Dispatch R57

Cortical plasticity: It’s all the range!Stephen J. Martin and Richard G.M. Morris

When rats learn a motor skill, synaptic potentials in themotor cortex are enhanced. A new study has revealedthat this learning-induced enhancement limits furthersynaptic potentiation, but not synaptic depression.These findings support the view that activity-dependentsynaptic plasticity is the brain’s memory mechanism.

Address: Department of Neuroscience, University of Edinburgh,Crichton Street, Edinburgh EH8 9LE, UK.E-mail: Stephen.Martin@ed.ac.uk

Current Biology 2001, 11:R57–R59

0960-9822/01/$ – see front matter © 2001 Elsevier Science Ltd. All rights reserved.

It is widely believed that a change in the efficacy of infor-mation transmission at synapses in the brain underlies theformation of memories [1]. Until recently, however, a criti-cal line of evidence was missing — evidence that changesin synaptic efficacy occur during learning, and are inducedby an activity-dependent mechanism analogous to thatinvolved in the induction of long-term potentiation(LTP). There have been several recent reports [2–5] thatlearning can result in an enhancement of the evokedresponse in a variety of brain structures. But is an LTP-like mechanism responsible for this enhancement? Rioult-Pedotti et al. [6] have recently used the technique ofocclusion to show that a form of motor cortical plasticityinduced by skill learning really does involve a mechanismsimilar to LTP.

Motor cortical representations are highly plastic, andcapable of substantial functional reorganization [7,8].Animals trained in a skilled reaching task show an expan-sion of the wrist and digit representation in the caudalforelimb area of the primary motor cortex (M1); thisexpansion occurs at the expense of the elbow/shoulderrepresentation, which shrinks with training [9]. Currentevidence suggests that the underlying circuitry necessaryto support such changes is present before motor learningoccurs. Local blockade of GABA-ergic inhibition results ina reorganization of the motor cortical map, an observationthat suggests the existence of a widespread system oflatent horizontal connections whose influence is normallymasked by feed-forward inhibition [10].

Having established a likely substrate for functional corticalreorganization, we now need a mechanism. One way inwhich the transmission of information between neuronsmight be enhanced is by a long-lasting, activity-depen-dent increase in the efficacy of synaptic transmission, suchas LTP. However, neural network modelling studies

suggest that the opposite phenomenon, long-term depres-sion of synaptic efficacy (LTD), acting in concert withLTP, may also be critical for efficient memory storage[11]. Both LTP and LTD can be induced in thelayer II/III horizontal connections of motor cortical slices[12,13]. The experimental protocol for LTP inductioninvolves focal application of the GABAA antagonist bicu-culline, followed by a series of trains of high frequencyelectrical stimulation. This form of LTP is blocked byNMDA receptor antagonists [14]. LTD can be inducedsimply by low frequency stimulation of the horizontal con-nections [12].

In a previous study, Rioult-Pedotti et al. [15] had trainedrats in a skilled reaching task and then measured the sizeof stimulus-evoked potentials ex vivo in the forelimbrepresentation area of M1 slices. Rats learned to reachthrough a hole in a small plastic box with their preferredpaw, and to grasp and retrieve small food pellets. Afterthree to five daily practice sessions — all that was neces-sary to see a striking improvement in the rats’ skill inretrieving pellets — the strength of synaptic connectionsamong M1 neurons was increased. Coronal brain slicescontaining both hemispheres were prepared, and stimulat-ing and recording electrodes were placed bilaterally inlayer II/III of the forelimb area of M1 (Figure 1a). Mostmotor cortical neurons lie contralateral to the limb thatthey control; therefore the hemisphere contralateral to thepreferred forelimb is termed the ‘trained’ hemisphere,whereas the ipsilateral hemisphere is referred to as‘untrained’, and serves as a within-subject control.

The first key finding was that, in rats that had acquiredthe motor skill, evoked potentials were as much as 50%larger in the trained, relative to the untrained, hemi-sphere. This increase did not occur if NMDA receptorswere blocked [16]. The second key finding was that thelearning-induced enhancement of the evoked potentialwas associated with a partial occlusion of artificiallyinduced LTP. As LTP can be saturated by repeated trainsof inducing stimulation, the implication is that prior skilllearning itself employs an LTP-like mechanism to ‘use up’some of the plasticity available at motor cortical synapses.

Rioult-Pedotti and colleagues’ new paper [6] puts themicroscope on this altered capacity for further plasticityafter learning. Is synaptic plasticity shut down by learning?Or might there be a ‘synaptic modification range’ overwhich synaptic efficacy can vary, with learning simplypushing synaptic strength one way or the other? A way ofdistinguishing these alternatives is to look at LTD. Many

studies have shown that, following LTP induction, it ispossible to depress or ‘depotentiate’ the newly elevatedlevel of synaptic efficacy by means of long trains of low-frequency stimulation [17]. It follows that, if skill learninghas moved synaptic efficacy upwards within the synapticmodification range, then LTD should be enhanced. Thisis exactly what the new paper reports [6]. As in the previ-ous study, synaptic plasticity was examined in the fore-limb region of M1 in cortical slices taken from rats trainedfor five days on the skilled reaching task. Whereasrepeated attempts to induce LTP resulted in a smallerpotentiation in the trained versus the untrained hemi-sphere, a larger amount of LTD could be induced.

As Rioult-Peddotti et al. [6] point out, these findingspresent an immediate puzzle. If the learning of one skilluses up almost all of the available capacity for synapticenhancement, how can additional skills ever be learned?If moving synaptic efficacy within a predeterminedrange were the only neural mechanism used to induceand store memory traces, one might reasonably predict

that learning one skilled reaching task would impair thelearning of another skill. As the authors note [6], such aresult has been reported in humans [18], but a direct testof this prediction has not yet been reported usinganimals, in which the underlying mechanisms might beprobed electrophysiologically.

The results of such an experiment might, however, beinconclusive: the types of skill learned, and the neuronalpopulations used, are likely to be critical in determining theoutcome. For instance, if two skills ‘compete’ for synapticplasticity within overlapping populations of synapses, thereis likely to be interference, but if a second skill can belearned by ‘borrowing’ elements of the first skill, such asspecific sequences of limb movements, but otherwise usesa distinct pool of synapses, learning of the second skillmay even be facilitated. Moreover, the time period duringwhich interference can occur might be limited by the fine-tuning of relevant synaptic connections that accompaniesovertraining on a skilled task, and/or by the induction ofsynaptic growth. Either or both processes might result in a

R58 Current Biology Vol 11 No 2

Figure 1

(a) Configuration of stimulating and recordingelectrodes in cortical slices containing M1.Electrodes were placed symmetrically in theforelimb representation area of thehemisphere contralateral to the preferredforelimb (trained hemisphere), and ipsilateralto the preferred forelimb (untrainedhemisphere). Key: wm, white matter. (b) After3–5 days of skill learning, evoked responseswere larger in the trained hemisphere than inthe untrained hemisphere. Saturation of eitherLTP or LTD revealed a reduced capacity forLTP and an enhanced capacity for LTD in thetrained hemisphere. This is consistent with thenotion of a synaptic modification range whoseupper and lower limits remain fixed, despitethe fact that baseline evoked responses arelarger after training. The fact that learning-induced LTP occludes tetanus-induced LTPsuggests that the two phenomena sharecommon mechanisms. (Adapted withpermission from 15].)

2 mm

wmwm

M1M1

StimulateRecord

StimulateRecord

Untrained hemisphere

Trained hemisphere Untrained hemisphere

Synapticmodificationrange

(a) Electrode configuration in cortical slices containing M1

Syn

aptic

str

engt

h +

–

Saturated LTP

Saturated LTD

Baseline

Saturated LTP

Saturated LTD

Baseline

Syn

aptic

str

engt

h +

Learning-inducedLTP

Capacityfor tetanus-inducedLTP / LTD

Trained hemisphere

–

(b) Increase in synaptic strength within a fixed modification range after learning

Current Biology

shift in the synaptic modification range — towards ahigher floor and/or a higher ceiling.

Evidence that a shift in both floor and ceiling occursfollowing an extended period of training was reportedby Rioult-Pedotti et al. [19] at this year’s Society forNeuroscience meeting in New Orleans. A characteristic ofmotor memories is that they undergo a process of consolida-tion over time. Such memories continue to develop evenin the absence of further training, and progress from aninitially labile condition to a more durable, permanentform [18]. The increase in the floor of the synaptic modifi-cation range may represent just such a process of consoli-dation, with existing patterns of synaptic enhancementbecoming permanently established over time. At the sametime, new synapses may grow, raising the ceiling of themodification range. New motor learning, and with it newincreases in synaptic efficacy, may then be possible withno discernible interference between newly acquired andwell established skills.

A separate issue concerns the nature of the informationrepresented by a learning-induced increase in the efficacyof a large number of cortical synapses. The learning ofwhat appears to be a fairly modest motor skill — reachingthrough a hole — causes a surprisingly large and long-lasting increase in the layer II/III evoked response. Mindyou, learning to reach for food pellets is likely to be amajor event in the life of a hungry laboratory rat. The realissue is not so much the size of the evoked response as thespatial distribution of synaptic changes, and we currentlyhave no idea how increases in synaptic efficacy among thehorizontal connections of the forelimb region of M1 canencode a complex spatiotemporal sequence of move-ments. The same can be said of the expansion of the fore-limb representation region of the motor cortical map. Thenub of the representational question is whether theprecise pattern of changes in synaptic strengths consti-tutes an engram of the motor program for the execution ofthe task, or whether such changes have some ancillaryinformation processing role.

We have argued recently that the generic ‘synapticplasticity and memory’ hypothesis has to satisfy fourformal experimental criteria, one of which is ‘detectability’— the idea that changes in synaptic efficacy should bedetectable somewhere in the brain as learning occurs [1].The work of Rioult-Pedotti et al. [6] complements recentamygdalar studies [4,5], and suggests that meeting thiscriterion is feasible. The observation that skill learning isassociated with LTP of layer II/III horizontal connectionswithin M1, and an augmented capacity for the inductionof LTD within an unchanged synaptic modification range,provides strong evidence that the functional reorganiza-tion of this region during learning depends on establishedprinciples of activity-dependent synaptic plasticity. Many

loose ends remain, but an understanding of the neuralmechanisms of memory seems to be within range.

References1. Martin SJ, Grimwood PD, Morris RGM: Synaptic plasticity and

memory: an evaluation of the hypothesis. Annu Rev Neurosci 2000,23:649-711.

2. Moser E, Moser MB, Andersen P: Synaptic potentiation in the ratdentate gyrus during exploratory learning. NeuroReport 1993,5:317-320.

3. Roman FS, Truchet B, Marchetti E, Chaillan FA, Soumireu-Mourat B:Correlations between electrophysiological observations ofsynaptic plasticity modifications and behavioral performance inmammals. Prog Neurobiol 1999, 58:61-87.

4. Rogan MT, Stäubli UV, LeDoux JE: Fear conditioning inducesassociative long-term potentiation in the amygdala. Nature 1997,390:604-607.

5. McKernan MG, Shinnick-Gallagher P: Fear conditioning induces alasting potentiation of synaptic currents in vitro. Nature 1997,390:607-611.

6. Rioult-Pedotti M-S, Friedman D, Donoghue JP: Learning-inducedLTP in neocortex. Science 2000, 290: 533-536.

7. Buonomano DV, Merzenich MM: Cortical plasticity: from synapsesto maps. Annu Rev Neurosci 1998, 21:149-186.

8. Sanes JN, Donoghue JP: Plasticity and primary motor cortex. AnnuRev Neurosci 2000, 23:393-415.

9. Kleim JA, Barbay S, Nudo RJ: Functional reorganization of the ratmotor cortex following motor skill learning. J Neurophysiol 1998,80:3321-3325.

10. Jacobs KM, Donoghue JP: Reshaping the cortical motor map byunmasking latent intracortical connections. Science 1991,251:944-947.

11. Willshaw D, Dayan P: Optimal plasticity from matrix memories:what goes up must come down. Neural Computation 1990,2:85-93.

12. Hess G, Donoghue JP: Long-term potentiation of horizontalconnections provides a mechanism to reorganize cortical motormaps. J Neurophysiol 1994, 71:2543-2547.

13. Hess G, Donoghue JP: Long-term depression of horizontalconnections in rat motor cortex. Eur J Neurosci 1996, 8:658-665.

14. Hess G, Aizenman CD, Donoghue JP: Conditions for the inductionof long-term potentiation in layer II/III horizontal connections ofthe rat motor cortex. J Neurophysiol 1996, 75:1765-1778.

15. Rioult-Pedotti M-S, Friedman D, Hess G, Donoghue JP:Strengthening of horizontal cortical connections following skilllearning. Nat Neurosci 1998, 1:230-234.

16. Margolis DJ, Donoghue JP, Rioult M-G, Rioult-Pedotti, M-S: Role ofNMDA receptors in skill learning and learning-induced synapticstrengthening. Soc Neurosci Abstr 1999, 25:888.

17. Wagner JJ, Alger BE: Homosynaptic LTD and depotentiation: dothey differ in name only? Hippocampus 1996, 6:24-29

18. Shadmehr R, Brashers-Krug T: Functional stages in the formationof human long-term motor memory. J Neurosci 1997, 17:409-419.

19. Rioult-Pedotti M-S, Donoghue JP: Neocortical learning regulatessynaptic strength by altering the synaptic modification range. SocNeurosci Abstr 2000, 26:194.

Dispatch R59