separate day (>24 h) to complete session two. In accordance with the Tri-Council guidelines in Canada, all participants gave written, informed consent prior to participating in the study, and potential risks were fully disclosed. Prior to TMS, all participants were screened for contraindications to magnetic stimulation using a safety checklist [12]. To determine limb dominance, the Edinburgh handedness inventory [13] was used. This information was gathered because all evoked responses elicited by TMS and TMES (see *Stimulation Conditions* below) were taken from the dominant arm. Additionally, all participants filled out a Physical Activity Readiness Questionnaire for Everyone (PAR-Q+, Canadian Society for Exercise Physiology (CSEP)) to screen for any contraindications to physical activity. Participants also refrained from caffeine for 12 h

and alcohol for 24 h prior to each experimental session. All procedures were performed in compliance with the Declaration of Helsinki and were approved by the Interdisciplinary Committee on Ethics in Human Research (ICEHR no. 20181196-HK) at Memorial University of Newfoundland.

#### *2.2. Experimental Setup*

Many of the experimental procedures and recording techniques herein are similar to those described previously [6,7,14]. All sessions were conducted with participants seated upright on an arm cycle ergometer (SCIFIT ergometer, model PRO2 Total Body, Tulsa, OK, USA). The seat height of the ergometer was adjusted so that participants' shoulders were approximately in line with the axis of rotation of the arm cranks, and the seat distance was manipulated to a position in which participants were at a comfortable distance (i.e., no reaching or trunk variation during cycling) from the hand pedals. The seat height and distance were recorded for each participant during the familiarization session and were used for the subsequent sessions. Arm cycling trials were performed in an asynchronous cranking pattern with the forearms fixed in a pronated position. Wrist braces were worn to limit the amount of wrist flexion and extension during cycling as a means to diminish the influence of shortand long-latency reflex connections that have been shown to exist between the wrist flexors and the biceps brachii (see Figure 1) [15].

**Figure 1.** Experimental setup for arm cycling trials showing participant seated on the ergometer instrumented with surface EMG electrodes on the biceps and triceps brachii. Arrows point to the site of each stimulation technique. All arm cycling trials were conducted in the forward direction. Abbreviations: TMS, transcranial magnetic stimulation; TMES, transmastoid electrical stimulation; BB, biceps brachii; TB, triceps brachii; EMG, electromyography.

For this study, participants were required to cycle at 6 different power outputs: 25, 50, 100, 150, 200, and 250 Watts (W) all at a constant cadence of 60 revolutions per minute (rpm). These cycling conditions were repeated at two different stimulation intensities (see *Stimulation Conditions* below), for a total of 12 cycling trials.

#### *2.3. Electromyography Recordings*

Surface electromyography (EMG) was recorded from the biceps brachii of the dominant arm using pairs of disposable Ag-AgCl surface electrodes (MediTraceTM 130 Foam Electrodes with conductive adhesive hydrogel, Covidien IIC, MA, USA). Electrodes were positioned approximately 2 cm apart (center to center) over the midline of the biceps brachii and on the lateral head of the triceps brachii in a bipolar configuration. A ground electrode was positioned on the lateral epicondyle of the dominant arm. To reduce the impedance for EMG recordings, the skin was thoroughly prepared by removing hair (via a handheld razor), abraded to remove dead skin cells (via abrasive paper), and cleaned using isopropyl alcohol swabs prior to electrode placement. The EMG signals were amplified (×300, CED 1902 amplifier; Cambridge Electronic Design Ltd., Cambridge, UK), and bandpass filtered using a 3-pole Butterworth filter with cut-off frequencies of 10–1000 Hz. All analog signals were digitized at a sampling rate of 5000 Hz and stored on a laboratory computer for off-line analysis (CED 1401 interface and Signal 5.11 software; Cambridge Electronic Design Ltd., Cambridge, UK).

#### *2.4. Stimulation Conditions*

Recordings were made of the motor responses in the biceps brachii to three different stimulation techniques: (1) brachial plexus stimulation at Erb's point, (2) magnetic stimulation of the motor cortex (i.e., TMS), and (3) electrical stimulation between the mastoids at the cervicomedullary junction (i.e., TMES). Motor responses were evoked during arm cycling at the 6 o'clock position, which corresponds to the mid-elbow flexion phase of arm cycling and when biceps brachii activity is relatively the largest (for a more detailed explanation of the phases of arm cycling see review by [16]). Stimulations were triggered automatically when the right hand passed a magnetic sensor on the ergometer, at either the 6 o'clock or 12 o'clock position for right-handed and left-handed participants, respectively. The intensities for all three stimulation techniques were set during arm cycling at a constant cadence of 60 rpm and power output of 25 W. For TMS and TMES, two different stimulation intensities were used: (1) a weak stimulation intensity (set to evoke responses equal to ~10% Mmax), and (2) a strong stimulation intensity (set to evoke responses equal to ~40% Mmax). These response amplitudes were chosen to provide insight into potential differences in excitability at different portions of the motoneurone pool as cycling intensity increased. All participants had prior experience with each of the stimulation procedures before participating.

#### *2.5. Brachial Plexus Stimulation*

For both sessions, single rectangular pulses (200-μs duration, 90–275 mA) were delivered via a DS7AH constant current stimulator (Digitimer Ltd., Welwyn Garden City, Hertfordshire, UK) to the brachial plexus at Erb's point to elicit maximal compound muscle action potentials (maximal M-wave (Mmax)) in the biceps brachii. The cathode was placed in the supraclavicular fossa and the anode over the acromion process. Stimulus intensity was initially set at 25 mA and was gradually increased until the size of the M-wave plateaued (i.e., Mmax). At this point, the stimulation intensity was increased by 10% (supramaximal) to ensure that Mmax was elicited throughout the remainder of the study.

#### *2.6. TMS*

TMS was delivered over the vertex of the motor cortex to elicit MEPs in the biceps brachii using a Magstim 200<sup>2</sup> magnetic stimulator (Magstim, Whitland, Dyfed, UK) and circular coil (13.5 cm outside diameter). The vertex was measured and marked on the participant's scalp with a felt-tip permanent marker. One investigator ensured proper and consistent coil placement directly over vertex throughout the experiment. The coil was held firmly against the participant's skull, parallel to the floor with the direction of current flow-oriented to preferentially activate either the left or right motor cortex, depending on hand dominance (i.e., "A" side up for right-handed participants, "B" side up for left-handed participants). Initially, TMS intensity was set at 25% of maximal stimulator output (MSO) and was increased until MEPs were observed in the biceps brachii equal in amplitude to ~10% Mmax. Once found, a trial consisting of 8 TMS was performed to ensure that the average MEPs were ~10% Mmax. This stimulation intensity was recorded as the weak stimulation intensity, and was then used for the remainder of the experiment. For the strong stimulation intensity, the same procedures were performed except the %MSO was increased until MEPs from the biceps brachii were equal in amplitude to ~40% Mmax. Once again, a trial consisting of 8 TMS was performed to ensure that the intensity of TMS would evoke MEPs equal to ~40% Mmax. Once determined, this intensity was recorded and then used as the strong intensity for the rest of the experiment.

#### *2.7. TMES*

TMES was delivered (200 μs pulse-width duration, DS7AH, Digitimer Ltd., Welwyn Garden City, Hertfordshire, UK) to the corticospinal axons at the cervicomedullary junction to elicit CMEPs in the dominant arm biceps brachii. Self-adhesive Ag-AgCl surface electrodes were placed on the skin at the grooves between the mastoid processes and the occipital bone, with the anode and cathode on the side corresponding to each participant's dominant and non-dominant arm, respectively. Similar to the procedures for setting the stimulation intensities for TMS (see *TMS* above), the intensity of electrical stimulation was gradually increased (initially from 25 mA) until the amplitudes of the CMEPs were equal in amplitude to ~10% Mmax (for the weak stimulation intensity) and ~40% Mmax (for the strong stimulation intensity). Trials of 8 CMEPs were evoked at each stimulation intensity and the average was calculated. These stimulation intensities were recorded, and were then used for the remainder of the experiment. The latency of responses was monitored carefully to ensure that stimulation did not activate the corticospinal axons at or near the ventral roots, which would be indicated by a reduction in latency by ~2 ms [17,18].

#### *2.8. Protocol*

Following familiarization, participants were randomly assigned to complete either session one (TMS) or session two (TMES) first. For both sessions, the procedures were identical with the exception of the stimulation type. Following EMG preparation and ergometer modifications, stimulation intensities were determined (see above). In both sessions, Mmax was determined first followed by the setting of stimulation intensities for the weak and strong stimulations for either TMS (session one) or TMES (session two). Once stimulation intensities were determined, participants began the 12 cycling trials consisting of six power outputs (25, 50, 100, 150, 200, and 250 W) performed at a constant cadence of 60 rpm with either the weak or strong stimulation intensity (i.e., six cycling trials at each stimulation intensity). The order of the cycling trials was randomized for each participant. While cycling, as the dominant hand passed the 6 o'clock position, one Mmax and either six MEPs or six CMEPs (depending on the session) were evoked in a randomized order. The time between stimulations was 5–6 s. The total length of each trial was approximately 30 s. To reduce the potential influence of fatigue, one-minute rest periods were given following completion of the lower power output trials (i.e., 25, 50, 100 W), and two-minute rest periods were given after the higher power output trials (i.e., 150, 200, 250 W). Additionally, half-way through the 12 trials (i.e., after trial six), a 5-min rest period was given before the remainder of the trials were completed.

#### *2.9. Data Analysis*

For analysis of Mmax, MEP, and CMEP, the averaged peak-to-peak amplitudes from each cycling trial were measured from the biceps brachii of the dominant arm. Since Mmax is thought to represent the maximal response of the motor system [4], averaged MEPs (*n* = 6) and CMEPs (*n* = 6) from each trial were normalized to the Mmax within each cycling trial. Response latencies of all evoked responses were carefully monitored throughout all cycling trials as well. The latency for each response was classified as the duration from the stimulus artifact to the initial deflection in the voltage trace from baseline and was averaged across the total number of stimulation trials. Additionally, since the level of voluntary muscle contraction could potentially have an influence on changes in MEP and CMEP amplitudes, pre-stimulus EMG was measured from the rectified virtual channel created for the biceps and triceps brachii as the mean of a 50 ms window immediately prior to the stimulation artifact [14]. For two participants who completed CMEPs (*n* = 8), pre-stimulus EMG from the triceps brachii was not available due to a technical error during data collection. Therefore, the final sample size for CMEP pre-stimulus EMG data from the triceps brachii was *n* = 6.

#### *2.10. Statistical Analysis*

Group data are presented as means ± SD in the text and means ± SE in the figures (with *n* in the legends). All statistics were performed using IBM's SPSS Statistics (IBM SPSS Statistics for Windows, Version 23.0. Armonk, NY, USA: IBM Corp.). Mauchly's test was employed to assess the assumption of sphericity for repeated measures analysis. In cases where sphericity was violated, the appropriate correction was applied (i.e., Greenhouse Geisser or Huynh-Feldt) and the degrees of freedom were adjusted. Separate two-way repeated-measures ANOVAs were used to assess the effects of stimulation intensity and cycling intensity (and any interaction) on the Mmax, MEP, and CMEP amplitudes (both normalized to Mmax), the average pre-stimulus EMG, and the MEP/CMEP ratios. Post hoc pairwise comparisons were made between means using the Bonferroni correction. Additionally, because one of our aims was to examine the effects of cycling intensity on corticospinal excitability measures within each stimulation intensity (weak and strong), separate one-way repeated-measures ANOVAs were conducted for both the weak and strong stimulus on Mmax, MEP, and CMEP amplitudes (normalized to Mmax), pre-stimulus EMG, and MEP/CMEP ratios as cycling intensity increased. If a main effect was identified, post hoc pairwise comparisons were made between means using the Bonferroni correction. Independent samples *t*-tests were conducted to compare whether MEPs and CMEPs (normalized to Mmax) at both stimulation intensities were matched appropriately. To compare between MEP and CMEP amplitudes (normalized to Mmax) at each power output, independent sample *t*-tests were used with a Bonferroni correction. Paired samples *t*-tests were conducted on MEP/CMEP ratios between stimulation strengths (weak vs. strong) at each power output. All statistics were performed on group data and statistical significance was set at *p* < 0.05.

#### **3. Results**

Evoked responses (i.e., Mmax, MEPs, and CMEPs) were recorded from the dominant arm biceps brachii at two different stimulation intensities while participants performed arm cycling bouts over a range of contraction strengths. MEPs and CMEPs (normalized to Mmax) were evoked on separate days but were initially matched to equal 10% (weak stimulus) and 40% (strong stimulus) of the Mmax on each day. MEPs and CMEPs were not significantly different when either the weak or the strong stimulation intensity were examined (*p* > 0.05 for both conditions), suggesting that the responses were indeed matched initially between days.

#### *3.1. Biceps Brachii Evoked Responses*

#### 3.1.1. MEP Amplitude

Figure 2 (top panel) and Figure 3A show representative and grouped data, respectively for MEP amplitudes from the biceps brachii during arm cycling across the various contraction intensities. Figure 2 shows evoked potential traces from one participant during arm cycling with the weak stimulation intensity. In this example, the amplitudes of the MEPs show a progressive and generally consistent increase from the lowest (25 W) to the highest (250 W) arm cycling/muscle contraction intensity. Results from the two-way ANOVA on MEP amplitudes showed a significant main effect for both stimulation intensity (strong > weak, *F*5,40 = 96.81, *p* < 0.001) and cycling intensity (*F*1,8 = 65.30, *p* < 0.001). Bonferroni post hoc tests revealed that MEP amplitudes at 25 W and 50 W were not different from one another (*p* = 0.187) but were significantly smaller than MEP amplitudes evoked

during the 100, 150, 200, and 250 W trials (*p* < 0.05 for all comparisons). Additionally, there was a significant interaction between the intensity of stimulation and the intensity of cycling on MEP amplitudes (*F*5,40 = 65.30, *p* < 0.001). Further analysis, through use of one-way ANOVAs for each stimulation intensity, showed a significant main effect for cycling intensity on MEP amplitudes at both the weak (*F*5,40 = 55.61, *p* < 0.001) and strong (*F*5,40 = 41.28, *p* < 0.001) stimulation conditions. Using the weak stimulation, Bonferroni post hoc tests revealed that MEP amplitudes increased as cycling intensity increased up until 200 W (200 W > 150 W >100 W > 50 W > 25 W, *p* < 0.05 for all comparisons) after which MEPs plateaued (*p* > 0.05). Using the strong stimulation, MEP amplitudes similarly increased with cycling intensity, however, a peak was observed at 100 W (100 W > 50 W > 25 W, *p* < 0.05 for all comparisons), at a lower power output than that observed using the weaker stimulation condition (i.e., 200 W). Beyond 100 W, there were no further increases in MEP amplitudes (*p* > 0.05).

**Figure 2.** Raw traces for MEPs (top row), CMEPs (middle row), and Mmax (bottom row) from the biceps brachii of a single participant (*n* = 1) across arm cycling power outputs using the weak stimulation intensity. Each MEP and CMEP waveform represent the average of six evoked potentials. Arrows indicate the stimulus artifact, and dashed lines portray the initial amplitudes of evoked potentials with the *weak* stimulation (~10% Mmax). In this example, MEP and CMEP amplitudes show a general progressive increase as power output increases towards 250 W, while Mmax gradually decreases.

#### 3.1.2. Biceps Brachii Pre-stimulus EMG

Figure 3C shows group data for biceps brachii pre-stimulus EMG prior to MEPs during arm cycling. Results from the two-way ANOVA showed that mean biceps brachii pre-stimulus EMG in the 50 ms preceding an MEP was not different between the weak and strong stimulation intensity (*F*1,8 = 1.42, *p* = 0.267). Therefore, the average pre-stimulus EMG was pooled between the weak and strong stimulation conditions, which are represented in Figure 3C. There was a significant main effect on biceps brachii pre-stimulus EMG for cycling intensity (*F*1.76,14.12 = 29.33, *p* < 0.001), but, there was no interaction between stimulation intensity and cycling intensity (*F*1.96,27.35 = 1.96, *p* = 0.137). To further examine changes in pre-stimulus EMG with cycling intensity, one-way ANOVAs were performed. Pre-stimulus EMG increased as cycling intensity increased up until 200 W (Figure 3C, *p* < 0.05), and no differences were observed between the 200 W and 250 W conditions (*p* = 1.00).

**Figure 3.** (**A**,**B**) Normalized grouped data (means ± SE) of the peak-to-peak amplitudes for MEPs (**A**) and CMEPs (**B**) obtained from the biceps brachii at each power output examined. MEPs and CMEPs were normalized to Mmax at each corresponding cycling intensity. In both A and B, filled data points represent when the weak stimulus was used, while unfilled points represent data from the strong stimulus. For clarity, circles were used for MEPs, while triangles were used for CMEPs. In some cases, data points are bigger than SE bars. \* Significant difference between illustrated data points. † Significant main effect for stimulation strength (*p* < 0.05). (**C**,**D**) Pre-stimulus EMG (means ± SE) from the biceps brachii which has been pooled and averaged between both stimulation intensities for the TMS session (**C**) and TMES session (**D**), respectively. # Significant difference between all data points. (**E**,**F**) Pre-stimulus EMG (means ± SE) from the triceps brachii which has been pooled and averaged between both stimulation intensities for the TMS session (**E**) and TMES session (**F**), respectively. \$ denotes significant difference from all previous power outputs. ¥ denotes significant difference from the 25 W condition. € denotes significant difference from the 25, 50, and 100 W conditions. ¢ denotes significant difference from the 25, 50, 100, and 150 W conditions.

#### 3.1.3. Triceps Brachii Pre-Stimulus EMG

Figure 3E shows group data for triceps brachii pre-stimulus EMG prior to MEPs. Similar to the biceps, results from the ANOVA showed no effect of stimulation intensity on triceps brachii EMG activity prior to a MEP (Figure 3E, *F*1,8 = 0.100, *p* = 0.760), but there was a significant main effect of cycling intensity (*F*1.62,12.94 = 19.32, *p* < 0.001). Also, there was no significant interaction between cycling intensity and stimulation intensity (*F*5,40 = 0.803, *p* = 0.554). To further examine the effect of cycling intensity on triceps brachii pre-stimulus EMG, one-way ANOVAs were performed using the pooled data. Results from these tests indicated that as cycling intensity increased, triceps brachii pre-stimulus EMG values were only significantly different at 150 W and 200 W. Specifically, triceps brachii pre-stimulus EMG was larger at 150 W than at 100 W (*p* = 0.006) and was larger at 200 W than 150 W (*p* = 0.044).

#### 3.1.4. CMEP Amplitude

Figure 2 (middle panel) and Figure 3B show representative and grouped data, respectively for CMEP amplitudes during the arm cycling bouts. Figure 2 portrays data from one participant from the weak stimulation intensity condition. Similar to the MEP amplitudes, in this example, CMEP amplitudes increase in a relatively consistent and progressive manner. The results from the two-way ANOVA on CMEP amplitudes showed significant main effects for both stimulation intensity (strong > weak, *F*1,7 = 91.50, *p* < 0.001) and cycling intensity (*F*3.81,26.65 = 20.16, *p* < 0.001), however, there was no significant interaction between the two factors (*F*5,35 = 1.34, *p* = 0.271). For cycling intensity, Bonferroni post hoc analysis revealed that CMEPs at 25 and 50 W are smaller than those at all other cycling intensities (i.e., 100, 150, 200, and 250 W) (*p* < 0.05 for all comparisons). To decipher specific effects of cycling intensity within each stimulation condition, separate one-way ANOVAs for the weak and strong stimulation conditions were performed on CMEP amplitudes. The results from the one-way ANOVAs showed a significant main effect for cycling intensity on CMEP amplitudes at both the weak (*F*5,35 = 21.11, *p* < 0.001) and strong (*F*5,35 = 9.95, *p* < 0.001) stimulation conditions. For the weak stimulation condition, Bonferroni post hoc analyses revealed that CMEP amplitudes increased up until 150 W (150 W > 100 W > 50 W > 25 W; *p* < 0.05 for all comparisons), after which CMEP amplitudes did not change (*p* > 0.05). When the strong stimulation intensity was used, post hoc analyses revealed that CMEP amplitudes increased up until 100 W (100 W > 50 W > 25 W, *p* < 0.05 for all comparisons), after which CMEPs plateaued (*p* > 0.05).

#### 3.1.5. Biceps Brachii Pre-Stimulus EMG

Figure 3D shows group data for biceps brachii pre-stimulus EMG prior to CMEPs during arm cycling. Results from the two-way ANOVA showed that mean biceps brachii pre-stimulus EMG in the 50 ms preceding CMEPs was not influenced by stimulation intensity (*F*1,7 = 0.02, *p* = 0.906), thus the data was pooled between the weak and strong stimulation conditions as shown in Figure 3D. There was a significant main effect on biceps brachii pre-stimulus EMG for cycling intensity (*F*1.49,10.41 = 43.08, *p* < 0.001), but, there was no interaction between stimulation intensity and cycling intensity (*F*5,35 = 1.22, *p* = 0.320). To further examine changes in pre-stimulus EMG with cycling intensity, one-way ANOVAs were performed using the pooled data. Similar to MEPs, pre-stimulus EMG for CMEPs increased as cycling intensity increased up until 200 W (Figure 3D, *p* < 0.05), and there was no difference between the 200 W and 250 W conditions (*p* = 0.885).

#### 3.1.6. Triceps Brachii Pre-Stimulus EMG

Figure 3F shows group data for triceps brachii pre-stimulus EMG prior to CMEPs. Similar to above, results from the two-way ANOVA showed no effect of stimulation intensity (*F*1,5 = 0.761, *p* = 0.423) and thus, the data was pooled between the week and strong stimulation intensities (Figure 3F). There was, however, a significant main effect of cycling intensity (*F*1.31,6.55 = 14.04, *p* = 0.006) on triceps brachii pre-stimulus EMG, but no significant interaction (*F*5,25 = 0.961, *p* = 0.460). To further examine the effect of cycling intensity on triceps brachii pre-stimulus EMG, one-way ANOVAs were performed using the pooled data. Results from these tests indicated that triceps brachii pre-stimulus EMG values for CMEPs were only increased at 150 W, 200 W and 250 W compared to the 25 W condition (*p* < 0.05 for all comparisons). However, triceps brachii pre-stimulus EMG was not significantly different with increased cycling intensity from 150 W to 250 W (*p* > 0.05 for all comparisons).

#### 3.1.7. MEP/CMEP Ratios

Although MEPs and CMEPs were evoked on separate days, the responses were initially matched in amplitude to approximately 10% or 40% Mmax for the weak and strong stimulation conditions, respectively (*p* > 0.05 for both stimulation conditions). Thus, MEP amplitudes were expressed relative to CMEP amplitudes and multiplied by 100% to obtain MEP/CMEP percentages for each participant (Figure 4). This was done in an attempt to isolate whether changes in overall excitability could be attributed to changes in supraspinal and/or spinal excitability. Values greater than 100% indicate that MEP amplitudes are larger than CMEP amplitudes, suggesting that supraspinal excitability may be increased. Similarly, values less than 100% indicate that MEP amplitudes are less than CMEP amplitudes, suggesting that changes in spinal excitability are important factors in maintaining excitability of the corticospinal pathway. Results from the two-way ANOVA revealed a significant main effect for stimulation intensity (weak > strong, *F*1,7 = 6.94, *p* = 0.034) and cycling intensity (*F*5, 35 = 9.71, *p* < 0.001). Bonferroni post hoc tests revealed that MEP/CMEP at 25 W and 50 W were not different from one another (*p* = 0.413) but were significantly smaller than MEP/CMEP at 100, 150, 200, and 250 W trials (*p* < 0.05 for all comparisons). As well, there was a significant interaction effect (*F*5, 35 = 8.18, *p* < 0.001) between stimulation intensity and cycling intensity on MEP/CMEP ratios. To examine changes in MEP/CMEP with increased power output, one-way ANOVAs were conducted within each stimulation intensity. Results from the one-way ANOVAs showed a significant main effect for cycling intensity on MEP/CMEP ratios at both the weak (*F*5,35 = 9.44, *p* < 0.001) and strong (*F*5,35 = 4.60, *p* = 0.003) stimulation conditions. When the weak stimulation was used, Bonferroni post hoc analysis revealed that MEP/CMEP were only significantly larger than that at 25 W at 150 W (*p* = 0.037), and 200 W (*p* = 0.05). When the strong stimulation intensity was used, MEP/CMEP were significantly larger at 50 W than at 25 W (*p* = 0.026) but were not different for any other comparison. To compare changes in MEP/CMEP between the weak and strong stimulation intensities, paired samples *t*-tests were performed at each power output. Thus, a total of six comparisons were made. The *t*-tests revealed that the MEP/CMEP ratios were not significantly different at 25 W (*t*(7) = 1.22, *p* = 0.261) or 50 W (*t*(7) = 0.52, *p* = 0.622) when either the weak or strong stimulus was used. However, MEP/CMEP ratios were significantly larger at 100 W (*t*(7) = 2.51, *p* = 0.041), 150 W (*t*(7) = 3.24, *p* = 0.014), 200 W (*t*(7) = 3.03, *p* = 0.019), and 250 W (*t*(7) = 2.41, *p* = 0.047) when the weak stimulation was used.

**Figure 4.** Comparison of MEP/CMEP ratios for the weak (filled circles) and strong (unfilled circles) stimulation intensities as power output increased from 25 W to 250 W. \* represents significant difference between stimulation intensities at each given power output (*p* < 0.05). In some cases, SE bars were smaller than the symbols for the data points.

#### 3.1.8. Mmax Amplitude

For both the TMS and TMES sessions, the results from the two-way ANOVA revealed similar effects on biceps brachii Mmax amplitudes. For both sessions, there was no effect of stimulation intensity (TMS: *F*1,8 = 0.093, *p* = 0.769, TMES: *F*1,7 = 1.06, *p* = 0.337), but there was a significant main effect for cycling intensity (TMS: *F*5,40 = 15.66, *p* < 0.001; TMES: *F*1,7 = 8.89, *p* < 0.001) on Mmax amplitudes (Figure 5). As cycling intensity increased Mmax amplitudes decreased (Figure 5A,B). Additionally, there was no interaction observed between factors on either day (TMS: *F*5,40 = 0.836, *p* = 0.532, TMES: *F*5,35 = 0.430, *p* = 0.825). Since there was no effect of stimulation intensity on Mmax values, the averages from each stimulation condition (weak and strong) were pooled across the cycling intensities for each session (as shown in Figure 5). For cycling intensity, Bonferroni post hoc analyses indicated that Mmax values decreased for the TMS and TMES session as cycling intensity increased from 25 to 250 W.

**Figure 5.** Changes in Mmax amplitudes with increasing power output pooled between stimulation intensities for the TMS (**A**) and TMES (**B**) session. \* denotes significant main effect of power output on Mmax amplitude. Mmax decreased by approximately 24.9 and 31.7% as power output increased from 25 to 250 W for the TMS and TMES sessions, respectively.

#### **4. Discussion**

This study shows that the amplitudes of TMS-evoked MEPs and TMES-evoked CMEPs increase with power output and plateau, but do not decrease in amplitude as has been previously shown by others during intense tonic contractions [1,4]. MEP amplitudes were much larger than CMEP amplitudes as power output increased regardless of stimulation strength, despite being initially matched in amplitude (Figures 3A,B and 4). This finding suggests that supraspinal factors mediate the change in overall corticospinal excitability observed during arm cycling as intensity increases. Importantly, stimulus strength had a substantial effect on MEP and CMEP amplitudes as cycling power output increased. Responses evoked by the weak stimulation (10% Mmax) increased up to approximately 200 W for MEPs (Figures 3A and 4) and 150 W for CMEPs (Figures 3B and 4), whereas with the strong stimulation (40% Mmax), responses reached a peak at 100 W for both MEPs and CMEPs and did not change afterward. Thus, the MEP/CMEP ratio used as a measure of supraspinal excitability was influenced by stimulation strength, which would lead to different conclusions on mechanisms of enhanced corticospinal excitability during arm cycling as power output increases.

#### *4.1. Modulation of Corticospinal and Spinal Excitability with Cycling Intensity*

Past research involving isometric contractions has shown that biceps brachii MEPs and CMEPs increase up until a peak at ~75–90% MVC [1–3], a finding which has been attributed to the motor unit firing and recruitment characteristics of the biceps brachii during progressively stronger isometric contractions [19,20]. Following the peak, there is a subsequent decline in responses as contraction intensity approaches 100% MVC [1] which is thought to reflect the inability for some motoneurones to fire in response to artificial excitatory input at strong contraction strengths, given the high degree of voluntary input to the motoneurone pool and the associated changes in their intrinsic properties [1]. In the present study, we did not observe a decline in corticospinal excitability as arm cycling intensity increased to the maximum intensity employed. Instead, we observed a plateauing of responses for both MEPs and CMEPs at intensities below 250 W, which were differentially influenced by stimulus strength (Figure 3A,B). Our results, however, do coincide with findings from the only other study to examine corticospinal excitability changes during a locomotor-like output over a wide range of contraction intensities [5]. In that study, MEPs and CMEPs from the knee extensors during leg cycling increased in amplitude up to 300 W, after which there was a plateauing, but no decline as cycling

intensity increased to 400 W [5]. Taken together, these studies suggest task-dependent changes in corticospinal and spinal excitability may be present, a finding we have previously reported [10,16,21].

In the current study, MEP and CMEP amplitudes increased at the lower, but not higher power outputs (Figure 3A,B), suggesting that the increase in overall corticospinal excitability at the low intensities (i.e., 25 to 100 W) is partially generated by increased spinal excitability. These finding are partially supported by biceps brachii pre-stimulus EMG values which increase for both stimulation types (Figure 3C,D) at the low cycling intensities, but are not significantly different between the highest cycling intensities (200 and 250 W). While this may explain the enhanced spinal excitability at the low power outputs, it does not explain why we observed a plateau in CMEP amplitudes beyond 150 W for the weak stimulus and 100 W for the strong stimulus in the present study, since EMG was still increasing beyond these power outputs. It is noted, however, that Weavil and colleagues showed increased EMG and workloads without changes in MEP and CMEP amplitudes. During isometric contractions, the biceps brachii is capable of recruiting additional motor units during contractions up to and beyond 90% MVC [19,20], which help to explain why CMEPs continue to increase beyond 90% MVC [1]. Corticospinal excitability to the biceps brachii is also task- [10,16] and forearm position-dependent [21] which is an important consideration when a comparison to tonic contractions is made. However, the lack of increase in CMEP amplitudes beyond 150 W and 100 W during arm cycling in the current study, while MEPs and background EMG are still increasing is unlikely to be explained by reaching the maximum motor unit recruitment of the biceps, given that these cycling intensities are not maximal, at least relative to a sprint test [7]. It is possible, however, that motoneurone recruitment strategies during a rhythmic motor output such as arm cycling may be different from those observed during isometric contractions (Power et al., 2018), and therefore could cause motoneurones to be maximally recruited sooner than 90% of maximal cycling power. Work in adult decerebrate cats and rats, for example, demonstrated that spinal motoneurones are characterized by changes in their electrical properties during locomotor outputs that would act to enhance their recruitment and firing [22–24]. These same changes in motoneurone excitability do not occur during tonic motor output [23].

#### *4.2. Modulation of Supraspinal Excitability with Cycling Intensity*

In the current study, MEP/CMEP ratios increased with power output, in particular when the weak stimulation intensity was used (Figure 4) suggesting that supraspinal excitability was enhanced to a larger degree than spinal excitability. It is plausible that changes in the excitability of interneuronal circuits and/or interhemispheric connections may be involved. During tonic contractions, short-interval intracortical inhibition (SICI) is reduced as muscle contraction intensity increases [25–27], a finding that is thought to downregulate the activity of the inhibitory neurons which project onto corticospinal cells involved in producing the movement. We recently showed that SICI was present during arm cycling, albeit not different than a tonic contraction [28]. Thus, it is possible that reductions in SICI during arm cycling as power output increases may underlay increases in MEP amplitudes as has been shown during tonic contractions.

Another potential mechanism involves cortical spread from the non-dominant to the dominant motor cortex as we have previously hypothesized [6,7,14]. Since arm cycling is a bilateral motor output it is possible that cortical excitation arising from the active, non-dominant motor cortex could facilitate excitability in the dominant motor cortex, which could reduce the input required to induce an MEP by a given TMS pulse. However, when the strong stimulation intensity was used, the changes in MEP/CMEP ratios were less marked and did not increase as cycling intensity increased suggesting a ceiling effect in the MEP amplitudes had been reached.

#### *4.3. Di*ff*erences between Stimulation Intensities*

This study highlights the importance of stimulation intensity selection for experimental design during locomotor outputs. Notably, MEPs continued to increase with cycling intensity up until approximately 200 W when elicited with weak stimulation intensity (10% Mmax), while they plateaued at approximately 100 W under strong (40% Mmax) stimulation. This led us to conclude that supraspinal excitability increases with increased power output, an effect only observed when weak stimulus intensity was used. In contrast, using the strong stimulation intensity leads one to believe, perhaps falsely, that spinal factors were driving the change in overall corticospinal excitability as a function of power output, a conclusion also reached by Weavil and colleagues (2015) who used a strong stimulation intensity (MEPs and CMEPs were ~50% Mmax). The use of a weak stimulation intensity yielded a more precise measure of corticospinal excitability in this specific study as MEPs were less susceptible to ceiling effects than at the strong stimulation.

#### *4.4. Methodological Considerations*

An important methodological consideration in interpreting the current data is that we did not make the power outputs relative to each individual as we have recently done in two separate studies during arm cycling [6,7]. In Spence et al. (2016) we used 5 and 15% of peak power output determined by a sprint test (modified Wingate) while in Lockyer et al. (2018) we used 20, 40, and 60% of peak power output determined via a standard incremental aerobic test (20 W increases every two minutes) [29]. These methods were not without limitations, however. The former used a sprint test to prescribe aerobic cycling intensity at 60 RPM and the latter incremental test resulted in most of the participants reaching a similar peak power output of ~120 W. In the present study we used absolute power outputs as has been used by others [5,30] and all participants were able to cycle well above the aerobic test maximum power output of 120 W obtained in our prior work. We were thus able to have participants cycle at supramaximal intensities, albeit we did not quantify exertion levels. Additionally, the sample size of (*n* = 9) for MEPs and (*n* = 8) for CMEPs was not determined by a power analysis and therefore, it is unclear whether a larger sample size would have influenced the present results.

#### **5. Conclusions**

The present study describes the influence of stimulation strength over a wide range of cycling intensities on corticospinal and spinal excitability during arm cycling. We have demonstrated that corticospinal excitability to the biceps brachii is increased with cycling intensity during low power outputs, a finding that is partially mediated by spinal factors. As cycling intensity increases, however, it appears as though supraspinal factors may play more of a role in modulating overall corticospinal excitability. Additionally, this study highlights the importance of stimulation intensity selection to assess corticospinal excitability during motor output. It is concluded that the use of a weaker stimulation intensity provides a more precise measure of corticospinal excitability during locomotor outputs at high intensities as they are less susceptible to potential ceiling effects.

**Author Contributions:** All authors participated in data collection, analysis, interpretation and manuscript preparation. All authors have approved the submitted version of this manuscript.

**Funding:** This research was funded by the Natural Sciences and Engineering Research Council of Canada, grant number (NSERC-#RGPIN-2015-05765).

**Acknowledgments:** This study was supported by PGS-D NSERC funding to E.J.L. as well as NSERC Discovery Grant to K.E.P. We would like to thank Thamir Alkanani for technical support and the participants for volunteering their time.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
