Molecular Psychiatry(1997) 2, 188191 1997 Stockton Press All rights reserved 13594184/ 97 $12.00NEWS & VIEWSRemodeling of cortical motor representations afterstroke: implications for recovery from brain damageRecent neurophysiological results in non-human primates suggest that the adjacent, undam-agedcortexmayassumethefunctionof thedamagedcortex, but onlywhenrehabilitativetherapy is provided after injury.After stroke or other injury to the cerebral cortex, it is motor representations can be altered by peripheralnerve lesions in both developing and adult rats.3not uncommon for patients to experience gradualrecovery of sensorimotor functions. For example, after Further, our own studies in primates have shown thatacquisition of manual skills is accompanied by task- damage to themotor cortex, aperiod of paralysis of thecontralateral musculature is often followed by limited specic changes in motor hand representations,4dem-onstrating that behavioral experience can shape the recovery of voluntary movement, strength and motorcoordination. However, complete recovery of motor functional organization of cerebral cortex.Over the past several years we have conducted abilities, especially skilled use of the hand, is rare.Historically, a recurrent theory to explain recovery detailed neurophysiological studies of the motor handrepresentation in adult squirrel monkeys usingconven- of function following brain injury has been that otherbrain regions must take over the damaged function. tional intracorticalmicrostimulation techniques.Inthese experiments conducted in anesthetized animals, Though this theory has considerable appeal in light ofrecent demonstrations of functional plasticity of rep- representational maps of the hand are derived fromhundreds of closely spaced microelectrodepen- resentational maps in the cerebral cortex, very littledirect empirical evidenceexists in either brain-injured etrations250m apart. At each site, brief trains oflow-intensity current pulses are injected to dene the humans or in animalmodels.1To test this theorydirectly, we recently introduced a primate model for movement or movements evoked at thelowest possiblecurrent levels. Since the corticospinal neurons that examining functional recovery of motor abilities afterbrain injury. Theadvantageof this model isthat wecan project to spinal motor neuron pools are located at aconstant depth relative to the cortical surface, one can track therecovery of manual skill after asmall injury tothe motor cortex, and determine the neurophysiolog- reliably evokemovements at each sitein themotor cor-tex with currents averagingonly 1015A. Fromthese icalcorrelates, if any, of behavioralcompensation.From these studies, it would appear that neurophysio- data, two-dimensional maps of the motor topographyaredrawn. Theseproceduresare widely used in alarge logicalremodeling can occur in the healthy tissueadjacent to a damaged part of the motor cortex, and number of neurophysiological laboratories to study theorganization of the motor cortex. under certain conditions, this reorganized tissue maysubserve the damaged function. Recently we used these techniques to determinewhether neurophysiological changes occur in motor The primary motor cortex (also known as area 4 orM1) is located in a narrow strip of cerebral cortex representations following cortical injury. In our initialstudy, we examined spontaneous recovery of manual within the precentral gyrus. M1 contains an orderlyrepresentation of thebody musculature(themotor map skills after injury.5First, squirrel monkeys were testedbriey on a task requiring skilled use of the hand to or homunculus), with muscles of thelower extremitiesrepresented in the medial portion, and those of the retrieve pellets fromfood wells of different diameters.Then, a few weeks later, a baseline motor map of the upper extremities in the lateral portion. Though it wasonce thought that motor representations are static in hand was derived in the motor cortex contralateral tothe monkeys preferred hand. This map was used to adults, this view is no longer tenable. It is now widelyheld that functional representations are dynamically dene a small portion of the hand representation thatwas subsequently damaged by permanent occlusion of maintained throughout life.2 For example,corticala small vascular bed on the surface of the cortex. Theinfarct targeted 2530% of the hand representation,leaving a large portion of undamaged hand represen-Correspondence: RJ Nudo, PhD, Department of Neurobiology andtation adjacent to the injury.Anatomy, University of Texas Health Science Center at Houston,Houston, TX 77030, USA The infarct resulted in an immediate and profoundNews &Views189decit in skilled use of the preferred hand, as evi- neuronalplasticity was responsible for recovery ofmotor function in this study, then it is likely that the denced by adecreasein retrieval efciency and atend-ency to use the non-preferred (unimpaired) hand to neural substrate was located in another motor struc-ture, either cortical or subcortical. retrieve the pellets. At periodic intervals, manual skillwas retested in probe trials. More detailed testing was In a second study, we examined functionalremodeling ofcorticalmotor representations when not conducted, in order to minimizetheeffectsof train-ing in this study of spontaneous recovery. After 34 monkeys underwent an intensivere-trainingprocedureafter cortical injury, simulating the effects of physical months, skilllevels were similar to those obtainedprior to the infarct. Then, the motor cortex was re- rehabilitation.6First, squirrel monkeys weretrained onthe manual skill task. Daily training consisted of two examined. The results showed that in the healthy cor-tex adjacent to the infarct, more proximal (ie arm and 30-minutesessions per day. In addition, monkeysworeajacket that contained asleeveextendingthe length of shoulder) movement representations replaced thehandrepresentations (Figure1). Thus, relatively small thenon-preferred hand, restrictingits use. Themonkeycould usethe restricted forelimb for climbing and sup- lesions in representations of the hand result in wide-spread reduction in the spared hand representations port, but was unable to grasp small objects with thengers. The jacket was worn by the monkeys through- adjacent to the lesion, and apparent increases in theadjacent proximal representations. This loss occurred out theexperiment, except duringthesurgical and neu-rophysiological mapping procedures. This allowed us despite an apparent recovery of motor skill. Thus, ifFigure1Two-dimensional maps of the hand representation in primary motor cortex showing functional remodeling beforeand a few months after an ischemic infarct, as might occur in stroke. Data were derived from hundreds of microelectrodepenetrations using conventional microstimulation procedures. In animals that spontaneously recovered motor hand skills(Lesion/no rehab), the adjacent, undamaged hand representation was invaded by more proximal representations. In animalsthat underwent intensive retrainingof motor hand skills (Lesion/rehab), the hand representation expanded, invadingregionsformerly occupied by representations of elbow and shoulder. These results suggest that after an injury to motor cortex, use oftheaffected limbis important in functional remodelingin theremaining, healthy tissue. It implies that theadjacent, undamagedtissuecan take over the damaged function. The dotted line in the pre-lesion map shows the siteof intended infarct. Becauseof tissue necrosis, the lesion appears smaller in post-lesion maps. The inset shows a dorsal view of a squirrel monkey brainand the scope of the neurophysiological mapping procedures.News &Views190to examine the motor performance of the preferred improvechronic motor function. For example, Taub etal demonstrated that forced use of the impaired limb (impaired) hand, both before and after the infarct.Initially, retrieval of pellets fromthe smallest well was in human stroke patients for 14 days, by constraint ofthe unaffected limb, resulted in long-term improve- difcult to perform. Monkeys were gradually encour-aged to retrieve pellets fromthesmallest well by asim- ment of motor function in the impaired limb.7 Severallines of evidencefromboth human and animal models ple titration procedure. Eventually the monkeys wereable to use their ngers in a skilled, stereotyped man- using neurophysiological, positron emission tomogra-phy (PET) and functional magnetic resonance imaging ner to efciently retrieve pellets without error fromthesmallest well. Oncecriterion performancewas attained (MRI) are beginning to shed light on the exact neuralsubstrates underlying recovery of function. For (usually afew weeks), apre-injury baselinemotor mapwas derived. Then a small ischemic infarct was made example, PET imaging of human patients with tumorslocated in the hand area of the motor cortex has in a portion of the hand representation, as before.Within 5 days, the monkeys began a retraining pro- revealed that blood ow changes related to voluntarynger movements are shifted compared with the unaf- cedure on this same task using the same titration pro-cedure used prior to the injury. Recovery of skilled fected motor cortex.8Other imagingstudies in humansand animals have demonstrated that motor hand rep- hand use occurred gradually over a few weeks. Oncehand skill was re-attained, neurophysiologicaltech- resentations are alterableby usein normal, intact indi-viduals.9niques were used to deriveasecond, post-injury motormap. While the cortical tissue underlying the infarct Despite evidence that use of the impaired upperextremity has a positive effect on the neurophysiolog- was necrotic, and thus unresponsive to microstimu-lation,stimulation ofthe adjacent,healthytissue ical consequences after stroke, recent studies in ratspaint a very different picture. Restricting the unim- resulted inevoked movements at current levelscomparable to those obtained prior to the infarct. In pairedforelimb inrats withsensorimotor cortexlesions results in prevention of dendritic growth in the contrast to the monkeys that received no post-injurytraining, recovered monkeys showed a net expansion uninjured cortex, exaggeration of the neuronal injury,and more severe behavioral decits.10Before attribu- of the hand representations in the zone immediatelysurrounding the infarct (Figures 1, 2). Finger represen- ting these results to species differences, it should benoted that several other experimental factors differed tations invaded cortical territory formerly occupied byelbow and shoulder representations. between this study in rats and our studies in monkeys(type of lesion, type and degree of restriction, specic These studies suggest that motor experience follow-ing injury to the motor cortex plays amajor rolein the behavioral paradigm, etc). However, theseresults pointto the need for further studies to address a number of subsequent physiological reorganization that inevi-tably occurs in the adjacent,intact tissue.In the important issues central to the understandingof recov-ery of function. For example, when should physical absence of post-injury rehabilitation, the surroundingtissue undergoes a further functional loss in the areal rehabilitation begin? How intensiveshould the therapybe? Do compensatory strategies developed in the clini- representation of the affected body part. These studiesalso imply theconverse: rehabilitation prevents further cal setting transfer easily to real-life situations? Theseanimal models should stimulate novel hypotheses losses ofhand representation, and may direct theadjacent tissue to take over the damaged function. regardingrecovery of function after cortical injury andthe role that post-injury experience may play in shap- In humans, forced use of the impaired limb may,under a restricted set of circumstances, signicantly ing behavioral outcome.Figure2 Summary of changes in the representation of the hand after damage to primary motor cortex. In control animals, theneurophysiological procedures were conducted, but no lesions were made.News &Views191logical model of the bases of recovery from stroke. ProgIt should not be assumed that neuronal mechanismsBrain Res 1987; 71: 249266.for recovery of motor function exist only in thehealthy2 Kaas JH. Plasticity of sensory and motor maps in adultmotor cortex adjacent to the lesion. It is also possible,mammals. Annu Rev Neurosci 1991; 14: 137167.asothers havesuggested, that other cortical motor areas3 Donoghue JP, Suner S, Sanes JN. Dynamic organization ofin the sameor opposite hemispheremay play a roleinprimary motor cortex output to target muscles in adultrecovery. Still others have suggested that subcorticalrats.II.Rapidreorganizationfollowing motornervestructures are responsible for recovery. However, thelesions. Exp Brain Res 1990; 79: 492503.healthy motor cortex provides a convenient window4 Nudo RJ, Milliken GW, JenkinsWM, Merzenich MM. Use-into the topographic organization of the motor system.dependent alterations of movement representations in pri-Whether similar changes occur in parallel in othermary motor cortex of adult squirrel monkeys. J Neuroscimotor structures remains for future studies.1996; 16: 785807.It is generally assumed that recovery of motor func-5 Nudo RJ, Milliken GW. Reorganization of movement rep-tion can beimproved by various types of physical ther-resentations in primary motor cortex followingfocal isch-apy. Theseresultsdemonstratethat useof theimpaired emic infarcts in adult squirrel monkeys. J Neurophysiol1996; 75: 21442149. limb after a stroke in motor cortex can signicantly6 NudoRJ, WiseBM, SiFuentes F, Milliken GW. Neural sub-inuence the neurophysiological organization of thestrates for the effects of rehabilitative training on motorremaining, intact tissue, and presumably, alter therecovery after ischemic infarct.Science 1996; 272:eventual behavioral outcome. Integrative neuro-17911794.physiological/behavioralapproaches may provide a7 Taub E, Miller NE, Novack TA, Cook EWI, Fleming WC,means to develop and test new therapeutic strategiesNepomuceno CS, ConnellJS, Crago JE. Technique tobased on innate neuronal mechanisms of brain plas-improvechronic motor decit after stroke. Arch Phys Medticity. If the rules governing cortical plasticity in sen-Rehabil 1993; 74: 347354.sory and motor cortex can be generalized to other8 Seitz RJ, HuangY, Knorr U, Tellmann L, HerzogH, Freundcorticalareas, these studies may lead to a generalHJ. Large-scaleplasticity of the human motor cortex. Neu-neurophysiological model of recovery of function. Ifroreport 1995; 6: 742744.such a general model is attainable, it may then be9 KarniA, Meyer G, Jezzard P, Adams MM, Turner R,applicable to less easily understood regions of the cer-Ungerleider LG. Functional MRI evidence for adult motorebral cortex, such as those involved in complex cogni- cortex plasticity duringmotor skill learning. Nature1995;377: 155158. tive functions.10 KozlowskiDA, James DC, Schallert T. Use-dependentexaggeration of neuronal injury after unilateral sensori-RJ Nudomotor cortex lesions. J Neurosci 1996; 16: 47764786.Department of Neurobiology and AnatomyUniversity of Texas Health Science Center at Houston,Houston, TX 77030, USAReferences1 Jenkins WM, Merzenich MM. Reorganization ofneo-cortical representations after brain injury: a neurophysio-