B R A I N R E S E A R C H 1 3 2 0 ( 2 0 1 0 ) 4 7 – 5 9

ava i l ab l e a t www.sc i enced i r ec t . com

www.e l sev i e r . com/ loca te /b ra i n res

Research Report

Influence of activity-induced axonal hypoexcitability ontransmission of descending and segmental signals

Alessandro Rossi⁎, Alessia Biasella, Cristiano Scarselli, Pietro Piu, Federica GinanneschiClinical Neurophysiology, Department of Neurological Neurosurgical and Behavioural Sciences, University of Siena,Viale Bracci 1, 53100 Siena, Italy

A R T I C L E I N F O

⁎ Corresponding author. Fax: +39 0577233476.E-mail address: rossiale@unisi.it (A. RossAbbreviations: ADM, abductor digiti minim

compound sensory action potential; EMG, elroot mean square; SD, standard deviation; S

0006-8993/$ – see front matter © 2009 Publisdoi:10.1016/j.brainres.2009.12.023

A B S T R A C T

Article history:Accepted 7 December 2009Available online 22 December 2009

In this experiment, the changes in excitability of motor axons produced after natural activityweremeasured inninehealthy subjects using 1minofmaximal voluntary contractions (MVC)of the abductor digiti minimi (ADM) by studying the relationship between stimulus intensityapplied to theulnarnerve and the size of theADMcompoundmuscle action potential (CMAP).On cessation of the contraction, there was a prominent right-shift of the input–output curve:the intensity required to produce a control CMAP ∼60% of maximum, generated a post-contraction response ∼25% of maximum. Similar changes occurred in the input–outputcurves obtained by recording the ulnar nerve volley evoked by same test stimulus for CMAP.Motor-evoked potential (MEP) and F-waves (and H-reflex in one subject) were recorded fromADMbefore and after 1minofMVC.On cessation of contraction, theMEP input–output curvesexhibited a significant right-shift: the stimulus required to evoke apre-contractionmaximumMEP (∼60%ofmaximumCMAP) generated a post-contraction response∼65%of initial values.OneminuteofMVCproduced similar decreases of F (∼35%)- andH (∼30%)-ADMresponses.Allresponses recovered their control value in 15–20min after the end of contraction. The almostidentical depressive effect produced by 1min of MVC on peripherally and centrally generatedmuscle responses suggests a common conditioning factor. These findings are discussedwithin the context of activity-induced motor axonal hyperpolarizion.

© 2009 Published by Elsevier B.V.

Keywords:Axonal hyperpolarizationCorticospinal volleyMotoneuron excitabilityVoluntary activity

1. Introduction

Activity-induced hypoexcitability occurs in humanmotor andsensory axons when they conduct trains of impulses (Milleret al., 1995; Kiernan et al., 1997; Vagg et al., 1998). Activation ofthe electrogenic Na+-K+ pump is presumably the mechanismresponsible for hyperpolarization, as found in animal studies(BostockandGrafe, 1985; Gordonet al., 1990;Morita et al., 1993).

i).i; CMAP, compoundmuscectromyography; MEP, mE, standard error; TMS, tra

hed by Elsevier B.V.

By a computerized threshold-tracking procedure, Vagg et al.(1998) demonstrated that excitability of motor axons in themedian nerve innervating the thenar eminence significantlydecreased after 1 min of maximal voluntary abduction of thethumb. It was accompanied by appropriate changes in nodaland internodal properties (i.e., increase in rheobase, decreasein strength–duration time constant, and increase in axonalsupernormality), indicating that axonal hypoexcitability was

le action potential; CNAP, compound nerve action potential; CSAP,otor evoked potential; MVC, maximal voluntary contraction; RMS,nscranial magnetic stimulation

48 B R A I N R E S E A R C H 1 3 2 0 ( 2 0 1 0 ) 4 7 – 5 9

actually due to hyperpolarization. Theoretically, hyperpolar-ization in motor axons following their voluntary activationcould be a limiting factor for impulse conduction if it is largeenough to overcome the safetymargin for impulse conduction(Bostock and Grafe, 1985). Activity-induced axonal hyperpolar-ization could also involve its proximal segment and, therefore,the trigger zone for action potential, i.e., hillock/first node ofRanvier. If so, thesynaptic current transferred to the somafromthe set of activated synapses by the same descending orsegmental volley could fail to initiate action potential in all thetarget motoneurons activated before contraction.

Effects of prolonged activity on axonal membrane proper-ties have been more or less ignored in human experiments.For example, after sustained voluntary contraction, theamplitude of both motor-evoked potential (MEP) produced bytranscranial magnetic stimulation (TMS) and compoundmuscle action potential (CMAP) are transiently depressed(Gandevia et al., 1999; Kalmar and Cafarelli, 2004). Decline inMEP amplitude observed during or after voluntary contractionwould appear, in the absence of peripheral data, to beevidence of central failure (Brasil-Neto et al., 1993; McKayet al., 1995; Zanette et al., 1995; Gandevia et al., 1996; Liepertet al., 1996; Samii et al., 1996; Taylor et al., 1996; Samii et al.,1997; Pitcher and Miles, 2002). In fact, since peripheral failuremay actually confound the interpretation of MEP, the impor-tance of analyzing CMAP before attributing changes incortically evoked potentials to cortical mechanisms has beenstressed (Kalmar and Cafarelli, 2004). Although the extent ofactivity-induced CMAP failure may be dependent upon theoverall demands of the contraction protocol, the mechanismresponsible for it has not yet been elucidated. Reduction in

Fig. 1 – Time course of average force (A) and electromyographic aof the fifth finger in eight subjects. Force is expressed in Newtonare SE of the mean. Sketch of experimental paradigm is also sho

sarcolemmal excitability, particularly in fast-twitch musclefibers (Fowles et al., 2002), has been considered a contributoryfactor (see also discussion). However, as already mentioned,evidence also exists of prominent activity-dependent hyper-polarization in motor axons lasting many minutes after theend of contraction (Vagg et al., 1998).

Thus, the primary purpose of this article is to place ourfindings of CMAP, MEP, and F-wave changes after 1 min ofmaximal voluntary contraction (MVC) of the abuctor digitiminimi muscle (ADM), within the context of activity-inducedmotor axonal hyperpolarizion. It was based on the observationthat magnitude and recovery time of activity-induced axonalhypoexcitability and those of post-contraction depression ofevoked descending (i.e., MEP) and segmental (i.e., F- and H-waves) responses were quantitatively almost identical.

2. Results

2.1. Excitability of ulnar nerve motor axons innervatingthe ADM before and after 1 min MVC

In all subjects, MVCmaintained for 1min reduced output forcefrom 9.12Newton (N)±0.57 standard error (SE) to 6.67N±0.51 SEand electromyographic activity from 0.70 mV±0.02 SE to 0.45mV±0.03 SE (data refer to average of first and last 5 min,respectively) (Fig. 1). The time courses of force (N) and EMG root-mean-square (RMS) (mV) showed similar behavior of the trendcomponentof those series, asmeasuredby theHollander test onlinear regression angular coefficients (−0.028534 and −0.004508,respectively): h=1.5421, P=0.9371.

ctivity (EMG) (B) during 1min of maximal isometric abductions (N) and EMG, as root-mean-square (RMS). Vertical barswn in the figure.

49B R A I N R E S E A R C H 1 3 2 0 ( 2 0 1 0 ) 4 7 – 5 9

Fig. 2 shows the input–output curves elicited by stimula-tion at elbow and wrist, respectively before and after 1 min ofMVC. This relationship had a sigmoid shape with a thresholdreflecting activation of the most excitable axons, a slopeindicating input–output efficiency (gain) and a top valuereflecting the total number of motor units (virtually 100%)recruited by maximum stimulation. There was no significantdifference between input–output relationships obtained bystimulation applied to wrist and elbow either before or after1 min MVC: pre-MVC wrist–elbow: rank sum=674, P=0.79;post-MVC wrist–elbow: rank sum=2601, P=0.60. Likewise, nosignificant difference was observed in the variability of CMAP(estimated by standard deviations) elicited by elbow and wriststimulation: Levene's test: f=3.85, P=0.06. After MVC, a similarincrease in the peak of variability occurred for both sites ofstimulation: pre-wrist=1.64, post-wrist=2.28; pre-elbow=1.56,post-elbow=2.44. The similarity of the elbow and wrist curvesensured stability of stimulation and eliminated the possibilitythat curve changes reflected contraction-induced shift of thestimulating electrodes at the wrist, since they were placedclose to the contactingmuscle. The similarity of the elbow andwrist curves made analytical description of both input–outputrelationships superfluous, so that all the data described belowrefer to wrist stimulation.

After 1 min of MVC, there was a significant change in theparameters of the Boltzmann sigmoidal function fitted to theexperimental data (Fig. 2B). Post-contraction bottom value ofthe curvewasnot significantly differentwith respect to control(t=5.72, P=0.08). Gain of the input–output relationship de-creased significantlywith respect to control: slope=56.31±2.58SE and 40.08±0.65 SE in pre- and post-contraction, respectively

Fig. 2 – Averaged input–output curves fitted to the Boltzmann svoluntary contraction (MVC) of the abductor digiti minimi (ADM)relation to stimulus intensities applied to the ulnar nerve at elbowthe ulnar nerve at wrist. (C) Superimposed scatter points from albefore (■) and after (□) 1 min of maximal voluntary contraction. Cpercent), and the stimulus intensity was normalized to unit and eof mean.

(h=23.40, P<0.0001). Note that, since the slope value of theBoltzmann function is computed point by point as the firstderivate of the function, we used the angular coefficient of thetangent line at the inflectional point (i.e., the pointcorresponding to 50% of the abscissa scale, also called V50) asthe most representative slope value.

Top value of the post-contraction input–output curveswas the same as in control, although at higher intensity(mean=1.25 times above pre-contraction intensity requiredto obtain maximal response). Indeed, the intensity required(2.90×Th) to obtain a maximal pre-contraction CMAP evokeda post-contraction CMAP of 72.4%±11.38% at maximum(rank sum=73, P=0.0047). Average decrease in CMAP curvevalue (from bottom to top) was 17.19%±5.03 standarddeviation (SD), and average intensity required to restoreCMAP to its control of the curve was 31.10%±4.80 SD abovepre-contraction intensity (Figs. 5A, B). The input–outputcurves after MVC did not change significantly when thesize of CMAP was measured in terms of area rather thanamplitude (rank sum=1725, P=0.097). Also, duration ofnegative phase as well as the general shape of potentialsdid not changed, suggesting that the decrease in amplitudewas not due to slowing of conduction of the actionpotentials across the fatigued muscle fibers. Latency ofpre- and post-contraction CMAP did not show any signifi-cant change: wrist stimulation, pre-contraction =2.92 ms(±0.92 SD), post-contraction=2.74 ms (±0.71 SD) (t=1.52,P=0.34); elbow stimulation, pre-contraction =7.25 ms (±1.64SD), post-contraction =6.83 ms (±1.86 SD) (t=2.74, P=0.96).

Fig. 3A shows the input–output curves elicited by stimula-tion of ulnar nerve at wrist and recorded from the same nerve

igmoidal function before (■) and after (□) 1 min of maximal. (A) Size of the compound muscle action potential (CMAP) in. (B) Size of CMAP in relation to stimulus intensities applied tol the subjects obtained by stimulation of ulnar nerve at wrist,MAPs were normalized to their maximum (expressed inxpressed in multiples of threshold (× Th). Vertical bars are SE

Fig. 3 – Averaged size compoundnerve action potential (CNAP, diagramA), example ofwaveforms of CNAP (A1) and compoundmuscle action potential (CMAP, diagram B) evoked by the same stimulus intensities applied at ulnar nerve at wrist. (A) CNAPinput–output curves fitted to the Boltzmann sigmoidal function before (■) and after (□) 1min ofmaximal voluntary contraction ofthe abductor digiti minimi. Recording from ulnar nerve at elbow and stimulation from the same nerve at wrist. (B) CMAPinput–output curves fitted to the Boltzmann sigmoidal function before (■) and after (□) 1min ofmaximal voluntary contraction ofthe abductor digiti minimi. Gray segments in abscissa indicate the increase in intensity required to obtain maximum CMAPafter MVCwith respect to control. Note that intensity requested to obtainmaximumCMAP after contraction (open squares in B)evoked a supra-maximum CNAP (open squares in A). Vertical bars are SE of mean.

Fig. 4 – Averaged motor-evoked potentials (MEP)input–output curves fitted to the Boltzmann sigmoidalfunction before (■) and after (□) 1 min of maximal voluntarycontraction of the abductor digiti minimi muscle (diagramA).Upper insert on the right, are superimposed MEP responsesfrom all the subjects before (■) and after (□) contraction. Dataare fitted to the Boltzmann equation, and each point is themean of 3 measurements (diagram B). Vertical bars are SE ofthe mean.

50 B R A I N R E S E A R C H 1 3 2 0 ( 2 0 1 0 ) 4 7 – 5 9

at elbow, before and after 1min ofMVC. A samplewaveformofcompound nerve action potential (CNAP) is given in Fig. 3A1.Fig. 3B illustrates the corresponding curves from ADM elicitedby the same nerve stimulation. Nerve-to-nerve input–outputrelationship had a sigmoid shape. Pre- and post-contractionbottom values showed no significant difference (t=2.42,P=0.168). Post-contraction slope of the curve (52.78%±3.68SE) decreased significantly with respect to its control value(47.85%±2.31 SE) (h=23.40, P<0.0001). Average decrease in thenerve curve (from bottom to top) was 12%. The stimulusintensity required to evoke a pre-contraction nerve potentialamplitude corresponding to maximum CMAP (100%) evoked apost-contraction ulnar nerve potential 27% lower (ranksum=73, P=0.0047) and a CMAP 29% lower than their pre-contraction values (rank sum=56, P=0.0073). The currentrequired to obtain a post-contraction CMAP 100% of controlevoked a nerve potential of∼130%. This apparent paradoxwaspresumably due to contribution of orthodromic volley fromsensory fibers co-activated by our stimulation (see alsodiscussion). In order to verify the effect of MVC on sensoryaxons, we analyzed the compound sensory action potential(CSAP) of the ulnar nerve at wrist after stimulation of digitalnerves of the little finger. It did not change in relation tocontraction: amplitude: 5.1 μV±2.33 SD and 4.68 μV±2.36 SD(P=0.156, t=1.619); latency 2.36 ms±0.21 SD and 2.14 ms±0.35SD (t=1.23, P=0.21) in pre- and post-contraction, respectively.

2.2. Excitability of evoked corticospinal pathway to ADM,before and after 1 min of MVC

In addition to peripheral changes, 1 min of MVC reduced theMEP input–output curve (Fig. 4A). There was a significantdifference between pre- and post-contraction MEP input–

output curves (Wilcoxon rank sum test=749, P<0.0001). Slopevalue of MEP curves was 103.30±4.47 SE and 66.11±2.98 SE forpre- and post-contraction, respectively (h=764.62, P<0.0001).

Fig. 5 – (A) percent decrease of post-contraction CMAP andMEP input–output curves. It gives the average decrease of CMAP andMEP (from minimum and maximum) caused by 1 min MVC. (B) percent increase in stimulus intensity required to recoverpost-contraction CMAP and MEP input–output curves to their control values. (C) Relationship between motor-evoked potential(MEP) by transcranial magnetic stimulation and compound muscle action potentials (CMAP) by wrist stimulation. Curve “a":relationship between the quantiles of pre-contraction MEP and CMAP distribution. Curve “b": relationship between thequantiles of post-contraction MEP and CMAP distribution. Quantiles (xq) have been computed every 10% of the distribution(q=0.1, 0.2, 0.3… 1). Curves “a” and “b” are described by exponential and first-order polynomial equations (straight line),respectively. First derivative of curve “a” confirmed that it holds a straight line up to the three-fourths of the distribution. Curve“c”: diagram bisector. Numbers on the right are angular coefficient values of curves; these values are reported with the SE.

51B R A I N R E S E A R C H 1 3 2 0 ( 2 0 1 0 ) 4 7 – 5 9

The threshold value after MVC was not significantly differentfrom control value (Δ=1.06±1.03 SD; t=2.97, P=0.456), whereasthe top value of the curve (i.e., maximum MEP) exhibited alarge drop: 62.90%±18.06 SD of control value (rank sum=100,P=0.00015). Average decrease in input–output MEP was 15.34±4.58 SD, and average intensity of TMS required to restore MEP

Fig. 6 – Size (A), persistence (B), and latency (C) of F-waves beforecontraction of the abductor digiti minimi muscle. Vertical bars ar

size to its control value from bottom to top was 36.00%±3.80SD above pre-contraction intensity. Figs. 5A and B comparesthe effect of 1 min of MVC on MEP and CMAP sizes. Fig. 5Cillustrates the correlation betweenMEP and CMAP curves. Thediagram shows the quantile function (i.e., the inverse of thecumulative distribution function: 2.xq=F-1(q)= inf {X | F(X)≥q,

(pre-MVC) and after (post-MVC) 1 min of maximal voluntarye SE of mean.

52 B R A I N R E S E A R C H 1 3 2 0 ( 2 0 1 0 ) 4 7 – 5 9

∀q∈(0,1)}) of pre- and post-contraction CMAP and post-contraction MEP. Quantiles were computed at every 10% ofthe distributions. This method was used to extract any linearcomponent of the relationship. Post-MEP/post-CMAP (curve B)was a straight line almost up to three-fourths of thedistribution, while pre-MEP/pre-CMAP showed nonlinearbehavior (exponential). The angular coefficient (1.45±0.01 SE)of the linear regression fitted to the curve B was very closeto one (i.e., close to the diagram bisector), but for curve A,the angular coefficient was far from the diagram bisector(1.95±0.03 SE).

2.3. Excitability of spinal ADM motoneurons, before andafter 1 min of MVC

One minute of MVC reduced the response of spinalmotoneurons innervating ADM, evoked antidromically bysupramaximal stimuli (F-waves) or orthodromically by weakstimuli just above motor threshold (H-reflex). Average areasof F-waves (Fig. 6A) were 0.83 (mV/ms)±0.06 SE and 0.51 (mV/ms)±0.04 SE (rank sum=6046, P=0.0001) pre- and post-

Fig. 7 – Diagram from a single subject, showing MEP (A), H-reflevoluntary contraction of the abductor digiti minimi muscle, respan example of the above responses.

contraction, respectively. F-wave persistence (Fig. 6B) (i.e.,number of responses evoked by 20 stimuli) decreased from91.6%±1.4 SE to 81.6%±5.3 SE (i.e., by 11%) after 1 min of MVC(χ2=0.8130, P=0.9368). F-wave latency (Fig. 6C) did notchange after contraction: 28.19 ms±0.24 SE and 27.89 ms±0.25 SE (rank sum=7970, P=0.2927). In one exceptionalsubject, H-reflex from the ADM was obtained by adequatestimulation of the ulnar nerve at wrist (Fig. 7A). After 1 minof MVC, H-reflex amplitude (peak to peak) significantlydecreased: 1.92 mV±0.15 SE and 1.34 mV±0.09 SE (i.e., by∼30%) (W=47, P=0.0137). Figs. 7B and C also shows the effectof 1 min of MVC on MEP (∼42% decrease) and F-waves (∼41%decrease) in the same subject. Fig. 8 shows the time courseof post-contraction recovery of CMAP, MEP, H-reflex, andF-waves to their control values. To avoid overlap ofsymbols, the time course is represented by four-orderregression lines fitted to experimental points. All responsesrecovered within 15–20 min from the end of contraction.Finally, in order to made direct comparison of data, theeffect of MVC on CMAP, MEP, and F-wave size is summa-rized in Fig. 9.

x (B), and F-wave (C) size before and after 1 min of maximalectively. Vertical bars are SE of the mean. Upper insert gives

Fig. 8 – Time course of recovery of the MEP, CNAP, CMAP, F-wave and H-reflex from the end of 1 min of maximal voluntarycontraction of the abductor digiti minimimuscle. Auto-normalized and averaged responses are expressed by spline curves: 272points were calculated with the x-values ranging from 2 to 22. Gray area in diagram identifies the range of elapsed time. Insertson the right showMEP, CMAP and CNAP elapsed time in one representative subject. Each point is the average of 3–5 recordingsand one SD of mean.

53B R A I N R E S E A R C H 1 3 2 0 ( 2 0 1 0 ) 4 7 – 5 9

3. Discussion

The present study has confirmed the excitability changes thatoccur in motor axons following a voluntary contraction (Vagget al., 1998). One minute of MVC resulted in a prominent rightshift of ulnar nerve input–output curves (Figs. 2, 3), due toaxonal hypoexcitability. Indeed, (1) there was a significantdecrease in slope (∼30%); (2) the same stimulus intensityapplied to the ulnar nerve evoked significantly lower muscleresponses after MVC (∼20% average decrease in CMAP); (3)post-contraction depression lasted many minutes, waning in∼20 min; (4) there was a prominent increase in the currentrequired (∼30% average increase) to produce the control targetpotential; (5) post-contraction input–output curves reachedthe same plateau value as control curves, indicating thatvirtually all ADM motor axons could be still activated.

Post-contraction axonal hypoexcitability was strictlyparalleled by depression of descending (MEP) and segmental(F- and H-responses) evoked volleys. Our interpretation isthat activity-induced motor fibers hypoecitability, likely dueto activity-induced axonal hyerpolarization, involved the

trigger zone for action potentials, leading increase in spinalmotoneuron firing threshold. The rationale for the hypoth-esis that depression of the evoked responses was, at least inpart, indirectly due to axonal hypoexcitability will bediscussed below.

Asmentionedabove, in addition toCMAP, 1minofmaximalisometric abduction of digiti minimi also resulted in depres-sion of the MEP input–output curve (Fig. 4): (1) on cessation ofcontraction, the slope of the curve decreased by ∼40%; (2) theaverage decrease in the MEP curve was ∼15% and a ∼35%increase in TMS was required to recover control value; (3) theTMS required to generate a maximum pre-contraction MEP,evoked a response ∼65% of maximum after MVC; (4) pre-contraction value was recovered ∼20 min from the end ofcontraction. Post-contraction decrease was also observed inresponses evoked by motoneurons innervating the ADM. Oncessation of contraction, the average size of F-waveswas∼30%lower than control, recovering in ∼20 min (Fig. 6). Finally, thesize of the post-contraction H-reflex of ADM obtained in onesubject (Fig. 7) was 0.6 mV lower than control (∼30% decrease),with a similar recovery time as observed with CMAP, MEP, andF-waves.

Fig. 9 – The figure summarizes CMAP (A, D), MEP (B, E) and F-wave (C, F) data. In upper diagrams, post-contraction responsesare expressed in percent of their control value. In lower diagrams, CMAP control size (open column in D) corresponded tothe size of maximum MEP, whereas MEP (open column in E) and F-wave (open column in F) as percentage of maximum CMAPsize. Vertical bars represent one SD of mean.

54 B R A I N R E S E A R C H 1 3 2 0 ( 2 0 1 0 ) 4 7 – 5 9

3.1. Putative factors responsible for activity-induceddepression of evoked responses

3.1.1. Muscle fibersIt is documented that natural activity may be associated withelevation of extracellular K+ concentrations, which couldreduce the amplitude of the muscle response (Overgaard etal., 1999; Pedersen et al., 2003, Sejersted and Sjogaard, 2000).The influence of elevated [K+] on M-waves may occurthrough a decrease in the number of excitable muscle fibersdue to a decrease in sarcolemmal excitability as a result ofdepolarization-induced slow inactivation of sodium channels(Ruff et al., 1988, Sjogaard, 1990; Lindinger and Sjogaard,1991; McKenna, 1992; Cairns et al., 1997). However, studies inwhich venous potassium and M-wave have been measured

simultaneously during voluntary exercise failed to demon-strate a relationship between these two variables (West et al.,1996; Unsworth et al., 1998; Jammes et al., 2005, Shushakov etal., 2007). In the present study, by monitoring the volleygenerated in the ulnar nerve (i.e., the CNAP), by the samestimulus used to evoke CMAP in ADM, parallel depression ofthese responses was observed after 1 min of MVC. Thissuggested that contraction-induced CMAP depression wascontingent on fewer motor axons being recruited by thestimulus with respect to control. Indeed, if CMAP depressionwas caused by a decrease in the number of excitable musclefibers, then nerve volley would be the same before and aftercontraction. We therefore conclude that the mechanismresponsible for post-contraction depression acts proximallywith respect the muscle membrane.

55B R A I N R E S E A R C H 1 3 2 0 ( 2 0 1 0 ) 4 7 – 5 9

3.1.2. Spinal motoneuron firing thresholdUsing a computerized threshold-tracking procedure, Vagget al. (1998) observed that 1 min of maximal voluntaryabduction of the human thumb resulted in a prominentincrease in the current required (∼30% using 0.1-ms teststimulus) to produce a CMAP ∼60% of maximum that wanedover 15 min. The increase in current required to produce thetarget potential was accompanied by an increase in rheobase,a decrease in strength–duration time constant, and anincrease in axonal supernormality, indicating that thehypoexcitability was due to axonal hyperpolarization.

The possibility that motor axonal hyperpolarization, inaddition to CMAP depression, could be responsible for long-lasting, post-contraction depression of cortical and spinal-evoked responses need to be considered. Theoretically, axonalhyperpolarization could reduce nerve, cortical, and spinalresponses by (1) producing conduction blocks, which mayoccur if hyperpolarization is large enough to overcome thesafety margin for impulse conduction; (2) increasing the firingthreshold of spinal motoneurons, as a possible consequenceof axon hillock/first node of Ranvier hyperpolarization. Thefirst hypothesis is unlikely, since motor axons did not showsigns of conduction block after MVC. Indeed, (a) the latenciesof CMAP (fromwrist or elbow to ADM) and CNAP (fromwrist toelbow) did not change after contraction. In fact, there was nodifference between parameters of the input–output curveobtained by electrical stimulation applied at wrist or elbow; (b)the plateau phase of CMAP input–output curves reached thesame value as in control condition, indicating that conductionwas preserved in all axons; finally (c) no significant changewas observed in F-wave latency with respect to control afterMCV. The second hypothesis was that axonal hyperpolariza-tion also caused an increase in the firing threshold of spinalmotoneurons. A long-standing hypothesis is that actionpotentials initiate in the axon hillock. However, the actualsite of initiation is not completely resolved (Adams, 1992;Johnston et al., 1996). Coombs et al. (1957) first noted thataction potentials might initiate in the axon, at the first node ofRanvier. More recently, Costa and Johnston (1996) confirmedthat orthodromic action potentials may actually initiate in theaxon beyond the hillock region of the neuron. Since a strictmembrane similarities among axonal nodes and even be-tween the initial segment and the nodes exists (Palay et al.,1968; Peters et al., 1968), it would be reasonable to assume thatthe axon hillock/first node of Ranvier and peripheral motoraxon properties will change in a similar manner. Althoughactivity-dependent changes in the axon hillock have neverbeen studied directly in humans, we suggest that post-contraction axonal hyerpolarization also involved initialsegment/first node of Ranvier. If so, the synaptic currenttransferred to the soma by the set of activated synapsesthrough descending or segmental volleys may not be suffi-cient to initiate action potentials in all the axons activated bythe same volley in control condition. The following findingssupport the possibility of increased firing threshold of spinalmotoneurons as a consequence of axonal hyperpolarization:(1) the increase in stimulus intensity required to compensatefor post-contraction depression of MEP and CMAP was of thesame order; (2) ADM responses resulting from descending(MEP) or segmental (F- and H-responses) volleys underwent

quantitatively similar post-contraction depression and withinthe same order as CMAP; (3) the relationship between post-contraction MEP and CMAP (quantiles–quantiles plot) showedlinearity, not present before contraction; finally, (4) oncessation of contraction, MEP and CMAP (and F- and H-responses) exhibited similar, slow recovery time lasting15–20 min; a time fully compatible with that required torestore axonal excitability after contraction (Applegate andBurke, 1989; Bostock and Bergmans, 1994; Kiernan et al., 1997).Therefore, these data are coherent with the conclusion thatthe decreases in motoneuron excitability (reflected by de-crease of F-wave persistence and amplitude and H-reflexamplitude) are a result of the activity-induced changes inmotoneuron membrane properties, the same change alsoresponsible for decreasing axonal excitability.

Changes inmotoneuron firing threshold in human subjectsafter periods of prolonged activation have not been previouslyreported, other than by using a stimulation protocol of 90 s of18-Hz supramaximal electrical stimulation of peripheral nerve(Butler andThomas, 2003). In this study, F-wave analysis of thethenar motoneuron pools in paralyzed patients by chroniccervical spinal cord injury and healthy subjects revealedreduction in excitability of spinal motoneurons as a conse-quence of activity-dependent changes in membrane proper-ties involving axon hillock and peripheral motor axons. It is ofinterest that the stimulation rate used in that study (18 Hz)about corresponds to the average discharge rate of motoraxons over a 1 min of maximal voluntary contraction ofintrinsic hand muscles (Bigland-Ritchie et al., 1983, 1986;Gandevia et al., 1990; Thomas, 1997). In the present study,1minofMVCproduced similar depressive effects on responsesregardless their size. Indeed, MEP (∼60 of maximum CMAP)and F-wave (∼ 1.5%maximumCMAP) underwent similar post-contraction depression (∼35% of their control value). Becauseof nonlinearity in the output of themotoneuron pool (Hultbornet al., 1996), small and large amplitude responses could exhibitdifferent sensitivity to the same conditioning event. However,F-wave does not reflect re-excitation of the motor unitsaccording Henneman's size principle, while evidence existsthat recurrent discharges of antidromically activated moto-neurons occur irrespective of their peripheral excitability orconduction characteristics (Kimura et al., 1984). It could beplausible that MEP and F-wave (and H-reflex) occurred in acommon portion of the motor pool.

In summary, the present study shows that 1 min of MVCof ADM resulted in prominent ulnar nerve axonal hypoexcit-ability, likely due to hyperpolarization that occurs whenaxons conduct trains of impulses. Axonal hypoexcitabilitydid not produce critical impairment of the safety margin forimpulse conduction, and therefore, post-contraction depres-sion of CMAP, MEP, F-waves, and H-reflex was not caused byaxonal conduction failure. Our proposal is that activity-induced axonal hyperpolarization may involve the triggerzone for action potentials. It could lead to increasedmotoneuron firing threshold, which would reduce thenumber of motoneurons activated by a given descending orsegmental volley. This should not be interpreted against thenotion of fatigue-induced excitability changes at spinal andsupraspinal level. Indeed, there is experimental evidence ofreduction in spinal motoneuron excitability (Butler et al.,

56 B R A I N R E S E A R C H 1 3 2 0 ( 2 0 1 0 ) 4 7 – 5 9

2003; Gandevia et al., 1999), as well as in cortical excitability(Brasil-Neto et al., 1994; McKay et al., 1995; Samii et al., 1996)in relation to fatiguing contractions. Rather, we suggest thatactivity-induced motor axonal excitability changes could be acofactor in limiting activity during fatiguing contractions,although its role may be quantitatively different for differentmuscles. Indeed, activity-dependent hyperpolarization islarger in hand than in limb muscles (Kuwabara et al., 2002).It could be also contingent on the relative contribution ofmotor unit recruitment and rate coding to force productionin each muscle, i.e. the level of motoneurone firing rate couldbe an important factor in determining the magnitude of theactivity-dependent hyperpolarization (Kiernan et al., 2004).During strong contractions, the lower firing rate of moto-neurones innervating large limb muscles with respect tohand muscles (Gelli et al., 2007) could also explain theweaker peripheral changes observed in the former than inthe latter (Kalmar and Cafarelli, 1999; Biro et al., 2007).

4. Experimental procedures

The subjects were nine healthy adult subjects (4 females and 5males) with an age range of 30–54 years; all gave informedconsent to the experimental procedures, which had theapproval of the Committee on Experimental ProceduresInvolving Human Subjects, University of Siena. They sat in areclining armchair with their right arm disposed in a horizon-tal plane in a neutral position. The hand and first four fingerswere secured to a rigid hand piece. The fifth finger was placedat about 10° horizontal abduction and secured to a strain gaugesensor to measure its isometric abduction force. Special carewas taken to prevent anywrist and finger joint changes duringADM contractions. The skin under stimulating and recordingelectrodes was careful prepared with abrasive paste. Handand forearm skin temperature was maintained above 32 °C byinfrared lamp.

Subjects performed maximal isometric voluntary abduc-tion of the little finger for 1 min against resistance. Maximalcontractions were used to ensure that all relevant motoneur-ons innervating the ADMwere active. Ongoing target force andelectromyographic activity/signal were displayed on an oscil-loscope in front of the subject. The EMG signals wereamplified, filtered (10–250 Hz), and stored in digital form.The data were collected at a sampling rate of 100 Hz for themechanical tracings and at 2048 Hz for the EMG signals (DelSanto et al., 2007).

The relationship betweenstimulus intensity andamplitudeof CMAP was used to assess the excitability of ulnar nervemotor axons innervating the ADM before and after 1 min ofMVC. The nerve was stimulated at the wrist and elbow.Recording electrodes were 10 mm in outer diameter Ag/AgClsurface electrodes, placed, according to the belly–tendontechnique, the negative one on the motor point of ADMmuscle, and the reference one, 2 cmapart, on the distal tendonof the samemuscle (medial aspect of the first interphalangealjoint).

The stimulating electrode was initially placed over thewrist and its position adjusted until the site with the lowestthreshold for eliciting a CMAP of 0.1 mV was established.

Electrical stimuli (rectangular pulses of 0.5-ms duration,delivered by a constant current stimulator, once per second)were applied in a fully randomized sequence, with intensityranging from threshold (CMAP of 0.1 mV, baseline–negativepeak) to that required to obtain maximal CMAP beforecontraction (i.e., control condition). After contraction, supra-maximal stimuli (i.e., above the intensity required to obtainmaximal CMAP in the control) were added to the sequence.

In order to a verify the post- exercise changes in excitabilityof the ulnar axons, the relationship between stimulusintensity and size of CNAP was studied. The nerve wasstimulated at the wrist, and the evoked volley was recordedat the ulnar sulcus in the elbow using bipolar surfaceelectrodes taped to the skin 3 cm apart. The stimulatingelectrode was initially placed over the wrist and its positionwas adjusted until the site with the lowest threshold foreliciting a CNAP of 50 μV was established.

In five of the nine subjects, the effect of MVC on cutaneousafferents in the digital nerves of the little finger was alsomeasured. The digital nerves were stimulated using ringelectrodes around the proximal phalanx of the little finger,and the evoked CSAPwas recorded from the ulnar nerve at thewrist using bipolar surface electrodes taped to the skin 3 cmapart. The stimulus was 0.1 ms in duration, one per second. Inall cases, stimulus intensity was expressed in multiples of thepre-contraction threshold (× Th).

MEPs in the ADM were elicited by TMS over the contralat-eral motor cortex. A Magstim 200 stimulator (Magstim, Whit-land, Dyfed, UK) and a figure-of-eight coil (outer diameter8 cm) were used to deliver magnetic stimuli. The stimulatingcoil was positioned over the subject's cortex, orientated at anangle of 45° to the midline and tangential to the scalp, so thatthe induced current flow was in a posterior–anterior directionalong themotor strip (Rothwell, 1997; Ziemann et al., 1998). Todetermine the optimal site of stimulation, the intensity wasfixed at 80% of maximal stimulator output, and the coil wasmoved until the site eliciting the largest MEP in the ADM waslocated. This site was defined as the “hot spot” andmarked onsubject's hand with a felt pen; this served as visual referenceagainst which the coil was positioned and maintained by theexperimenter. All further stimuli were delivered with the coilsecured in this site by clamps.

A headrest placed behind the subject's occipital regionensured that his head position was constant during theexperimental sessions. Individual resting motor thresholdwas determined for each subject by detecting the minimumintensity at which four out of eight consecutive stimuli yieldeda response of at least 50 μV (peak-to-peak amplitude). TMSwas applied with the same paradigm as for ulnar nervestimulation except for the inter-stimulus interval, which wasset at 5 s. As for peripheral stimulation, TMS intensity wasexpressed as multiples of pre-contraction threshold. Input–output curves, expressing the relationship between stimulusintensity and MEP or CMAP amplitudes, were fitted to theBoltzmann sigmoidal function by the Levenberg-Marquardnonlinear least mean squares algorithm (Ginanneschi et al.,2005, 2007b). Each experimental session for the same subjectwas separated by an interval of at least 2 days.

In four subjects, TMS and peripheral electrical stimuli weredelivered alternately. In these cases, the intensity of electrical

57B R A I N R E S E A R C H 1 3 2 0 ( 2 0 1 0 ) 4 7 – 5 9

stimulation was adjusted so as to obtain a CMAP of the samesize as MEP.

Responses of spinal motoneurons innervating ADM weretestedbyanalysis of F-waves inall subjects, andH-reflex inoneexceptional subject, by conventional technique (Ginanneschiet al., 2007a). Thirty F-waves and ten H-responses wererecorded before and after 1 min of MVC. Electrical stimulationapplied to the ulnar nerve at wrist was fixed above maximalADM direct motor response for F-response and just abovedirect motor response for H-reflex. Because of the polyphasicnatureof F-wave, its sizewasmeasured in termsof area. Size ofH-reflex, as well as all the other recorded responses, wasmeasured as peak to peak. Latencies of CMAP, CNAP, F-wave,and CSAP were measured at the onset. Stimuli were deliveredat random intervals, 1- to 5-s interval for F-wave (0.5-msduration) and 10–15 s (1-ms duration) for H-reflex. Conven-tionally, stability of electrical stimulation to group I afferentfibers responsible for the H-reflex is checked using a smalldirectmotor responseas reference. This couldnot be applied inour case, since excitability of motor axons was expected tochange (see results). Hence H-reflex data were only accepted ifthe reflex recovered control size during the post-contractionperiod. Persistence of the F-wave (i.e., number of responseswith respect to number of stimuli applied)was estimated every20 stimuli.

In order to explore the time course of post-contractionrecovery of responses (CMAP, MEP, F- and H-responses), fiveresponses evoked by constant stimulation (TMS and electricalstimulation of the ulnar nerve) were evoked approximatelyevery 2 min from the end of contraction until responsesrecovered their control value. To verify the reproducibility ofthe results, experiments were replicated two or three times ineach subject for a total of 22 sessions.

5. Statistical analysis

Originally the amplitude data (output) were recorded at levelsof delivered intensities (input), which were different from onesubject to another. In order to draw an averaged input–outputcurve through the observed scatter points, we needed to referto common input values. Afterwards, the amplitudes of eachsubject could be reproduced in the same range of intensitieson the basis of his/her own estimated Boltzmann equations(bottom, top, V50, slope) parameters. In other words, since theestimated amplitudes spanned the same input space, wecould average amplitudes of the different subjects point bypoint, in that common interval. The t-test and ANOVAparametric test were used to compare the means of seriesafter assessment of homoscedasticity (equal variance of thedata sets to be compared by Levene's test). Indeed, compar-ison of two (or more) means is reliable when the data comefrom sets having equal variance and normal distribution.Otherwise, a nonparametric test should be used. We adoptedthe Wilcoxon rank sum test (or Mann-Whitney U-test) forequal medians, a two-sided test of the hypothesis that thedistributions of two independent samples, in generic vectors xand y, have equal medians. Small P-values cast doubt on thevalidity of the null hypothesis. The two datasets are assumedto come from continuous distributions that are identical,

except possibly for a location shift, but are otherwise arbitrary.Since normality was rejected (significance α=0.05), a nonpara-metric test was used to assess the significance of thedifference between stimulus intensities applied to wrist andelbow and CMAP amplitudes evoked in ADM before and afterMVC. Hollander's test was used to compare angular coeffi-cients of linear regressions. For the parameters reported, itwas also given either the SD, when interest was focused in thespread of the data or the SE, when interest was focused in theprecision of the estimate.

Acknowledgments

This study was founded by a grant from the Italian Ministry ofHealth. Dr. Pietro Piu attends the Doctorate School in AppliedNeurological Sciences.We are grateful to Dr. FedericaDominicifor assistance with the experiments.

R E F E R E N C E S

Adams, P.R., 1992. The platonic neuron gets the hots. Curr. Biol. 2,625–627.

Applegate, C., Burke, D., 1989. Changes in excitability of humancutaneous afferents following prolonged high-frequencystimulation. Brain 112, 147–164.

Bigland-Ritchie, B.R., Johansson, R.S., Lippold, O.C., Smith, S.,Woods, J.J., 1983. Changes in motoneurone firing rates duringsustained maximal voluntary contractions. J. Physiol. 340,335–346.

Bigland-Ritchie, B.R., Dawson, N.J., Johansson, R.S., Lippold, O.C.,1986. Reflex origin for the slowing of motoneurone firing ratesin fatigue of human voluntary contractions. J. Physiol. 379,451–459.

Biro, A., Griffin, L., Cafarelli, 2007. Reflex gain of muscle spindlepathways during fatigue. Exp. Brain Res. 177, 157–166.

Bostock, H., Bergmans, J., 1994. Post-tetanic excitability changesand ectopic discharges in a human motor axon. Brain 117,913–928.

Bostock, H., Grafe, P., 1985. Activity-dependent excitabilitychanges in normal and demyelinated rat spinal root axons.J. Physiol. 365, 239–257.

Brasil-Neto, J.P., Pascual-Leone, A., Valls- Solé, J., Cammarota, A.,Cohen, L.G., Hallett, M., 1993. Post-exercise depression ofmotor-evoked potentials: a measure of central nervous systemfatigue. Exp. Brain Res. 93, 181–184.

Brasil-Neto, J.P., Cohen, L.G., Hallett, M., 1994. Central fatigueas revealed by postexercise decrement of motor evokedpotentials. Muscle Nerve 17, 713–719.

Butler, J.E., Thomas, C.K., 2003. Effects of sustained stimulation onthe excitability of motoneurons innervating paralyzed andcontrol muscles. J. Appl. Physiol. 94, 567–575.

Butler, J.E., Taylor, J.L., Gandevia, S.C., 2003. Responses of humanmotoneurons to corticospinal stimulation during maximalvoluntary contractions and ischemia. J. Neurosci. 23,10224–10230.

Cairns, S.P., Hing, W.A., Slack, J.R., Mills, R.G., Loiselle, D.S., 1997.Different effects of raised [K+]o on membrane potential andcontraction in mouse fast- and slow-twitch muscle. Am. J.Physiol. Cell. Physiol. 273, C598–C611.

Coombs, J.S., Curtis, D.R., Eccles, J.C., 1957. The generation ofimpulses in motoneurones. J. Physiol. (Lond.) 139, 232–249.

Costa, M.C., Johnston, D., 1996. Axonal action-potential initiation

58 B R A I N R E S E A R C H 1 3 2 0 ( 2 0 1 0 ) 4 7 – 5 9

and Na+ channel densities in the soma and axon initialsegment of subicular pyramidal neurons. J. Neurosci. 16,6676–6686.

Del Santo, F., Gelli, F., Ginanneschi, F., Popa, T., Rossi, A., 2007.Relation between isometric muscle force and surface EMG inintrinsic hand muscles as function of the arm geometry. BrainRes. 1163, 79–85.

Fowles, J.R., Green, H.J., Tupling, R., O'brien, S., Royhuman, B.D.,2002. Neuromuscular fatigue is associated with alteredNa+-K+-ATPase activity following isometric exercise. J. Appl.Physiol. 92, 1585–1593.

Gandevia, S.C., Macefield, G., Burke, D., McKenzie, D.K., 1990.Voluntary activation of human motor axons in the absence ofmuscle afferent feedback. The control of the deafferentedhand. Brain 113, 1563–1581.

Gandevia, S.C., Allen,G.M., Butler, J.E., Taylor, J.L., 1996. Supraspinalfactors in human muscle fatigue: evidence for suboptimaloutput from the motor cortex. J. Physiol. (Lond.) 490,529–536.

Gandevia, S.C., Petersen, N., Butler, J.E., Taylor, J.L., 1999. Impairedresponse of humanmotoneurones to corticospinal stimulationafter voluntary exercise. J. Physiol. (Lond.) 521, 749–759.

Gelli, F., Del Santo, F., Popa, T., Mazzocchio, R., Rossi, A., 2007.Factors influencing the relation between corticospinal outputand muscle force during voluntary contractions. Eur. J.Neurosci. 25, 3469–3475.

Ginanneschi, F., Del Santo, F., Dominici, F., Gelli, F., Mazzocchio, R.,Rossi, A., 2005. Changes in corticomotor excitability of handmuscles in relation to static shoulder positions. Exp. Brain Res.161, 374–382.

Ginanneschi, F., Dominici, F., Milani, P., Biasella, A., Rossi, A.,Mazzocchio, R., 2007a. Changes in the recruitment curve ofthe soleus H-reflex associated with chronic low back pain.Clin. Neurophysiol. 118, 111–118.

Ginanneschi, F., Dominici, F., Milani, P., Biasella, A., Rossi, A.,2007b. Evidence of altered motor axon properties of the ulnarnerve in carpal tunnel syndrome. Clin. Neurophysiol. 118,1569–1576.

Gordon, T.R., Kocsis, J.D., Waxman, S.G., 1990. Electrogenic pump(Na+/K(+)-ATPase) activity in rat optic nerve. Neuroscience 37,829–837.

Hultborn, H., Ilert, M., Nielsen, J., Paul, A., Ballegard, M., Wiese, H.,1996. On the mechanism of the post-activation depressionof the H-reflex in human subjects. Exp. Brain Res. 108,450–462.

Jammes, Y., Steinberg, J.G., Mambrini, O., Bregeon, F., Delliaux, S.,2005. Chronic fatigue syndrome: assessment of increasedoxidative stress and altered muscle excitability in response toincremental exercise. J. Intern. Med. 257, 299–310.

Johnston, D., Magee, J.C., Colbert, C.M., Christie, B.R., 1996. Activeproperties of neuronal dendrites. Annu. Rev. Neurosci. 19,165–186.

Kalmar, J.M., Cafarelli, E., 1999. Effects of caffeine onneuromuscular function. J. Appl. Physiol. 87, 801–808.

Kalmar, J.M., Cafarelli, E., 2004. Central fatigue and transcranialmagnetic stimulation: effect of caffeine and the confound ofperipheral transmission failure. J. Neurosci. Methods 138,15–26.

Kiernan, M.C., Mogyoros, I., Hales, J.P., Gracies, J.M., Burke, D., 1997.Excitability changes in human cutaneous afferents inducedby prolonged repetitive axonal activity. J. Physiol. 500,255–264.

Kiernan, M.C., Lin, C.S., Burke, D., 2004. Differences inactivity-dependent hyperpolarization in human sensory andmotor axons. J. Physiol. 558, 341–349.

Kimura, J., Yanagisawa, H., Yamada, T., Mitsudome, A., Sasaki, H.,Kimura, A., 1984. Is the F wave elicited in a select group ofmotoneurons? Muscle Nerve 7, 392–399.

Kuwabara, S., Cappelen-Smith, C., Lin, C.S., Mogyoros, I., Burke, D.,

2002. Effects of voluntary activity on the excitability of motoraxons in the peroneal nerve. Muscle Nerve 25, 176–184.

Liepert, J., Kotterba, S., Tegenthoff, M., Malin, J.P., 1996. Centralfatigue assessed by transcranial magnetic stimulation. MuscleNerve 19, 1429–1434.

Lindinger, M.I., Sjogaard, G., 1991. Potassium regulation duringexercise and recovery. Sports Med. 11, 382–401.

McKay, W.B., Tuel, S.M., Sherwood, A.M., Stokic, D.S., Dimitrijevic,M.R., 1995. Focal depression of cortical excitability inducedby fatiguing muscle contraction: a transcranial magneticstimulation study. Exp. Brain Res. 105, 276–282.

McKenna, M.J., 1992. The roles of ionic processes in muscularfatigue during intense exercise. Sports Med. 13, 134–145.

Miller, T.A., Kiernan, M.C., Mogyoros, I., Burke, D., 1995.Activity-dependent changes in impulse conduction in normalhuman cutaneous axons. Brain 118, 1217–1224.

Morita, K., David, G., Barrett, J.N., Barrett, E.F., 1993. Posttetanichyperpolarization produced by electrogenic Na(+)-K+ pump inlizard axons impaled near their motor terminals.J. Neurophysiol. 70, 1874–1884.

Overgaard, K., Nielsen, O.B., Flatman, J.A., Clausen, T., 1999.Relations between excitability and contractility in rat soleusmuscle: role of the Na+-K+ pump and Na+/K+ gradients.J. Physiol. 518, 215–225.

Palay, S.L., Sotelo, C., Peters, A., Orkland, P.M., 1968. The axonhillock and the initial segment. J. Cell. Biol. 38, 193–201.

Pedersen, T.H., Clausen, T., Nielsen, O.B., 2003. Loss of forceinduced by high extracellular [K+] in rat muscle: effect oftemperature, lactic acid and β2-agonist. J. Physiol. 551,277–286.

Peters, A., Proskauer, C.C., Kaiserman-Abramof, I.R., 1968. Thesmall pyramidal neuron of the rat cerebral cortex: the axonhillock and the initial segment. J. Cell Biol. 39, 604–619.

Pitcher, J.B., Miles, T.S., 2002. Alterations in corticospinalexcitability with imposed vs. voluntary fatigue in human handmuscles. J. Appl. Physiol. 92, 2131–2138.

Rothwell, J.C., 1997. Techniques and mechanisms of action oftranscranial stimulation of the human motor cortex.J. Neurosci. Methods 74, 113–122.

Ruff, R.L., Simoncini, L., Stuhmer, W., 1988. Slow sodiumchannel inactivation in mammalian muscle: a possible rolein regulating excitability. Muscle Nerve 11, 502–510.

Samii, A., Wasserman, E.M., Ikoma, K., Mercuri, B., Hallett, M.,1996. Characterization of postexercise facilitation anddepression of motor evoked potentials to transcranialmagnetic stimulation. Neurology 46, 1376–1382.

Samii, A., Wassermann, E.M., Hallett, M., 1997. Post-exercisedepression of motor evoked potentials as a function ofexercise duration. Electroencephalogr. Clin. Neurophysiol. 105,352–356.

Sejersted, O., Sjogaard, G., 2000. Dynamics and consequences ofpotassium shifts in skeletal muscle and heart during exercise.Physiol. Rev. 80, 1411–1481.

Shushakov, V., Stubbe, C., Peuckert, A., Endeward, V., Maassen, N.,2007. The relationships between plasma potassium, muscleexcitability and fatigue during voluntary exercise in humans.Exp. Physiol. 92, 705–715.

Sjogaard, G., 1990. Exercise-induced muscle fatigue: thesignificance of potassium. Acta Physiol. Scand. Suppl. 593,1–63.

Taylor, J.L., Butler, J.E., Allen, G.M., Gandevia, S.C., 1996. Changesin motor cortex excitability during human muscle fatigue.J. Physiol. (Lond.) 490, 519–528.

Thomas, C.K., 1997. Fatigue in human thenar muscles paralysedby spinal cord injury. J. Electromyogr. Kinesiol. 7, 15–26.

Unsworth, K., Hicks, A., McKelvie, R., 1998. The effect of β-blockadeon plasma potassium concentrations and muscle excitabilityfollowing static exercise. Pflugers Arch. 436, 449–456.

Vagg, R., Mogyoros, I., Kiernan, M.C., Burke, D., 1998.

59B R A I N R E S E A R C H 1 3 2 0 ( 2 0 1 0 ) 4 7 – 5 9

Activity-dependent hyperpolarization of human motor axonsproduced by natural activity. J. Physiol. 507, 919–925.

West, W., Hicks, A., McKelvie, R., O'Brien, J., 1996. The relationshipbetween plasma potassium, muscle membrane excitabilityand force following quadriceps fatigue. Pflugers Arch. 432,43–49.

Zanette, G., Bonato, C., Polo, A., Tinazzi, M., Manganotti, P., Fiaschi,

A., 1995. Long-lasting depression ofmotor-evoked potentials totranscranial magnetic stimulation following exercise.Exp. Brain Res. 107, 80–86.

Ziemann, U., Tergau, F., Wassermann, E.M., Wischer, S.,Hildebrandt, J., Paulus, W., 1998. Demonstration of facilitatory Iwave interaction in the human motor cortex by pairedtranscranial magnetic stimulation. J. Physiol. 511, 181–190.