plasticity in cerebellar tactile maps in the adult rat

Plasticity in Cerebellar Tactile Mapsin the Adult Rat

CAROLY A. SHUMWAY,1,2* JOSEE MORISSETTE,3 PETER GRUEN,4

AND JAMES M. BOWER3

1Departments of Conservation and Research, New England Aquarium,Boston, Massachusetts 02110-3399

2Boston University Marine Program, Boston University, Boston, Massachusetts 022153Computation and Neural Systems Program, California Institute of Technology

Pasadena, California 911254USC Healthcare Consultation Center, Los Angeles, California 90033

ABSTRACTWe previously demonstrated that the fractured tactile cerebellar map within the crus IIa

folia of the cerebellar hemispheres reorganizes after deafferentation of the upper lip inneonatal rats (postnatal day [PND] 1–30). The present study examined the capacity of thismap to reorganize after deafferentation in adults and animals late in development (PND30–89). Several months after cauterization of the infraorbital branch of the trigeminal nerve,the tactile map in the granule cell layer of crus IIa reorganized, with representations of intactstructures expanding into the denervated area. The pattern of reorganization was similar toreorganization after neonatal lesions in that (1) all representations were from perioralstructures, (2) the reorganized map maintained a fractured somatotopy, and (3) the dener-vated area was predominantly and consistently invaded by the upper incisor representation.We conclude that the spatial pattern of reorganization is essentially the same regardless of theage of deafferentation. However, we also observed developmental differences in reorganiza-tion. First, more areas of crus IIa were nonresponsive in animals lesioned later indevelopment (PND 30–89). Second, we found a surprising degree of variability in the patternof tactilely evoked cerebellar field potentials of PND 30–40 animals compared with neonatesand adults, suggesting that this time period differs from other stages. The pattern of evokedpotentials reflects the two primary inputs to the map. Our data show that, although bothafferent pathways are capable of reorganization throughout development, their relativecontribution to the map appears to differ, depending on the age at which lesion occurs. J.Comp. Neurol. 413:583–592, 1999. r 1999 Wiley-Liss, Inc.

Indexing terms: cerebellum; mossy fiber; granule cell; somatosensory; field potential

Tactile maps in the granule cell layer of the rat cerebel-lum differ considerably in topography from their afferentstructures. In contrast to somatotopically organized tactilemaps such as those found in the somatosensory cortex(Welker, 1971; Chapin and Lin, 1984), cerebellar tactilemaps have a fractured organization (Shambes et al., 1978;Welker, 1987; Bower and Kassel, 1990). We have demon-strated previously that the fractured somatotopic mapsfound in this region of the cerebellum reorganize in aregular and repeatable manner after peripheral nervelesions in neonatal animals (Gonzalez et al., 1993). In thisstudy, we analyze the pattern of reorganization afterperipheral lesions in animals lesioned later in develop-ment and animals lesioned as adults.

As in our neonatal experiments, the current experi-ments examined the response of the tactile map in the crusIIa region of the cerebellar hemisphere to lesions of theinfraorbital branch of the trigeminal nerve. This nervesupplies much of the innervation of crus IIa in normaladult animals (Bower and Kassel, 1990; Gonzalez et al.,1993). When lesioned adult animals were mapped 2 months

Grant sponsor: NIH; Grant number: NS22205; Grant sponsor: NIMH;Grant number: NRSA award 1F32-MH09849-01.

*Correspondence to: Caroly A. Shumway, Departments of Conservationand Research, New England Aquarium, Central Wharf, Boston, MA02110-3399.

Received 27 August 1996; Revised 18 May 1999; Accepted 13 July 1999

THE JOURNAL OF COMPARATIVE NEUROLOGY 413:583–592 (1999)

r 1999 WILEY-LISS, INC.

later, we found that substantial reorganization of thetactile map had taken place. Although the pattern ofreorganization is in many ways similar to the pattern seenwith neonatal lesions, there were several differences inmap reorganization related to the age of the animal atdeafferentation. The similarity in the overall spatial pat-tern of the maps further supports the idea that thefractured structure of these maps is important to cerebel-lar function.

MATERIALS AND METHODS

Animals used

A total of 35 Sprague-Dawley albino rats were used inthis study: 15 adult control animals and 20 experimentalanimals. The infraorbital branch of the trigeminal nerve ofexperimental animals was lesioned at different times afterbirth, including postnatal day 30 (PND 30, n 5 8), PND 40(n 5 3), PND 77 (n 5 3), PND 80 (n 5 2), PND 85 (n 5 3),and PND 89 (n 5 1). Because not all experiments wereundertaken with each animal, the numbers of animalsused for any given experiment are presented in the resultsand figure legends. All research reported here was carriedout in accordance with the guidelines for animal useestablished by NIH.

Deafferentation

The infraorbital branch of the maxillary division of thetrigeminal nerve on the left side of the face was cauterizedin 1- to 3-month-old rats. Transection of that branch of thetrigeminal nerve removes all sensation from the upper lip,vibrissae, furry buccal pad, and anterior sinus hair on theleft side of the face but does not cause any obvious motordeficits. Nerve cauterization was performed on rats anaes-thetized with chloral hydrate (420 mg/kg). An incision wasmade between the occipital bone ridge and the caudal edgeof the vibrissae pad, and the wound was bathed in 2%lidocaine HCl. The infraorbital branch was then exposedby teasing away surrounding muscle. A cautery unit(Sybron) was used to interrupt the nerve for severalmillimeters. Care was taken to cauterize all of the multiplebranches of the nerve. As an additional precaution againstnerve regeneration, bone wax was inserted under theoccipital bone ridge in the nerve’s previous location. Othersteps taken to address potential regeneration included theelectrophysiological testing of all lesioned animals beforemapping to see whether any response occurred aftercutaneous stimulation of the ipsilateral upper lip or re-lated structures. Any animal with more than 10% ofpenetrations (i.e., six penetrations) responsive to thesestructures was eliminated from analysis. We also exam-ined the nerve of several experimental animals post mor-tem, including all animals in which there was electrophysi-ological evidence of regeneration along with a few in whichthere was not. After another application of the localanaesthetic, 2% lidocaine HCl, the wound was closed withsilk sutures. Postoperatively, animals were monitored forseveral hours until they had appeared to recover com-pletely from the anaesthesia. They were subsequentlyreturned to the Caltech animal care facility.

Animal preparation and electrophysiologicalprocedures

Surgical and tactile mapping procedures were identicalto those described in previous publications (Bower et al.,

1981; Bower and Kassel, 1990; Gonzalez et al., 1993).Briefly, before surgery, each rat was anaesthetized withintraperitoneal injections of sodium pentobarbital (12mg/kg body weight) and ketamine hydrochloride (50 mg/kgbody weight). Throughout the experiment, ketaminesupplements were given as needed to suppress reflexiveactivity. Animals were killed at the end of the experimentwith an overdose of sodium pentobarbital (50 mg/kg bodyweight).

The left cerebellar hemisphere was surgically exposedand covered with mineral oil (see Bower and Woolston,1983; Gonzalez et al., 1993 for further details). Multiunitactivity was recorded in the granule cell layer of crus IIa(400–700 µm below the brain surface) with glass micropi-pettes filled with 2 M NaCl (5–10 µm, 1–3 MV impedance).The central region of the exposed folial crown of crus IIawas finely mapped with 60 sequential perpendicular elec-trode penetrations (three rows of penetrations, 20 punc-tures per row). Depending on surface vasculature, thepenetrations were spaced 100–150 µm apart mediolater-ally and 100–200 µm apart rostrocaudally (see Gonzalez etal., 1993). The location of the electrode penetrations on thesurface of crus IIa was directly recorded on enlargedphotographs at the time of recording (Fig. 1). At eachelectrode penetration, neural signals were displayed on anoscilloscope, made audible by means of speakers followingstandard procedures, and the multiunit receptive field wasdetermined audibly with hand-held glass probes. Twoexperimenters independently rated responses subjectivelyon a scale from 1 (barely detectable) to 5 (maximum).

Fig. 1. Documentation of the position of 60 electrode penetrationswithin crus IIa for a single animal. The three rows of penetration are100–150 µm apart; each row generally contains 20 recording sites. Theseparation between points within a row is 100 µm. Left: anterior; top:medial. This photograph was scanned into a computer, the originalhand-written markings removed, and the penetration points on thisfigure added electronically with Adobe Photoshop.

584 C.A. SHUMWAY ET AL.

Field potential analysis

As in previous experiments (Gonzalez et al., 1993), fieldpotential recordings were made at the site of each elec-trode penetration to further quantify cerebellar responsesto tactile stimulation. After audibly determining the recep-tive field, the center of the receptive field was mechanicallystimulated with the blunt end probe (,1 mm in diameter)of a custom-built tactile stimulator. The stimulus pulseconsisted of a square wave (10- or 50-msec width) gener-ated by an IBM personal computer, with a total probeexcursion of 0.5 mm. The granule cell layer responses tostimulation of the peripheral receptive field were ampli-fied, selectively filtered (field potentials: 1–1,000 Hz band-pass), digitized, and stored on a MassComp 5700 labora-tory computer (Concurrent Computer Inc.) for furtheranalysis.

Subsequent to each experiment, field potential record-ings were carefully examined and measurements of thelatency and amplitude of the waveform components weretaken. Granule cell layer responses to peripheral tactilestimulation consist of two distinct components, one due toa direct mossy fiber projection of the trigeminal sensorynuclei, and the second due to indirect input largely throughthe forebrain (Morissette and Bower, 1996). We havepreviously reported age-related differences in the occur-rence of these two different peaks after neonatal lesions.The same techniques were used here to carefully quantifytheir distribution in animals lesioned late in developmentand adults (Gonzalez et al., 1993).

Map construction

Maps of tactile cerebellar regions were constructed bydrawing enclosed boundaries around adjacent electrodepuncture locations with common strongest receptive fields(Welker, 1987). In cases for which responses were of equalstrength, the boundary line was drawn through the site ofthe electrode penetration.

Statistical analysis of tactile responses

Statistical two-sample comparisons of perioral represen-tations between different experimental groups were con-ducted with a Mann-Whitney U test. Three-sample com-parisons used a factorial analysis of variance (ANOVA),followed by a posteriori Scheffe test. The significance levelwas set at 0.05. Bartlett’s test for homogeneity of variancewas used to compare variances across multiple samples,followed by an a posteriori Tukey-type test. This post hoctest used a logarithmic transformation of sample vari-ances (Zar, 1984). All measures of variability describedherein are standard errors (SE).

RESULTS

Tactile organization in normal adultcerebellum

A number of previous reports have detailed the patternof perioral representations found in crus IIa of the adultrat (Shambes et al., 1978; Welker, 1987; Bower and Kassel,1990). As shown in Figure 2A, the structures innervated bythe infraorbital branch of the trigeminal nerve (i.e., theupper lip, anterior sinus hair, furry buccal pad, and

vibrissae) constitute the largest total representation in thecrown of this folium. This is quantified in Figure 3, wherethe average percent representation of each body surfaceprojecting to crus IIa for 15 normal animals is shown bythe black bars. On average, 69% of the responses in thecrown are from ipsilateral upper lip and related structures(i.e., ipsilateral anterior sinus hair, furry buccal pad, andvibrissae).

Tactile reorganization 2 months afterdeafferentation

Figure 2B–E shows the pattern of tactile inputs to crusIIa 2 months after the infraorbital nerve was lesioned atseveral different postnatal ages, from late in development(PND 30) up to adulthood (PND 77–89). Extensive reorga-nization was found in all animals lesioned, regardless ofthe age at deafferentation. Although there is clear variabil-ity between the individual reorganized maps, close exami-nation reveals some striking similarities. For clarity, in thefollowing sections, we contrast the pattern of reorganiza-tion in these later developmental stages with our earlierstudy of reorganization after neonatal lesion (PND 1–30:Gonzalez et al., 1993).

Patchy organization of reorganized maps. Compari-son of the maps shown in Figure 2 indicates that differentbody surfaces were represented in ‘‘patches’’ within crusIIa both before (A) and after (B–E) peripheral lesions.Furthermore, the general size and distribution of thepatches were similar between normal and lesioned maps.The reorganization of maps after neonatal lesions showeda generally similar patchy structure (see Figs. 2 and 7 inGonzalez et al., 1993).

Change in proportional representation of perioral

surfaces. The body surfaces whose representations in-vaded the denervated region were always perioral regionsnormally represented in crus IIa; however, there was aclear change in the proportional representation of thesenon-upper lip–related structures compared with normaladults. Figure 3 summarizes the results from all threedevelopmental stages studied in this article (PND 30, 40,and adult). The results were pooled, because the propor-tion of the various face areas that invaded the denervatedarea was approximately the same, as judged by a factorialANOVA (not significant) (for individual means, see Table1). The upper incisor became the predominant structurerepresented in lesioned animals, covering on average 44%of the reorganized map, an increase of nearly ninefoldrelative to normal animals (Fig. 3; P 5 0.0001, as judged bya Mann-Whitney U test). A similar anisotropy was shownin the animals lesioned as neonates (Gonzalez et al., 1993).There was also a significant increase in the representationof the lower lip (from 6.5–20%, P 5 0.0001). The contralat-eral upper lip, lower incisor, and nose representation didnot increase significantly.

Increase in the number ofnonresponsive sites

Some electrode penetrations showed no response toperipheral tactile stimulation of any peripheral area (indi-cated with shading in Fig. 2). In normal animals, nonre-sponsive penetrations represented only 2% of the totalpenetrations, on average (Fig. 3). In animals deafferentedlate in development to adulthood, the mean number of

ADULT CEREBELLAR PLASTICITY 585

nonresponsive penetrations significantly increased to 8%(P , 0.05, as judged by a Mann-Whitney U test). Inaddition, a developmental comparison of the percentage ofnonresponsive recordings for animals lesioned from PND 1through PND 89 showed a roughly eightfold increase invariability of response in the animals lesioned late indevelopment to adulthood compared with neonates. Meanand standard errors are as follows: PND 1–16, 0.76 6 0.35;PND 30–40, 7.15 6 2.63; and PND 77–89, 7.83 6 2.42.(PND 1–16 and some PND 30 data used with permissionfrom Gonzalez et al., 1993.) The differences in varianceamong these three groups of lesioned animals are signifi-cant at the 0.001 level, as measured by Bartlett’s test forhomogeneity of variance (Zar, 1984). A Tukey-type mul-tiple comparison test (see Materials and Methods section)showed that the variance of the PND 30–40 animalsdiffered significantly from the PND 1–16 animals, as didthe variance of the adult animals. No significant differ-ence, however, was found between the PND 30–40 and theanimals lesioned as adults.

Variation and distribution of field potentials contain-

ing only the long-latency component. In our previousstudy of neonatal animals, we found that tactilely evokedfield potentials recorded in lesioned animals can some-times differ from those found in intact animals (Gonzalezet al., 1993). In particular, peripheral tactile stimulation inlesioned animals sometimes does not evoke the two-peaked field potential response characteristic of normaladults (lesioned, Fig. 4A; normal, Fig. 4B, also see Moris-sette and Bower, 1996). (The short-latency input reflectsinput from the trigeminocerebellar pathway [Watson andSwitzer, 1978; Woolston et al., 1981], whereas the long-latency input has been shown to primarily originate fromS1 [Morissette and Bower, 1996].) Instead, some fieldpotentials consist of only the second waveform, or long-latency component of the response.

Figure 5 shows maps from four different animals le-sioned at PND 30 with the locations of field potentialscontaining only the second waveform indicated by stars.Note that there appears to be no systematic spatial

Fig. 2. Organization of tactile inputs to the granule cell layers ofcrus IIa for five different rats. A: Tactile map from a normal adultanimal. B–E: Reorganized maps from animals deafferented at differ-ent postnatal days (PND). B, PND 30; C, PND 40; D, PND 77; E, PND85. For this and all subsequent map figures: Left: anterior; top: medial.Filled dots represent the location of the electrode penetrations. Dotted

lines around a patch indicate projections from contralateral struc-tures; solid lines, ipsilateral. Shaded areas indicate cerebellar loca-tions that did not respond to any tactile stimulation (nonresponsive).Ash, anterior sinus hair; Fbp, furry buccal pad; Li, lower incisor; LL,lower lip; N, nose; Ui, upper incisor; UL, upper lip; V, vibrissae.


distribution of these altered field potential responses.Field potentials consisting of just the second peak werefound for every receptive field type, comparable to thepercentage of representation found within the folium(Table 2).

Figure 6 shows that the number of recording sites withfield potentials consisting of only the long-latency compo-nent gradually and significantly increases with the age oflesion up to PND 40 (PND 1–16 and some PND 30 dataused with permission from Gonzalez et al., 1993). Regres-sion analysis was significant, as judged by an ANOVA (P 50.008).

Given these data, we expected that there should be alarge percentage of responses consisting of only the long-latency component in animals lesioned as adults as well.Surprisingly, inclusion of the adult data in the regressionanalysis dramatically changes the slope, resulting in nosignificant relationship with the age of lesion. This changeis due to a significant drop in the number of sites with fieldpotentials containing only the second waveform, or long-latency component in adults. Comparison of the meannumber of sites with these types of field potentials betweenPND 30–40 and adults showed a decline from 17.65 63.09% for PND 30–40 (n 5 11) to 5.71 6 1.36% for PND 77to PND 89 (n 5 6). These results were significant as judgedby a Mann-Whitney U test (P 5 0.02). Comparison of allthree experimental groups, using an ANOVA followed by aScheffe test, indicates that the adult percentage of re-sponses exhibiting the second waveform only was notstatistically distinguishable from the data from neonatallylesioned animals (, PND 17). In contrast, the PND 30–40animals did differ significantly from neonatally lesionedanimals (P , 0.05).

Animals lesioned 30 and 40 days after birth also showedconsiderable variability in the percentage of penetrationswith the long-latency component only (SE above, range,3.45–34.82%). The reorganized maps after adult lesionsshowed less variability, again more closely resemblingmaps from animals whose lesions were made early indevelopment (range for adults, 2.50–10.34%; range forPND 1–16: 0.40–10.60%). The differences in varianceamong these three groups of lesioned animals are signifi-cant at the 0.001 level, as measured by Bartlett’s test forhomogeneity of variance (Zar, 1984). A Tukey-type post hoc

Fig. 3. Comparison of the percentage of representation for perioralsurfaces in tactile maps of normal animals (black bars, n 5 15) and oflesioned animals (white bars, n 5 20). The data from animals lesionedat postnatal day (PND) 30 to PND 90 were pooled, because there wasno significant difference in the body surface representation for theseanimals (see Table 1). Each bar represents the mean percentage 6 SEof total recording sites responding to stimulation of a given face area.Asterisks indicate significant differences in the percentage of represen-tation of a particular face area between normal and deafferentedanimals (as judged by a Mann-Whitney U test, P , 0.01). The upperincisor and lower lip representations were significantly larger inlesioned animals relative to normal animals. IUL, ipsilateral upper lipand related ipsilateral structures (vibrissae, furry buccal pad, andanterior sinus hair); UI, upper incisor; CUL, contralateral upper lipand related contralateral structures (vibrissae, furry buccal pad, andanterior sinus hair); LL, lower lip; LI, lower incisor; N, nose; NR,nonresponsive.

TABLE 1. Map Organization in Normal Rats and Rats Lesionedat Different Postnatal Days (PND)1

Receptivefield type

Percentage of total recording sites

Normal(n 5 15)

Deafferented developmental stage

PND 30(n 5 8)

PND 40(n 5 3)

.PND 75(n 5 9)

IUL 69.28 6 3.53 0.32 6 0.32 0 0.17 6 0.17UI 5.28 6 1.98 44.18 6 3.10 44.94 6 1.89 42.76 6 3.95CUL 8.03 6 1.48 18.54 6 4.26 7.98 6 2.51 13.43 6 3.59LL 6.54 6 1.49 16.13 6 2.69 27.43 6 4.86 21.86 6 2.72LI 8.50 6 1.74 11.37 6 3.45 13.85 6 5.36 13.77 6 2.40N 0.77 6 0.49 1.26 6 0.53 4.72 6 2.40 0.18 6 0.18NR 1.60 6 0.87 8.20 6 2.92 1.08 6 1.08 7.83 6 2.42

1Numbers reflect the mean percent 6 SE of the total recording sites for a given receptivefield type. Abbreviations as in Figure 3.

Fig. 4. Examples of the two different types of field potentialsobserved in lesioned animals. Three superimposed waveforms elicitedupon tactile stimulation are shown for each of the two types of fieldpotential. The first peak, or short-latency component, is denoted by a‘‘1’’; the second peak, or long-latency component, is denoted by a ‘‘2.’’A: Field potentials recorded in location ‘‘A’’ (star) of the map shown inFigure 5D, contain only the second peak or long-latency component ofthe response to tactile stimulation. B: Field potentials recorded at theelectrode location marked ‘‘B’’ (solid dot) of the map shown in Figure5D, exhibit both peaks similar to what is observed in normal animals.


test (see Materials and Methods section) showed that thevariance of the PND 30–40 animals differed significantlyfrom that of both the early and the adult animals. Therewas no significant difference between the variances of theearly and adult animals. The data suggest that PND 30–40animals differ in this aspect of lesion-related reorganiza-tion.

The field potential data can be used to estimate therelative contribution of the different afferent pathways toreorganization after lesion at different developmentalstages, as shown in Figure 7 for total recording sites (A)and upper incisor recording sites only (B). It is clear thateven though the S1 contribution (black bars: second wave-form only) increases with age of lesion up to PND 40, asshown in the previous figure, in all cases, a significantportion of the map receives contributions from both affer-ent pathways (shaded bars).

TABLE 2. Comparison of Receptive Field Type Between All ResponsiveRecording Sites and Those Sites That Exhibited Only the Long-Latency

Component (2nd Peak) of the Tactilely Evoked Field Potentials1

Receptivefield type

Percentage of total recording sites

Allrecording sites

(n 5 17)

Sites with 2ndwaveform only

(n 5 17)

IUL 0.28 6 0.20 0UI 46.78 6 1.74 42.25 6 9.11CUL 16.51 6 2.94 29.84 6 9.83LL 19.73 6 2.07 18.48 6 6.76LI 15.03 6 2.21 3.51 6 3.20N 1.67 6 0.60 5.92 6 3.63

1Numbers reflect the mean percent 6 SE of the total recording sites for a given receptivefield type. The PND 30 to PND 90 animals were pooled together (n 5 17). Abbreviationsas in Figure 3.

Fig. 5. A–D: Tactile maps from four different animals that had theinfraorbital nerve cut 30 days after birth (PND 30). Each map showsthe location and distribution of two different types of field potentialsseen in deafferented animals after tactile stimulation of the face areacorresponding to the receptive field. Solid dots represent field poten-tials similar to those seen in normal animals, with both a short-latencyand a long-latency peak. Stars denote the electrode penetrations for

which the field potential contained only the second waveform, orlong-latency component (shown to originate from the somatosensorycortex, Morissette and Bower, 1996). Open circles show sites where thefield potentials were either not recorded or not analyzable. See Figure4 for examples of the two different types of field potentials, as recordedat the sites marked ‘‘A’’ and ‘‘B’’ in map D. Conventions as in Figure 2.


DISCUSSION

Spatial pattern of reorganization in adultsomatosensory maps

Our results show that fractured somatosensory maps inthe adult cerebellum substantially reorganize severalmonths after peripheral nerve lesion. The spatial patternof reorganization is similar to the pattern after recoveryfrom neonatal lesions in that (1) all representations werefrom perioral structures, (2) the reorganized maps main-tained a fractured somatotopy, and (3) the denervated areawas predominantly and consistently filled in by the upperincisor representation. Previous studies of reorganizationin other somatosensory areas after lesions in adults gener-ally have shown striking similarities to the neonatalpattern as well (somatosensory cortex: Merzenich et al.,1983a,b, 1984; Wall and Cusick, 1984; Calford and Tweed-ale, 1991; thalamus: Garraghty and Kaas, 1991; trigemi-nal nuclei: Waite, 1984; reviews: Snow and Wilson, 1991;Kaas, 1991, 1994).

Developmental differences in mapreorganization

In contrast to the general spatial pattern of the receptivefields, we found a complex relationship between the rela-tive contribution of different inputs to the map and the ageat lesion, as indicated by evoked cerebellar responses. Inother words, the relative proportion of different inputschanges with respect to age. Other studies in the somato-sensory system have shown developmental differences inthe total spatial extent of reorganization dependent on the

age of the lesioned animal, but the studies generally havenot compared the contribution from different sources.Most studies have reported that the spatial extent of adultreorganization is more restricted and variable than thatafter neonatal lesion (Jain et al., 1995). For example, spinalcord or peripheral nerve transection causes extensive re-organization in cortical areas, subcortical areas, or both, inneonates, but limited changes in adult rats (Waite, 1984;McKinley and Smith, 1990). We also found more limitedchanges in adult rats and animals lesioned late in develop-ment. However, a few studies have reported the opposite.For example, Wall and Cusick (1986) report a moreextensive reorganization of the hindpaw region of somato-sensory cortex after nerve section in adults compared witha similar section in neonatal rats (Wall and Cusick, 1984).

We had previously demonstrated that tactilely evokedfield potentials recorded in the granule cell layer of thenormal cerebellum consist of two peaks; one occurringroughly 8 msec; the other, about 18 msec after the onset ofthe peripheral stimulation (Morissette and Bower, 1996).The first peak, or short-latency component, reflects the

Fig. 6. Percentage of electrode penetrations with second waveformonly, or long-latency component, as a function of age at the time ofnerve lesion. Each point displays data from a single animal (opentriangles, normal animals; filled dots, lesioned animals). Each pointrepresents the percentage of the 60 electrode penetrations in whichthe field potential response elicited by tactile stimulation containedonly the second (or long-latency) component originating from thesomatosensory cortex (Morissette and Bower, 1996). Solid lines de-notes regression lines for PND 1–40 (R2 5 0.36) and PND 1–89 (R2 50.0005). The regression is significant for PND1–40 but not for PND1–89, as judged by an analysis of variance (PND 1–40: P 5 0.008; PND1–89: P . 0.05). (PND 1–20 and some PND 30 data were used withpermission from Gonzalez et al., 1993). Fig. 7. Comparison of the contribution of different afferent path-

ways to reorganization, as indicated by the tactilely evoked fieldpotentials. Ordinate is the mean percentage 6 SE of total recordingsites (A) or upper incisor (UI) recording sites (B); the abscissaindicates the age at the time of nerve lesion: PND 1–16 (n 5 8), PND30–40 (n 5 11), and PND 80–89 (n 5 6). The black bars represent thepercentage of the 60 electrode penetrations in which the field potentialresponse elicited by tactile stimulation contained only the second (orlong-latency) component originating from the somatosensory cortex(Morissette and Bower, 1996). The shaded bars represent the percent-age of penetrations in which the field potential response containedboth waveforms. (PND 1–20 and some PND 30 data used withpermission from Gonzalez et al., 1993.)


direct trigeminal input to crus IIa (Watson and Switzer,1978; Woolston et al., 1981), whereas the second peak, orlong-latency component, is primarily due to indirect inputthrough the somatosensory cortex (Morissette and Bower,1996). Our neonatal studies showed that these two compo-nents of the cerebellar granule cell layer field potential areaffected differently after peripheral injury on differentpostnatal days (Gonzalez et al., 1993). An increasingnumber of crus IIa sites lacked the short-latency compo-nent as lesions were made on later postnatal days, up toPND 30. Both the past and present results demonstratethat the largest number of responses consisting of only thelong-latency component occurred in animals lesionedaround PND 30–40. We had previously interpreted thisresult as suggesting that the direct trigeminal-cerebellarpathway became progressively less plastic with age rela-tive to the projection through somatosensory cortex (Gonza-lez et al., 1993).

On the basis of this interpretation, we had expected tofind that the adult lesions (i.e., at PND 77–89) in thecurrent study would produce even more map sites in whichonly the second waveform, or long-latency component, waspresent than the PND 30–40 animals. To our surprise, thiswas not the case. Animals deafferented as adults weremore likely to exhibit responses consisting of just thesecond waveform than normal animals, but had signifi-cantly fewer such sites than animals lesioned betweenPNDs 30 to 40. In fact, adult patterns of reorganizationwere statistically indistinguishable from that of animalslesioned as neonates, with respect to both the mean andvariance of the number of recording sites having theresponse consisting of the second waveform only. It ap-pears as though something about somatosensory mapreorganization may be quite different at PND days 30–40,relative to other stages of development and adults. Inother words, a 30-day brain differs from these other stages,at least with respect to map plasticity.

Recent anatomical studies also indicate that some devel-opmental events occur beyond PND 30. Although mostgranule cells are differentiated by PND 21, their dendritescontinue to grow and differentiate from PND 15–45, andnew postsynaptic elements continue to form until PND 60(Hamori and Somogyi, 1983). Increased expression ofnitric oxide synthase in putative basket and stellate cellsoccurs up to PND 30, with a subsequent 30% decline toadulthood (Li et al., 1997). In one of the afferent struc-tures, the somatosensory cortex, the volume of barrelcortex increases fivefold from PND 5 up to PND 60; thenumber of GABAergic neurons declines between PND5–10, stabilizes until PND 20, and then increases fromPND 20–60 (Micheva and Beaulieu, 1995).

Reorganization in the somatosensory systemas viewed from the cerebellum

Our analysis of the reorganization of cerebellar mapsmay have important implications for interpreting theplasticity seen in other somatosensory regions. Specifi-cally, because the first demonstration of plasticity insomatosensory cortex after peripheral lesions, there hasbeen considerable controversy concerning the exact loca-tion of that plasticity. As we have pointed out previously(Gonzalez et al., 1993), the fact that the spatial organiza-tion of somatosensory cortex and its afferent structures isvery similar makes it inherently difficult to determine theprecise location of any plastic change. For example, an

observed spread of representation of a particular fingercould in principle result from reorganization in any or allof the afferent brain structures or even the periphery(Waite and de Permentier, 1991; Jacquin et al., 1993; Joneset al., 1997). Determining which regions of the brain areresponsible and to what extent, therefore, is experimen-tally difficult, requiring simultaneous recording (e.g., Fag-gin et al., 1997).

Our results indicate that it is possible to infer plasticchanges in different regions of the somatosensory systemby recording in the cerebellum. Since we can clearlydistinguish cerebellar evoked responses arising directly inthe trigeminal system from those projecting through so-matosensory cortex (Morissette and Bower, 1996), we havea means of comparing the plasticity of these differentpathways. Although field potentials recordings have beenused in many previous electrophysiological studies ofcerebellar cortical circuitry (for review see Eccles et al.,1967), the exact sources of the field potentials are stillunclear and are an active area of research (Holt and Koch,1999). However, in our previous studies, we have demon-strated a direct relationship between the locations ofmultiunit granule cell layer responses and the largestamplitude evoked field potentials, which are also recordedin the granule cell layer (Bower, 1981). We have alsoshown, by using both mechanical and pharmacologicallesion techniques (Morissette and Bower, 1996), that dis-ruption of the somatosensory cortex removes both the longlatency field potential and multi-unit responses to periph-eral stimuli. Accordingly, although the exact source of thefield potentials is unclear, we are quite confident in theirability to distinguish between the two major afferentpathways projecting to these cerebellar areas. Further-more, previous studies suggest that both pathways repre-sent information from the same receptive field, in spatialregister with one another (Bower et al., 1981).

Our data suggest that both the trigeminal nuclei and theforebrain pathways are capable of reorganization at allstages of cerebellar development and into adulthood, be-cause the majority of penetrations receive input from bothafferent pathways, although to differing degrees (Fig. 7).Thus, it is likely that changes in both afferent structurescontribute to the reorganization seen in somatosensorycortex, regardless of the animal’s age at lesion. Our studyof reorganization after several months of recovery, there-fore, corroborates the same conclusion derived from arecent study of simultaneous reorganization in the ratcortex, thalamus, and trigeminal brainstem complex afterreversible deactivation (Faggin et al., 1997).

However, our data also suggest that the extent of thiscontribution may vary depending on the time of lesion. Itappears as though there is a developmental period (PND30–40) in which the cerebellar map is less plastic. In otherwords, reorganization in at least the rat’s trigeminalsystem may be considerably more restricted at PND 30–40than earlier or later in life. There is anatomic evidence forthe developmental loss of plasticity in the trigeminalcomplex (Waite and de Permentier, 1991; Jacquin et al.,1993). However, to our knowledge, no previous studieshave specifically compared reorganization in any somato-sensory areas of rats lesioned at PND 30–40 with reorgani-zation in rats lesioned as adults (PND 75–90). We haveconstructed a systems level model of the somatosensorycortex to explore these complex relationships (Morissette,1996). In a forthcoming paper, we consider the possible


mechanisms undergoing reorganization (Shumway et al.,1999).

Functional significance

The current results demonstrate that somatosensorymaps in the adult cerebellum reorganized extensivelyafter peripheral injury. The reorganized maps still main-tain several prominent features of normal maps. First, thereorganized maps still show a mosaic of patches receivingprojection from disjunctive body locations. Second, thereorganized maps generally have a large patch in themiddle of the folium surrounded by smaller patches.Finally, the reorganized maps continue to represent exclu-sively the rat’s perioral regions. In our view, the consisten-cies between reorganized and normal maps regardless ofthe developmental timing of the lesions make it likely thatthe detailed pattern of fractured tactile projections in crusIIa has important functional significance.

Recently, we have proposed that these regions of thecerebellar cortex may be specifically involved in control-ling the active acquisition of somatosensory data (Bower,1997). The fractured somatotopic maps are considered torepresent relationships between the different sensory sur-faces most involved in active sensory exploration in therat. Imaging experiments to test this hypothesis withrespect to the use of fingers in humans have generateddata entirely consistent with this proposal (Gao et al.,1996). Our finding that the upper incisor consistentlyreplaces the lesioned upper lip representation in crus IIa ofthe rat, regardless of the developmental stage at the timeof the lesion is, we believe, also consistent with thisproposal. Because the upper incisor is found in closeperipheral proximity to the inside of the upper lip, theupper incisor representation may convey some informa-tion about upper lip activation in the lesioned animals. Insome, but not all, lesioned animals, we have found that asharp tap on the upper lip in lesioned animals can induce aresponse in the area normally representing the upper lip,through indirect activation of the upper incisor. In theseanimals, if the lip is lifted up to prevent it from touchingthe upper incisor, no response is seen upon upper lipstimulation. Whether or not the upper incisor expansion isreally functionally significant (i.e, whether this activationsufficiently substitutes for the absent upper lip), willrequire analysis of behavioral changes after deafferenta-tion and subsequent reorganization.

ACKNOWLEDGMENTS

We are grateful to Mitra Hartmann for insightful criticalcomments of this manuscript and Erika Oller for her helpdrawing some of the figures. This work was supported byNIH grant N522205 to J.M.B. and NIMH NRSA awardIF32-MH09849-01 to C.A.S.

LITERATURE CITED

Bower JM. 1981. Tactile projections to cerebellar Purkinje cells. Doctoraldissertation, University of Wisconsin at Madison.

Bower JM. 1997. Is the cerebellum sensory for motor’s sake, or motor forsensory’s sake? The view from the whiskers of the rat. Prog Brain Res114:483–516.

Bower JM, Kassel J. 1990. Variability in tactile projection patterns tocerebellar folia crus IIa of the Norway rat. J Comp Neurol 302:768–778.

Bower JM, Woolston DC. 1983. Congruence of spatial organization of tactileprojections to granule cell and Purkinje cell layers of cerebellar

hemispheres of the albino rat: vertical organization of cerebellar cortex.J Neurophysiol 49:745–766.

Bower JM, Beermann DH, Gibson JM, Shambes GM, Welker W. 1981.Principles of organization of a cerebro-cerebellar circuit. Micromappingthe projections from cerebral (S1) to cerebellar. granule cell layer.tactile areas of rats. Brain Behav Evol 18:1–18.

Calford MB, Tweedale R. 1991. Acute changes in cutaneous receptive fieldsin primary somatosensory cortex after digit denervation in adult flyingfox. J Neurophysiol 65:178–187.

Chapin JK, Lin C-S. 1984. Mapping the body representation in the SIcortex of anesthetized and awake rats. J Comp Neurol 229:199–213.

Eccles JC, Ito M, Szentagothai J. 1967. The cerebellum as a neuronalmachine. Berlin: Springer.

Faggin BM, Nguyen KT, Nicolelis MAL. 1997. Immediate and simultaneousreorganization at cortical and subcortical levels of the somatosensorysystem. Proc Natl Acad Sci USA 94:9428–9433.

Gao J-H, Parsons LM, Bower JM, Xiong J, Li J, Fox PT. 1996. Cerebellumimplicated in sensory acquisition and discrimination rather than motorcontrol. Science 272:545–547.

Garraghty PE, Kaas JH. 1991. Functional reorganization in adult monkeythalamus after peripheral nerve injury. Neuroreport 2:747–750.

Gonzalez, L, Shumway C, Morissette J, Bower JM. 1993. Developmentalplasticity in cerebellar tactile maps: fractured maps retain a fracturedorganization. J Comp Neurol 332:487–498.

Hamori J, Somogyi J. 1983. Differentiation of cerebellar mossy fibersynapses in the rat: a quantitative electron microscope study. J CompNeurol 220:365–377.

Holt RH, Koch C. 1999. Electrical interactions via the extracellularpotential near cells bodies. J Comp Neurosci 6:169–184.

Jacquin MF, Renehan WE, Rhoades RW, Panneton WM. 1993. Morphologyand topography of identified primary afferents in trigeminal subnucleiPrincipalis and Oralis. J Neurophysiol 70:1911–1936.

Jain N, Florence SL, Kaas JH. 1995. Limits on plasticity in somatosensorycortex of adult rats: hindlimb cortex is not reactivated after dorsalcolumn section. J Neurophysiol 73:1537–1546.

Jones EG, Manger PR, Woods TM. 1997. Maintenance of a somatotopiccortical map in the face of diminishing thalamocortical inputs. ProcNatl Acad Sci USA 94:1003–1007.

Kaas JH. 1991. Plasticity of sensory and motor maps in adult mammals.Annu Rev Neurosci 14:137–167.

Kaas JH. 1994. The reorganization of sensory and motor maps in adultmammals. In: Gazzaniga MS, editor. The cognitive neurosciences.Cambridge, MA: MIT Press. p 51–71.

Li J, Chen S, Lin RCS, Smith SS. 1997. Cerebellar nitric oxide synthase isexpressed within granule cell patches innervated by specific mossy fiberterminals: a developmental profile. Dev Neurosci 19:274–282.

McKinley PA, Smith JL. 1990. Age-dependent differences in reorganizationof primary somatosensory cortex following low thoracic (T12) spinalcord transection in cats. J Neurosci 10:1429–1443.

Merzenich MM, Kaas JH, Wall JT, Nelson RJ, Sur M, Felleman D. 1983a.Topographic reorganization of somatosensory cortical areas 3b and 1 inadult monkeys following restricted deafferentation. Neuroscience 8:33–55.

Merzenich MM, Kaas JH, Wall JT, Sur M, Nelson RJ, Felleman D. 1983b.Progression of change following median nerve section in the corticalrepresentation of the hand in areas 3b and 1 in adult owl and squirrelmonkeys. Neuroscience 10:639–665.

Merzenich MM, Nelson RJ, Stryker MP, Cynader MS, Schoppmann A, ZookJM. 1984. Somatosensory cortical map changes following digit amputa-tion in adult monkeys. J Comp Neurol 224:591–605.

Micheva KD, Beaulieu C. 1995. Postnatal development of GABA neurons inthe rat somatosensory barrel cortex: a quantitative study. Eur JNeurosci 7:419–430.

Morissette J. 1996. Plasticity in mammalian somatosensory cerebellarmaps. PhD Thesis. Pasadena: California Institute of Technology.

Morissette J, Bower JM. 1996. Contribution of somatosensory cortex toresponses in the rat cerebellar granule cell layer following peripheraltactile stimulation. Exp Brain Res 109:240–250.

Shambes GM, Gibson JM, Welker W. 1978. Fractured somatotopy ingranule cell tactile areas of rat cerebellar hemispheres revealed bymicromapping. Brain Behav Evol 15:94–140.

Shumway C, Morissette J, Bower JM. 1999. Mechanisms underlyingreorganization of fractured tactile cerebellar maps following deafferen-tation in developing and adult rats. J Comp Neurol (in press).


Snow PJ, Wilson P. 1991. Plasticity in the somatosensory system ofdeveloping and mature mammals: the effects of injury to the central andperipheral nervous system. In: Ottoson D, editor. Progress in sensoryphysiology, Vol 11. New York: Springer-Verlag.

Waite PME. 1984. Rearrangement of neuronal responses in the trigeminalsystem of the rat following peripheral nerve section. J Physiol 352:425–445.

Waite PME, de Permentier P. 1991. The rat’s postero-orbital sinus hair: I.Brainstem projections and the effect of infraorbital nerve section atdifferent ages. J Comp Neurol 312:325–340.

Wall JT, Cusick CG. 1984. Cutaneous responsiveness in primary somatosen-sory (S-I) hindpaw cortex before and after partial hindpaw deafferenta-tion in adult rats. J Neurosci 4:1499–1515.

Wall JT, Cusick CG. 1986. The representation of peripheral nerve inputs inthe S-I hindpaw cortex of rats raised with incompletely innervatedhindpaws. J Neurosci 6:1129–1147.

Watson CRR, RC Switzer III. 1978. Trigeminal projections to cerebellartactile areas in the rat originate mainly from N Interpolaris and NPrincipalis. Neurosci Lett 10:77–82.

Welker C. 1971. Microelectrode delineation of fine grain somatotopicorganization of SmI cerebral neocortex in albino rats. Brain Res26:259–275.

Welker W. 1987. Spatial organization of somatosensory projections togranule cell cerebellar cortex: functional and connectional implicationsof fractured somatotopy (summary of Wisconsin studies). In: King JS,editor. New concepts in cerebellar neurobiology. New York: Wiley-Liss. p239–280.

Woolston DC, Kassel J, Gibson JM. 1981. Trigeminocerebellar mossy fiberbranching to granule cell layer patches in the rat cerebellum. Brain Res209:255–269.

Zar JH. 1984. Biostatistical analysis, 2nd ed. Englewood Cliffs, NJ:Prentice-Hall, Inc. p 181–183.


plasticity in cerebellar tactile maps in the adult rat

Documents