Neuroscience Letters 383 (2005) 339–344

Hindlimb stepping movements in complete spinal rats inducedby epidural spinal cord stimulation

R.M. Ichiyamaa, Yu.P. Gerasimenkoa,c, H. Zhonga, R.R. Roya,b, V.R. Edgertona,b,∗a Department of Physiological Science, University of California, Los Angeles, 1804 Life Science Building, 621 Charles E Young Drive,

Los Angeles, CA 90095, USAb Brain Research Institute, University of California, Los Angeles, CA 90095, USA

c Pavlov Institute of Physiology, Russian Academy of Sciences, St. Petersburg, Russia

Received 13 December 2004; received in revised form 15 April 2005; accepted 15 April 2005

Abstract

The locomotor ability of the spinal cord of adult rats deprived of brain control was tested by epidural spinal cord stimulation. The studieswere performed on six rats that had a complete spinal cord transection (T7–T9) and epidural electrode implantations 2–3 weeks before testingwas initiated. The stimulating epidural electrodes were implanted at the T12–L6 spinal segments. Epidural electrical stimulation of the dorsals ilitation wasu n at mosts comotora tiation ofl rom brainc l segment.F©

K

StaiatZcst(muc

liken.er-at inpinalhanin thelanti-enessfterroniclatedinto

-y inpor-igm

0d

urface of the spinal cord at frequencies between 1 and 50 Hz and intensities between 1 and 10 V without any pharmacological facsed. Stimulation at each of the lumbar spinal cord segments elicited some rhythmic activity in the hindlimbs. However, stimulatioegmental levels usually evoked activity in only one leg and was maintained for short periods of time (<10 s). Bilateral hindlimb loctivity was evoked most often with epidural stimulation at 40–50 Hz applied at the L2 segment. A necessary condition for ini

ocomotor activity was providing a specific amount (at least 5%) of body weight support. Therefore, the rat spinal cord isolated fontrol is capable of producing bilateral stepping patterns induced most readily by epidural stimulation applied at the L2 spinaurthermore, the induced stepping patterns were dependent on sensory feedback associated with weight bearing.2005 Elsevier Ireland Ltd. All rights reserved.

eywords:Epidural electrical stimulation; Locomotion; Spinal cord; Rats

pinal neural networks play an important role in the organiza-ion of locomotion. These spinal networks, often referred tos central pattern generators (CPGs), are capable of produc-

ng step-like efferent patterns in the absence of supraspinalnd/or afferent input. CPGs can be activated by administra-

ion of a variety of pharmacological agents[23]. Grillner andangger[14] also elicited locomotor activity by tonic electri-al stimulation of the dorsal columns or dorsal roots in acutepinal cats injected with nialamide andl-Dopa. Some fea-ures of hindlimb locomotion in acute and chronic spinal catsafferents intact) have been evoked in response to intraspinalicro-stimulation[1,19], as well as epidural spinal cord stim-lation [11–13,16]. In human patients with complete spinalord injuries, Dimitrijevic et al.[9] have shown that epidural

∗ Corresponding author.E-mail address:vre@ucla.edu (V.R. Edgerton).

stimulation of L2 was the most effective in inducing step-movements when the patients were in a reclined positio

The study of the neural control of locomotion is often pformed using the rat as a model considering the fact thsome ways the anatomical organization of the human scord rostro—caudally is more similar to that of the rat tthe cat, e.g., number of spinal segments and segmentslumbar enlargement. In addition, humans and rats are pgrade animals, whereas cats are digitigrade. The effectivof epidural electrical stimulation to induce locomotion aa complete spinal cord transection either in acute or chadult rats is unknown. Given the prominent use of the isospinal cord from neonatal rats as a model to gain insightthe neural control of locomotion in adult animals[4,5], the capability of the adult spinal rat to generate locomotor activitresponse to epidural stimulation becomes particularly imtant. From this perspective the epidural stimulation parad

304-3940/$ – see front matter © 2005 Elsevier Ireland Ltd. All rights reserved.oi:10.1016/j.neulet.2005.04.049

340 R.M. Ichiyama et al. / Neuroscience Letters 383 (2005) 339–344

represents a new model for studying spinal control of loco-motion in an unanesthetized adult in vivo preparation withoutpharmacological facilitation. Therefore, we examined the ca-pability of the adult spinal cord to produce locomotor patternsin response to epidural stimulation of the dorsal surface ofthe spinal cord. We identified the most effective sites on thespinal cord dorsal surface and the optimal stimulus parame-ters for eliciting hindlimb locomotor activity in response toepidural electrical stimulation.

All experimental procedures comply with the guidelinesof the National Institute of Health Guide for the Care and Useof Laboratory Animals. Six adult female Sprague–Dawleyrats (200–250 g) were anesthetized with a combination ofketamine (100 mg/kg) and xylazine (10 mg/kg). A deep levelof anesthesia was maintained during the procedures (supple-mental doses of ketamine administered, i.p. as needed). Allsurgical procedures were performed under aseptic conditions.A mid-dorsal skin incision was made from∼T4 to∼S5. Thefascia and muscles at the appropriate vertebral levels (seethe following) were removed. In all rats, a partial laminec-tomy was performed at a mid-thoracic level (∼T7–T9) andthe spinal cord was completely transected using fine scis-sors and forceps. Small cotton balls were used to dry thearea and two surgeons verified the completeness of the tran-section. Gel foam was inserted into the gap created by thetransection as a coagulant and to prevent reconnection of thec werep crals lec-t onerW nousp verte-b ationen l wireo thenw uram ition,t s top onso ec-t odesw lanta d asf . Ac -m in them

con-n de-s on-n rougha nts).M lo-c tionw tion

was delivered at 30, 40 or 50 Hz. An upper body harnesssupport system was used to place the rats on a treadmill toperform bipedal locomotion and standing[7]. During loco-motion, the speed of treadmill was 11 cm/s. An automatedbody weight support system was used to provide the spe-cific amount of body weight support that each rat neededto perform the required task. Video analysis (Peak MotionAnalysis System) of the hip, knee and ankle joints was ob-tained to determine segmental and joint angular kinematics.A six-camera system with retro-reflective markers placedon bony landmarks at the iliac crest, greater trochanter, lat-eral condyle, lateral malleolus and the distal end of thefifth metatarsal on both legs was used. This arrangement al-lowed us to perform a 3-D reconstruction of the synchronizedmovements of both legs and provided information in severalanatomical planes. Stick diagrams and trajectories were cal-culated using Motus (Peak System) software.

Before any spinal cord stimulation, the two electrodes ateach segment were connected (via an isolation unit) and usedas a single electrode. Electrical stimulation was applied sym-metrically from the midline of the spinal cord. Rectangu-lar pulses (200�s duration) were delivered with frequenciesfrom 1 to 50 Hz. Pulse duration was chosen based on previouswork by Dimitrijevic and colleagues[9,21]. The stimulationintensity ranged from 0 to a maximum of 13 V. Each rat wastested four times at different time points after the spinal cordw timu-l t was3 ula-t stainw t thee fullyd l im-p tep-p ng at ry toe mo-t T12a asm

eg-m y int lev-emt ionm a-t nga in-d eryp ther nlyt aring( se tos tim-u ion,

ut ends of the spinal cord. Partial laminectomies thenerformed at various lower thoracic, lumbar or higher sapinal segments for implantation of epidural stimulating erodes. Teflon-coated stainless steel wires (AS632; Co

ire, Chatsworth, CA, USA) were passed under the spirocesses and above the dura matter of the remainingrae between the partial laminectomy sites. The stimullectrodes were made by removing a small portion (∼1 mmotch) of the Teflon coating to expose the stainless steen the surface facing the spinal cord. The electrodesere secured in position by suturing the wire to the datter above and below the segment of interest. In add

he Teflon was gently pulled over the cut end of the wirerevent stimulation through this site. Different combinatif sites from T12 to L6 were implanted bilaterally with el

rodes across animals. Two, three or four pairs of electrere implanted in each rat, with all rats having an impt L2. The wires for each pair of electrodes were place

ar lateral to the midline of the spinal cord as possibleommon ground (indifferent) wire (∼1 cm of the Teflon reoved at the distal end) was inserted subcutaneouslyid-back region.The electrode wires were connected to an amphenol

ector that was cemented to the skull of the animal ascribed previously[24]. The amphenol connector was cected to a Grass S88 Stimulator (Grass Instruments) thstimulus isolation unit (Grass SIU5, Grass Instrumeuscle twitch threshold and the threshold for inducing

omotor activity under high frequency epidural stimulaere determined. Continuous epidural electrical stimula

as transected with a 2- to 3-day rest period between sation bouts. The duration of a testing session for each ra0–40 min. Within a given testing session, epidural stim

ion was delivered for as long as the rats were able to sueight-bearing stepping, and this time was recorded. And of experiments, the spinal cords of all rats were careissected and the location and placement of the epiduralants were verified. Statistical comparisons of bilateral sing duration between L2 and T12 were performed usi

-test. A comparison of the stimulus intensity necessalicit a muscle contraction to a single pulse versus loco

ion to constant stimulus was also performed betweennd L2 using at-test. All quantitative data are expressedean± (S.D.).In general, stimulation of any of the spinal cord s

ents tested (T12–L16) elicited some rhythmic activithe hindlimbs. However, stimulation at most segmentalls usually evoked activity only in one leg (Table 1) and wasaintained only for short periods of time (Fig. 1). In addi-

ion, stimulation of the L3–L6 segments often induced flexovements (Table 1). A necessary condition for the initi

ion of locomotion by epidural stimulation was providispecific amount of body weight support. Locomotion

uced by epidural stimulation during “air stepping” was voor. Hindlimb locomotor activity occurred only whenats bore 5–20% of their body weight. If the hindlimbs oouched the treadmill belt or there was too much load be>20%), no stepping patterns were observed in respontimulation. The spinal rats stepped only when epidural slation was delivered to the spinal cord. Without stimulat

R.M. Ichiyama et al. / Neuroscience Letters 383 (2005) 339–344 341

Table 1Effect of epidural stimulation at different spinal segment levels on weightsupported locomotor behavior

Behaviora Spinal segment level

T13 T12 L1 L2 L3 L4 L5 L6

Bilateral stepping 0/3 5/8 0/1 14/19 0/3 0/3 0/4 1/6Unilateral stepping 3/3 3/8 3/4 5/19 0/3 2/3 2/4 2/6Flexion movements 0/3 0/8 0/4 0/19 3/3 1/3 2/4 2/6Synchronous stepping 0/3 0/8 0/4 0/19 0/3 0/3 0/4 1/6

a The number of trials that produced the corresponding effect relative tothe total number of stimulation trials performed when stimulated epidurallyat 40 Hz.

the rat hindlimbs simply dragged on the moving treadmillbelt. Locomotor activity stopped the moment that the epidu-ral stimulation was turned off, even when the treadmill beltwas still moving.

At lower frequencies of stimulation (1–10 Hz) of the lum-bar segments, hindlimb muscle contractions were synchro-nized with the electrical pulses and, in most cases resulted inflexion movements. Hindlimb locomotor activity was evokedmost often when epidural stimulation at frequencies between30 and 50 Hz was applied to the L2 segment (Table 1). Thisphenomenon was observed in all animals tested. Such alter-nating activity was evoked in 14 out of 19 stimulation trials(Table 1) and the mean duration of this bilateral stepping was18.2± 13.5 s (Fig. 1). A mean duration of 7.3± 1.5 s of bilat-eral stepping was triggered by stimulation of the T12 segment(five out of eight stimulation trials). The difference betweenL2 and T12 was statistically significant (p< 0.01). Unilateral(two out of six stimulation trials) or alternating (one out of

Fig. 1. Mean duration of hindlimb rhythmic activity induced by epiduralstimulation (40 Hz) applied at different spinal segments.n, number of steps.Bars, S.D.

six stimulation trials) activity of the hindlimbs was observedsporadically with stimulation of the L6 segment.

The typical kinematics of the locomotor activity in re-sponse to stimulation of the T12, L2 and L6 segments at40 Hz are presented inFig. 2. Stick diagrams of one stepcycle show a locomotor pattern induced by epidural stimula-tion of the L2 segment that approximates that observed whencontrol rats step bipedally[18]. The stepping quality was notas good with stimulation at other spinal segments. For ex-ample, stimulation at T12 resulted in short step lengths, and

F timulat on bk timulat ulatL

ig. 2. Kinematics of the hindlimb movements induced by epidural snee and ankle angles during bipedal stepping induced by epidural s2 than the T12 or L6 segments.

ion at different spinal segments. The graphs illustrate the coordinatietweenion. Note that this pattern is more organized and consistent with stimion at the

342 R.M. Ichiyama et al. / Neuroscience Letters 383 (2005) 339–344

Fig. 3. The average thresholds for inducing a hindlimb response to a singlepulse of epidural stimulation and for inducing locomotor activity via 40Hzepidural stimulation at the T12, L2 and L6 spinal segments. The numbers ofobservations are listed in Table 1. Bars, S.D.

stimulation at L6 resulted in a prolonged stance and virtuallyno plantar placements. Moreover, the coordination betweenknee and ankle joints was more similar to that observed incontrol rats with stimulation at the L2 than the T12 or L6segments.

The threshold for generating a hindlimb muscle responsewith a single pulse and the threshold for inducing hindlimblocomotor activity at 40 Hz differed among the spinal seg-ments (Fig. 3). The mean threshold for a muscle response tosingle pulse stimulation at the L2 segment was 2.69± 0.43 Vand the threshold for inducing bilateral locomotor activityto 40 Hz was 5.76± 0.86 V. This difference was statisticallysignificant (p< 0.005).

The threshold to induce locomotor activity increased as afunction of the time after spinal cord transection. For exam-ple, epidural stimulation of the L2 segment, performed twoweeks after transection-induced hindlimb locomotor activityat an average stimulation strength of 3.39± 0.2 V, whereasthis value was 7.55± 1.2 V at 4 weeks after transection. Thisdifference was statistically significant (p< 0.05). The meanduration of the bouts of hindlimb locomotor activity, however,were significantly longer at 4 weeks (15.71± 7.31 s) than 2weeks (5.5± 3.5 s) after spinal cord transection. The averagethreshold for a hindlimb muscle response to a single pulse, incontrast, was similar at the two time points, i.e., 2.7± 0.8 Vand 3.39± 0.2 V at 2 and 4 weeks post-transection, respec-t

, al-t atsc liede nt atf ween3 eL bilat-e of

the hindlimbs was evoked sporadically with stimulation atthe L6 segment. Similar epidural stimulation of other spinalsegments was not as effective: epidural stimulation at theT13 or L1 spinal segments evoked rhythmic activity in onlyone leg, and stimulation at the L3, L4, or L5 segments pro-duced mainly flexion movements. In contrast to stimulationat the L2 segment, step-like movements produced by stimula-tion at the L6 segment reflected a predominantly peripheral,not central, origin since it was evoked only the moment thefoot touched the moving treadmill belt. This observation in-dicates that stimulation at L6 increases the excitability of thestepping-related neural networks, but that afferent input wasrequired to induce the stepping movements.

It should be emphasized that the observed stepping pat-terns were induced by epidural electrical stimulation with-out any pharmacological facilitation, as has been reported byMushawar et al.[20] using intraspinal stimulation in spinalcats. In contrast, hindlimb locomotor activity has been evokedmainly via a combination of electrical stimulation and phar-macological interventions in many previous reports. In acutespinal cats, for example, hindlimb locomotor activity wasevoked by stimulation of the dorsal roots only in the pres-ence of the noradrenergic precursorl-Dopa[2]. Grillner andZangger[14] also elicited hindlimb locomotor activity inacute, spinal cats in response to stimulation of the dorsal rootsor dorsal columns in conjunction with administration of L-D cedb ner-g

mbl ords ffica-c theL np e them tep-l do andL imbl f-f cordo tionz , i.e.,i

urals thed clos-e ed a sin-g entsa o ini-t ithf ssaryt telyt ndi-t tion

ively.We have demonstrated for the first time that rhythmic

ernating hindlimb locomotor activity in chronic spinal ran be induced effectively by electrical stimulation apppidurally to the dorsal surface of the L2 spinal segme

requencies between 30 and 50 Hz, pulse intensities betand 11 V, and a pulse width of 200�s. Stimulation at th

2 segment consistently produced the most successfulral stepping. Rhythmic unilateral or alternating activity

opa. In chronic spinal cats, locomotor activity was induy intraspinal stimulation only after clonidine (a noradreic agonist) administration[1].

The efficacy of epidural stimulation to induce hindliocomotor activity clearly varied according to the spinal cegment stimulated. In the present study, the most eious region for stimulation was clearly localized around2 spinal cord segment (Table 1; Fig. 1). In spinal humaatients, the upper lumbar segments also appear to bost effective regions to stimulate to induce a bilateral s

ike rhythm[9] and weight-bearing steps[3]. In decerebrater chronic spinal cats, in contrast, stimulation at the L45 segments was the most effective for inducing hindl

ocomotor activity[11–13,16]. Taking into account the dierences in the anatomical organization of the spinalf the mammals mentioned above, the effective stimulaone to induce locomotion seems to be at a similar siten the lumbar pre-enlargement area.

Theoretically, the electrical fields produced by epidtimulation could preferentially activate the afferents inorsal roots since they are the largest fibers and are thest to the stimulating electrodes[6,22]. Therefore, when wetermined the threshold for the leg muscle response tole epidural shock, we most likely were activating Ia affernd evoking monosynaptic reflexes in the leg muscles. T

iate hindlimb locomotor activity by epidural stimulation wrequencies between 30 and 50 Hz, however, it was neceo increase the strength of the stimulation by approximawo-fold above muscle twitch threshold. Under these coions, the electrical fields produced at the higher stimula

R.M. Ichiyama et al. / Neuroscience Letters 383 (2005) 339–344 343

intensities will excite adjacent spinal structures, in particularthe propriospinal neurons of the dorsolateral funiculus[6].Thus, the pathway(s), i.e. through the dorsal roots or throughthe propriospinal system of the dorsolateral funicular[17,25],responsible for initiating locomotion via epidural stimulationin the present study cannot be clearly identified. Additionally,pulse duration has been suggested to significantly affect theactivation of neural substrates in humans with incompletespinal cord injuries[3,22] and spinal cats[16]. Since we didnot manipulate this variable in our current experiments, therole of longer pulse durations in an adult spinalized rat isunclear.

Another issue is the significance of peripheral feedbackin the regulation of hindlimb locomotor activity via epiduralspinal cord stimulation. On one hand, there are data show-ing an independence from peripheral input for producingstepping patterns via epidural stimulation since such activ-ity can be evoked under conditions of fictive locomotion[16]. On the other hand, the locomotor pattern induced byepidural stimulation changes depending on foot contact onthe treadmill belt. On a moving treadmill belt decerebratedcats demonstrate well-coordinated hindlimb locomotion inresponse to epidural stimulation. When the cat hindlimbs areheld above the treadmill belt, however, their stepping patternchanges dramatically[12]. Under this condition, the sameepidural stimulation induces air rhythmic movements of theh d arei clesa l belt.A r ind cordi

e in-p tiono la-t ent otora ng’w witht tinga arec k inr l an-i anp

sibil-i rals seg-m ore,t pidu-r herala , thep la-t ingb y int

Acknowledgements

This work was supported by the Roman Reed Funds,CRPF VEC-9901, RFBR Grant 04-04-48772a and RGNF03-06-00315.

References

[1] D. Barthelemy, H. Leblond, S. Rossignol, Electrical stimulation ofthe lumbar spinal cord and importance of mid lumbar segments ineliciting spinal locomotion, Soc. Neurosci. Abstr. (2002) 69.19.

[2] N. Budakova, Stepping movements evoked by repetitive dorsal rootstimulation in a mesencephalic cat, Neurosci. Behav. Physiol. 5(1972) 355–363.

[3] M.R. Carhart, J. He, R. Herman, S. D’Luzansky, W.T. Willis, Epidu-ral spinal-cord stimulation facilitates recovery of functional walkingfollowing incomplete spinal-cord injury, IEEE Trans. Neural. Syst.Rehabil. Eng. 12 (2004) 32–42.

[4] J. Cazalets, S. Bertrand, Coupling between lumbar and sacral motornetworks in the neonatal rat spinal cord, Eur. J. Neurosci. 12 (2000)2993–3002.

[5] F. Clarac, E. Pearlstein, J. Pflieger, L. Vinay, The in vitro neonatalrat spinal cord preparation: a new insight into mammalian locomotormechanisms, J. Comp. Physiol. (2004) 343–357.

[6] B. Coburn, W. Sin, A theoretical study of epidural electrical stimu-lation of the spinal cord. Part I. Finite element analysis of stimulusfields, IEEE Trans. Biomed. Eng. 32 (1985) 971–977.

[7] R.D. de Leon, D.J. Reinkensmeyer, W.K. Timoszyk, N.J. London,ptive002)

ra-tion,

nal998)

[ .R.ev.

[ o-inal

[ cher,rated. I.

[ o-pinal

[ n in

[ bkin,ding

[ tim-28

[ u-xp.

[ on-xonaladult

indlimbs, but the movements are small in amplitude annduced mainly by the muscles at the hip, not by the must the ankle as observed during stepping on the treadmilfferent input also has been shown to be a critical factoetermining the stepping performance induced in spinal

njured humans via implanted epidural electrodes[3].In the present experiments we show that proprioceptiv

ut from hindlimbs plays an important role in the regulaf hindlimb locomotor activity induced by epidural stimu

ion. Robust hindlimb locomotor activity occurred only whhe rats bore at least 5% of their body weight. The locomctivity induced by epidural stimulation during ‘air steppias very poor. In other words, the contact of the paws

he treadmill belt was a necessary condition for initiand maintaining robust locomotor activity. These dataonsistent with the important role of peripheral feedbacegulating the locomotor performance of complete spinamals [8,10,17]and incomplete and complete spinal humatients[3,8,15,26].

In general, the present findings demonstrate the posty of inducing locomotion in chronic spinal rats by epidutimulation, and that the dorsal region of the L2 spinalent is a particularly effective site to stimulate. Furtherm

he robustness of the locomotor pattern in response to eal stimulation is dependent on the presence of peripfferent feedback associated with weight bearing. Thusossibility of a combination of centrally (epidural stimu

ion) and peripherally (locomotor training) induced steppeing an effective method for restoring locomotor activit

he absence of brain control should be explored further.

R.R. Roy, V.R. Edgerton, Use of robotics in assessing the adacapacity of the rat lumbar spinal cord, Prog. Brain Res. 137 (2141–149.

[8] V. Dietz, M. Wirz, A. Curt, G. Colombo, Locomotor pattern in paplegic patients: training effects and recovery of spinal cord funcSpinal Cord 36 (1998) 380–390.

[9] M. Dimitrijevic, Yu. Gerasimenko, M. Pinter, Evidence for a spicentral pattern generator in humans, Ann. NY Acad. Sci. 860 (1360–376.

10] V.R. Edgerton, N. Tillakaratne, A.J. Bigbee, R.D. de Leon, RRoy, Plasticity of the spinal neural circuitry after injury, Annu. RNeurosci. 27 (2004) 145–167.

11] Yu. Gerasimenko, V. Avelev, O. Nikitin, I. Lavrov, Initiation of lcomotor activity in spinal cats by epidural stimulation of the spcord, Neurosci. Behav. Physiol. 33 (2003) 247–254.

12] Yu. Gerasimenko, I. Lavrov, I. Bogacheva, N. Sherbakova, V. KuP. Musienko, Features of stepping pattern formation in decerebcats under epidural spinal cord stimulation, Ross Fiziol. Zh. ImM. Sechenova 89 (2003) 1046–1057 (in Russian).

13] Yu. Gerasimenko, A. Makarovskii, O. Nikitin, Control of locomtor activity in humans and animals in the absence of suprasinfluences, Neurosci. Behav. Physiol. 32 (2002) 417–423.

14] S. Grillner, P. Zangger, On the central generation of locomotiothe low spinal cat, Exp. Brain Res. 34 (1979) 241–261.

15] S.J. Harkema, S.L. Hurley, U.K. Patel, P.S. Requejo, B.H. DoV.R. Edgerton, Human lumbosacral spinal cord interprets loaduring stepping, J. Neurophysiol. 77 (1997) 797–811.

16] T. Iwahara, Y. Atsuta, E. Garcia-Rill, R. Skinner, Spinal cord sulation -induced locomotion in the adult cat, Brain Res. Bull.(1991) 99–105.

17] O. Kazennikov, M. Shik, G. Yakovleva, Stepping elicited by stimlation of the dorsolateral funiculus in the cat spinal cord, Bull. EBiol. Med. 96 (1983) 8–10 (in Russian).

18] M.D. Kubasak, H. Zhong, R.R. Roy, V.R. Edgerton, A. RamCueto, P.E. Phelps, Weight-bearing hindlimb stepping and aregeneration after OEG transplantation and step training in

344 R.M. Ichiyama et al. / Neuroscience Letters 383 (2005) 339–344

paraplegic rats, Program No. 714.10. 2004 Abstract Viewer/ItineraryPlanner. Society for Neuroscience, Washington, DC, 2004 (Online).

[19] V.K. Mushahwar, D.F. Collins, A. Prochazka, Spinal cord micros-timulation generates functional limb movements in chronically im-planted cats, Exp. Neurol. 163 (2000) 422–429.

[20] V.K. Mushahwar, D.M. Gillard, M.J. Gauthier, A. Prochazka, In-traspinal micro stimulation generates locomotor-like and feedback-controlled movements, IEEE Trans. Neural. Syst. Rehabil. Eng. 10(2002) 68–81.

[21] M.M. Pinter, F. Gerstenbrand, M.R. Dimitrijevic, Epidural electricalstimulation of posterior structures of the human lumbosacral cord:3. Control of spasticity, Spinal Cord 38 (2000) 524–531.

[22] F. Rattay, K. Minassian, M.R. Dimitrijevic, Epidural electrical stim-ulation of posterior structures of the human lumbosacral cord. 2.

Quantitative analysis by computer modeling, Spinal Cord 38 (2000)473–489.

[23] S. Rossignol, C. Chau, E. Brustein, N. Giroux, L. Bouyer, H. Bar-beau, T.A. Reader, Pharmacological activation and modulation of thecentral pattern generator for locomotion in the cat, Ann. N. Y. Acad.Sci. 860 (1998) 346–359.

[24] R.R. Roy, D.L. Hutchison, D.J. Pierotti, J.A. Hodgson, V.R. Edger-ton, EMG patterns of rat ankle extensors and flexors during treadmilllocomotion and swimming, J. Appl. Physiol. 70 (1991) 2522–2529.

[25] M. Shik, Recognizing propriospinal and reticulospinal systems ofinitiation of stepping, Motor Control 1 (1997) 310–313.

[26] A. Wernig, S. Muller, A. Nanassy, E. Cagol, Laufband therapy basedon “rules of spinal locomotion” is effective in spinal cord injuredpersons, Eur. J. Neurosci. 7 (1995) 823–829.