Muscle fatigue is specified as an exercise-induced decrease of the force generated by a muscle (Edwards et al., 1977). It is caused by alterations of structures involved in movement generation from the brain to the muscle cells. A progressive failure to drive motor neurons (MNs) during exercise has been termed central fatigue (Gandevia, 2001). It is associated with changes of excitability of the involved neural cells within the central nervous system (CNS) at spinal and supraspinal sites. Therefore, the size of motor evoked potentials (MEPs) elicited by transcranial magnetic stimulation (TMS) of the brain is altered after fatiguing contractions of muscles (Brasil-Neto et al., 1993; McKay et al., 1995; Liepert et al., 1996; Samii et al., 1997; Gandevia et al., 1999; Pitcher and Miles, 2002; Andersen et al., 2003; Rösler et al., 2009).

In a previous study on TMS-induced responses during muscular fatigue, we used a novel technique of TMS during a fatiguing exercise and demonstrated a dramatic decrease of the MEPs that could be evoked, indicating a decreased responsiveness of the MNs to synaptic input (Rösler et al., 2009). This decrease was paralleled by a significant drop of muscle force. We suggested an important supraspinal contribution mediated by intracortical inhibition and voluntary drive, while spinal mechanisms seem to have little effect on the MEP reduction. While the decrease of muscle force was fairly similar in all subjects, a remarkable inter-individual variability in the behavior of the TMS evoked responses was observed. A possible cause of these inter-individual differences were differences of the individual state of muscle training among the subjects. If this was the case, then muscular training could influence the reduced MN activation during a fatiguing exercise.

Muscular strength training is known to lead to adaptations not only within the muscle itself, but also within the CNS, as summarized in a review by Kidgell and Pearce (2011). Central nervous system adaptations to training have been inferred by the observation that at the beginning of a strength training program, muscle force increases without concomitant increase of muscle cross-sectional area (Lüthi et al., 1986; Griffin and Cafarelli, 2005). Hence, at the beginning of a training regimen, when structural adaptations within the muscle tissue have not yet occurred, CNS adaptations may precede and serve to increase the efficiency of a muscle generating its force output. Strength training does not increase the size of MEPs when the measurements are done in rested, unfatigued muscles at rest (Carroll et al., 2002; Jensen et al., 2005; Duchateau et al., 2006). The present experiments were done to test the hypothesis that a 3 weeks resistance training program could induce CNS adaptations which in turn would alter the MNs’ susceptibility to TMS during fatigue. In contrast to previous studies, we quantified the size of MEPs using the triple stimulation technique (TST), which increases the sensitivity for detection of small changes (Magistris et al., 1998; Humm et al., 2004). Moreover, we measured the MEPs not only in rested muscles but also during a fatiguing exercise, since the size of MEPs changes dramatically in fatigued muscles.

All subjects tolerated the experiments and the training well without any adverse event. All subjects completed all measurements, both before and after training. Cooperation during the training period was excellent, as judged by the diaries and personal communication. To account for the individual availability of the subjects for pre- and post-training measurements, we allowed some flexibility of the total duration of training. Thus, the duration of training ranged from 14 to 26 days (mean duration: 20.5 days, SD 3.9). Regression analysis revealed no statistically significant influence of training duration on training effects (in particular, on the increases of force and of CCI at the end of the 2 min exercise test).

In the present study, we analyzed if the exercise-induced drop of CCI was influenced by muscular training. Our data demonstrate that after approximately 3 weeks of little finger abduction training, the CCI decreased significantly less during exercise. Compared to the experiments in the untrained condition, we found 30% less CCI decrease during 2 min of isometric exercise after muscular training (Figures 3, 4). At the same time, the relative drop of force remained unchanged. During muscular fatigue, TMS fails to activate the entire spinal MN pool of a target muscle. This failure of activation was demonstrated in our study by the drop of the CCI (which is a measure of the size of the motor evoked potential) during a 2 min exercise. Previous studies demonstrated that it was related to the intensity and the duration of exercise, and varied between subjects (Andersen et al., 2003; Rösler et al., 2009). The differential adaptation of force and CCI during training does not support the idea that there is a straightforward causal link between the drop of CCI and the drop of force during motor fatigue.

We used a repetitive contraction protocol, introduced in a previous study (Rösler et al., 2009). Pilot experiments and the study by Andersen et al. (2003) had demonstrated that sustained work principally leads to similar alterations of TMS induced responses as those observed here. The results obtained in the present study before training were virtually identical to those of our previous study (Rösler et al., 2009). Facilitation maneuvers may additionally confound the fatigue induced changes of MN excitability measured by TMS responsiveness, as facilitatory and inhibitory inputs to MNs at both spinal and supraspinal sites may change unpredictably during a fatiguing exercise. To achieve activation of 100% of MNs by TMS, facilitation of the response by voluntary contraction is necessary (Magistris et al., 1998; Rösler et al., 2002). In the present study, we chose contractions of 20% of MVC to facilitate the TST responses throughout the entire exercise protocol. We have previously discussed the implications of the amount of pre-contraction on the facilitatory maneuver for the TST in fatiguing exercise (Rösler et al., 2009).

The technical implications of the TST have extensively been discussed previously, regarding the comparison to conventional MEPs (Magistris et al., 1998; Rösler et al., 1999; Rösler et al., 2002; Andersen et al., 2003; Humm et al., 2004), and to the use of the CCI (Rösler et al., 2009), and the small systematic error introduced due to peripheral desynchronisation and muscle-nerve ephaptic backfiring (Magistris et al., 1998), all not being critical for the interpretation of our data.

During all experiments, effects of peripheral fatigue on the recordings were observed. Peripheral fatigue leads to characteristic changes of the response configuration (broadening of the CMAP, small amplitude reduction). These changes do not influence the CCI, because they occur in both deflections of the TST recordings. As shown previously during a similar exercise protocol, the ratio between the second and first deflection remained constant, if TMS was replaced by peripheral stimulation at the brachial plexus within the TST protocol (Rösler et al., 2009). The observed changes of CCI with TMS are therefore caused by excitability changes within the central motor pathway, and are not influenced by peripheral effects of fatigue.

Our data shows that changes in MEPs (as measured by the CCI) are associated with the loss of muscular force during exercise. During fatigue, fewer MNs are elicited after TMS, suggesting that MNs become less responsive to synaptic input (Taylor and Gandevia, 2008). The present results demonstrate that this phenomenon is influenced by training. After training, our subjects’ CCI decreased significantly less than before training, indicating that the responsiveness to synaptic input during fatigue improved (Figures 3, 4). Previous studies using TMS in high resistance training analyzed the size of MEPs before and after training, and found unchanged or reduced MEP size, depending on facilitatory maneuver and examined muscle (Carroll et al., 2002; Jensen et al., 2005). Our present results show that the lack of a training effect on MEP size may only be observed if measurements are performed in a rested (i.e., unfatigued) muscle, but that differences may be observed if measurements are done during a fatiguing exercise. Indeed, the present measurements showed a CCI near 100% at the beginning of the exercise, before and after training. Hence, nearly 100% of the target muscle MN pool could be brought to discharge in the rested muscle, and this percentage was not affected by the training. That “central neural drive” increases with strength training has previously inferred by observations of surface EMG recordings, which indicates that greater number of motor units can be recruited voluntarily (Grimby et al., 1981). Increased ability of the CNS to activate motor units after strength is also demonstrated by the twitch interpolation technique. During strength training, the maximal voluntary force increases more than the stimulus evoked force, suggesting that a greater proportion of the muscle can be accessed by the CNS after training (McDonagh et al., 1983; Davies et al., 1985; Narici et al., 1989). It is well conceivable that the effect of our training program on the CCI during fatigue is related to the increased central neural drive as described in the above mentioned studies. A limitation of our study is the lack of a control group who underwent no or “sham” training to rule out that our findings are simply an effect of central learning mechanisms. However, previous studies have not found significant differences in motor cortex excitability between strength and skill training (Leung et al., 2015). Furthermore, our data do not pinpoint the exact underlying physiological mechanisms responsible for the less pronounced decrease of CCI during exercise after training. Changes of different neural structures at spinal and supraspinal sites could be involved (spinal MNs, cortico-spinal neurons, subcortical neurons contacted by corticospinal tract fibers and projecting to the spinal MNs, intracortical inhibitory and excitatory interneurons). Possible underlying mechanisms will be discussed subsequently.

From previous studies, there is evidence that resistance training does not lead to changes in the representational organization of the cortex. Adaptations in the motor cortex of monkeys are not induced by repetitive execution of a simple movement (Plautz et al., 2000). In rats, the training-induced reorganization of the movement representation within the motor cortex remains similar, whether the rats performed movement against a low or a high load (Remple et al., 2001). In humans, Carroll et al., 2002 demonstrated that the MEP size decreased after resistance training (but not after a “sham” training of similar movements without resistance), and that this decrease occurred equally if MEPs were elicited by transcranial magnetic stimulation or transcranial electrical stimulation. As electrical stimulation targets the corticospinal neurons directly on the axon distal to the axon hillock, and magnetic stimulation excites the corticospinal neuron trans-synaptically via intracortical circuits (Edgley, 1997), the results of Carroll et al. point to changes of functional properties at spinal sites rather than structural changes in the motor cortex. Finally, Z’Graggen et al. (2008) did not find an effect of resistance training on the TMS resting motor threshold. Summarized, these studies argue against the explanation of our results by structural cortical adaptations.

On the other hand, it is equally unlikely that afferent input from the periphery could explain our findings. A reflex inhibition of alpha motor neurons by group III and IV muscle afferents was shown to influence firing frequency of motor neurons during fatigue (Bigland-Ritchie et al., 1986; Woods et al., 1987; Garland and McComas, 1990; Duchateau and Hainaut, 1993), and may change with strength training (Aagaard et al., 2000; Kamen and Knight, 2004; Duchateau et al., 2006). Changing afferent negative feedback could thus influence the size of TMS-evoked responses during training. However, previous studies have shown that afferent negative feedback is probably not an important factor for the exercise induced drop of CCI. Rösler et al. induced muscular contractions by stimulating the ulnar nerve by stimulus trains of 20 Hz, simulating the voluntary exercise protocol as closely as possible (Rösler et al., 2009). While this “imposed” exercise led to similar fatigue of muscle force than voluntary exercise, the CCI remained almost constant throughout the 2 min of exercise, excluding an important contribution of muscle afferents to the CCI. Moreover, after voluntary exercise, the recovery of CCI was not delayed by hemostasis induced by putting a cuff around the upper arm; and recovery of conventional MEP amplitudes after sustained post-exercise muscle ischemia was not disturbed (Taylor et al., 2000). Taken together, it seems unlikely that afferent feed-back was involved in the exercise-induced drop of CCI, and thus it seems not probable as well that the observed training effect was caused by alterations of this feed-back mechanism. However, all previously mentioned studies on afferent feed-back mechanisms were not performed after a training period as our present study.

The less pronounced drop in CCI after training could be caused by a training induced reduction of intracortical inhibition during exercise (Rösler et al., 2009). Impaired intracortical inhibition has been a potential explanation for the reduced CCI decrease in patients suffering from multiple sclerosis (Scheidegger et al., 2012). In these patients intracortical inhibition is probably impaired by a cortical maladaptive process due to structural alterations in motor areas. In our healthy subjects after training, intracortical inhibition might be reduced as a transient physiological response to facilitate the motor task. Accordingly, short-latency intracortical inhibition was previously shown to decrease after 3 weeks of strength training in healthy subjects (Goodwill et al., 2012), and silent period duration was also shown to be reduced after 4 weeks of strength training (Kidgell and Pearce, 2010). Intracortical inhibition could be influenced by the level of voluntary effort perceived during the exercise, in healthy subjects (Rösler et al., 2009), and in patients with multiple sclerosis (Scheidegger et al., 2012) and amyotrophic lateral sclerosis (Vucic et al., 2011). Also, a recent systematic review on the effect of resistance training on human cortical excitability concluded that there is evidence favoring the reduction of intracortical inhibition in motor cortex, though pointing at a high degree of inconsistencies in the studies which were evaluated (Colomer-Poveda et al., 2019). In the present study we did not formally assess the level of perceived exertion of our subjects, but most of them reported a greater ease in performing the exercise task after the training.

In conclusion, 3 weeks of regular muscular training significantly influences fatigue induced drop of motor evoked potentials during an isometric exercise. This may be due to a greater responsiveness to synaptic input during fatigue and hence a greater proportion of muscle that can be activated by the CNS after regular muscle training.