| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Computation and Neural Systems Program, California Institute of Technology, Pasadena, California
Submitted 15 February 2005; accepted in final form 21 June 2005
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
). However, both the mechanisms responsible for reorganization and the site of reorganization remain subjects of considerable debate (reviews: Florence et al. 1997; Fox 2002; Kaas et al. 1999; Wall 1988). For example, despite extensive studies of the somatosensory region (S1) of cerebral cortex, it is unclear if lesion-induced S1 reorganization is caused by intrinsic cortical mechanisms, such as the immediate unmasking of "silent" projections and/or the sprouting of intracortical connections, or extrinsic effects, such as the reorganization of afferent somatosensory structures (reviews: Irvine and Rajan 1996; Jones 2000; Kaas 1991).
One confound is that S1 and its afferent structures are topographically organized. Because denervated areas of S1 are "filled-in" by adjacent representations (review: Kaas 1991), it is difficult to distinguish between intrinsic cortical reorganization and extrinsic reorganization of afferent structures, unless S1 and its afferents are examined simultaneously (Faggin et al. 1997; Lane et al. 1995). This problem does not exist for tactile cerebellar maps. While cerebellar afferent structures, like SI, trigeminal nucleus and the superior colliculus, are topographically organized (Chapin and Lin 1984; Welker 1971, 1987), cerebellar maps are fractured, with adjacent peripheral structures not necessarily adjacent within cerebellar cortex (Bower and Kassel 1990). This difference between intrinsic and afferent topography should make distinguishing between the possibility of intrinsic and extrinsic mechanisms of reorganization easier for cerebellar maps.
Previously, we have taken advantage of the fractured cerebellar somatotopy to better understand the reorganization of cerebellar maps following peripheral lesions (neonates: Gonzalez et al. 1993; Morissette 1996; adults: Shumway et al. 1999). We showed that lesions of the infraorbital branch of the trigeminal nerve result in an expanded representation of the upper incisor into the denervated upper lip–related representation in the center of folium crusIIa. This finding was surprising because micromapping of the entire crus IIa folium in normal rats showed that the upper incisor is minimally represented and varies in position, not necessarily adjacent to the upper lip representation it replaces after lesions (Bower and Kassel 1990).
The purpose of this study was to better understand the mechanism of cerebellar map reorganization by assessing the possible contributions of intrinsic and extrinsic influences. We first analyzed normal cerebellar maps in detail. We compared the location and frequency of adjacency of the upper incisor representation and other perioral structures relative to the upper lip, as well as the size of a given patch and extent of representation. We then explored the possibility of unmasking, by mapping crus IIA immediately before and after peripheral nerve lesion. Finally, we compared reorganized maps in the cerebellum with those in the S1 cortex from the same animals after peripheral lesion. Lesions were made at developmental stages from PND 1 to adult, and the regions were mapped 2–3 mo later. The results suggest that change in cerebellar maps reflects afferent and not intrinsic reorganization.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
) were also reanalyzed here (Figs. 4B and 5B only). Because not all experiments were undertaken with each animal, the numbers of animals used for any given experiment are presented in the results and figure legends. This research was carried out in accordance with the guidelines for animal use established by National Institutes of Health and was preapproved by the Caltech Institutional Animal Use Committee.
|
|
Nerve transection was performed on rats anesthetized either with avertine (125 mg/kg) or chloral hydrate (420 mg/kg). First, an incision was made between the occipital bone ridge and the caudal edge of the vibrissae pad. The infraorbital branch was exposed by teasing away surrounding muscle and cut. A cautery unit (Sybron) was used to interrupt the nerve for several millimeters. To prevent regeneration, care was taken to cauterize all of the multiple branches of the nerve, and bone wax was inserted under the occipital bone ridge in the nerve's previous location. Other steps taken to address the potential problem of regeneration included the electrophysiological testing of all animals to see if any response occurred after cutaneous stimulation of the ipsilateral upper lip or related structures. Regeneration was determined to have occurred if an animal had >10% of penetrations (i.e., 6 penetrations) responsive to these structures; these animals were eliminated from analysis. We also examined the nerve of several experimental animals postmortem (both those lesioned neonatally and those lesioned as adults): this included all animals in which there was electrophysiological indication of regeneration as well as a few in which there was no physiological evidence of regrowth. After application of a local anesthetic, 2% lidocaine HCl, the wound was closed with suture. The animals were monitored for several hours until they recovered from the anesthetic. They were subsequently returned to the animal care facility. Young rats were placed back with their mothers.
Receptive field mapping of lesioned rats after recovery
Both crus IIa and S1 cortex (for a subset of animals) were mapped in experimental animals following a 2- to 3-mo postlesion recovery period. Twelve intact, normal rats were used as controls. Surgical and tactile mapping procedures were identical to those used in previous studies (Bower and Woolston 1983; Gonzalez et al. 1993). Before surgery, each rat was anesthetized with intraperitoneal injections of pentobarbital sodium (12 mg/kg) and ketamine hydrochloride (50 mg/kg). Throughout the experiment, ketamine supplements were given as needed. Animals were killed at the end of the experiment with an overdose of pentobarbital sodium (50 mg/kg, ip).
In all animals, the left cerebellar cortex, (and in the subset of animals) right cerebral cortex, were surgically exposed, covered with mineral oil, and photographed. Multi-unit recordings were taken in the granule cell layer of crus IIa (400–700 µm below the brain surface) and layer IV of S1 cortex (500–1,100 µm). The 400- to 700-µm variation in the cerebellum was caused by the curvature of the granule cell layer across the surface of the folial crown. Multi-unit activity was recorded with glass micropipettes filled with 2 M NaCl (10 µm diam, 1-M
impedance). The central region of the exposed folial crown of crus IIa was finely mapped with 60 sequential perpendicular electrode penetrations (3 tracts, 20 punctures per tract). The S1 cortex was mapped with as many penetrations as required to determine the areal extent of the upper incisor representation. The location of the penetrations on the surface of crus IIa and the S1 cortex were directly recorded on enlarged photographs at the time of recording. In crus IIa, depending on surface vasculature, penetrations were spaced 100–150 µm apart mediolaterally and 100–200 µm apart rostro-caudally (see Gonzalez et al. 1993). In the S1 cortex, penetrations were spaced 100–200 µm apart in each direction. Near the apparent border of the upper incisor representation, penetrations were generally spaced 100 µm apart. For each electrode penetration, the multi-unit receptive field was determined audibly with handheld glass probes and observed visually with an oscilloscope. At least two experimenters independently rated responses subjectively on a scale from 1 (barely detectable) to 5 (maximal). The strength of the response was used in determining dominant and subdominant response patterns (strongest and second strongest, respectively), and in constructing tactile maps (see Map construction).
Receptive field mapping immediately before and after lesion
In three of the normal animals, the central region of crus IIa was mapped in detail. Then the infraorbital nerve was cut. Animals were anesthetized and prepared as described above. A topical application of 2% lidocaine was applied to the skin over the infraorbital nerve, and the infraorbital nerve was cauterized. The wound was bathed with 2% lidocaine, and the skin incision closed with suture. The area was immediately remapped, which took 
6 h. We did not notice any change in excitability during the hours of remapping, as measured qualitatively from auditory determination of response strength. This suggests that residual lidocaine effects of the cauterized nerve were minimal.
Electrode penetrations after deafferentation were positioned in the same location as the original map with the aid of the original coordinates and the location of the original penetrations relative to the surface vasculature on the photograph.
Map construction
As in previous experiments (Gonzalez et al. 1993), cerebellar and cerebral cortical maps were constructed by drawing enclosed boundaries around adjacent electrode puncture locations whose receptive fields were from the same body structure. When the neurons at a given penetration could be excited by stimulation of different perioral structures, the structure eliciting the strongest response was used (dominant response). In cases where responses were of equal strength, the boundary line was drawn through the site of the electrode penetration. Subdominant receptive fields were measured in the same manner. Note that these multiple responses at a given penetration necessitated that total number of responses per animal be used in normalizing across animals, not total number of penetrations. Frequency of adjacency to the ipsilateral upper lip and associated structures was calculated as follows: (number of patches for any given receptive field/total number of adjacent patches) x 100.
Field potential analysis
After qualitative determination of the strength of all possible receptive fields at a given electrode penetration by listening to the auditory monitor, we quantitatively determined the source of afferent input for the dominant response(s) with field potential recordings. Normal field potentials consist of two components; the first, short-latency component (10–22 ms) originating from the trigeminal complex (Watson and Switzer 1978; Woolsten et al. 1981), and the second, longer-latency component (21–37 ms) originating from S1 cortex (Morissette and Bower 1996). The center of the dominant receptive field was mechanically stimulated with the blunt end probe (<1 mm diam) of a custom-built tactile stimulator. The stimulus pulse consisted of a square wave (10- or 50-ms width) generated by an IBM personal computer, with a total probe excursion of 0.5 mm. The field potential response recorded from the granule cell layer (400–700 µm from the cortical surface) was preamplified at a gain of 600 (WPI preamplifier), filtered (high-pass 1 Hz, low-pass 1 kHz), digitized, and stored on a MassComp 5700 laboratory computer (Concurrent Computer) for further analysis. After the experiment, the latency and amplitude of the waveform components were taken, and the presence of one or both components of the field potential were noted. For this paper, only the second waveform was considered.
Statistical analysis of tactile responses
Statistical two-sample comparisons of perioral representations between different experimental groups were conducted with a Mann-Whitney U-test; three-sample comparisons used a factorial ANOVA, followed by an a posteriori Scheffé test. Multiple comparisons of receptive fields within maps were conducted with a one-way repeated-measures ANOVA, followed by a Scheffé test. The 
level was set at 0.05. All measures of variability reported herein are SE.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
TACTILE ORGANIZATION IN NORMAL ADULT CEREBELLUM.
In previous lesion studies, we reported that the upper incisor representation consistently replaces the denervated upper lip representation in crus IIa after peripheral lesions (Gonzalez et al. 1993; Morissette 1996; Shumway et al. 1999). These papers also showed, using evoked potentials, that the reorganization involves both cerebral cortical input as well as input from the sensory trigeminal nucleus (for a detailed analysis of the cerebellar evoked potentials, see Morissette and Bower 1996). As a first step toward understanding the mechanisms underlying the predominant expansion of this structure in crus IIa of deafferented animals, we compared the upper incisor representation in maps of normal animals with that of other perioral structures represented in crus IIa. We took this approach, because previous studies of the normal pattern of organization in somatosensory cortex have been helpful in understanding the pattern of reorganization in the cortex after lesion-induced change. For example, Schroeder et al. (1997) used a variety of electrophysiological techniques to show that the radial nerve provides latent, subdominant input to the medial cortical area in normal animals. By understanding the normal subdominant input, they were able to provide a convincing justification for why the radial, and not the ulnar, representation expands after median nerve lesions.
ADJACENCY TO IPSILATERAL UPPER LIP.
Plasticity studies in other parts of the somatosensory system have shown that denervated areas are generally filled in by adjacent representations (Wall and Cusick 1984; for review, see Kaas and Florence 1993). In this analysis, we examined whether a similar mechanism might account for the expansion of the upper incisor representation in crus IIa by quantifying the frequency of adjacency of different receptive field types to the central upper lip representation in normal animals (see METHODS). If the upper incisor representation in normal animals was found to be commonly adjacent to the upper lip representation in crus IIa, this would be supportive of an intrinsic mechanism of reorganization, i.e., reorganization occurring within the cerebellum itself, as reported for other somatosensory regions. Detailed micromapping of the entire normal crus IIa map by Bower and Kassel (1990) suggested, however, that the upper incisor is minimally represented in the folium, often not adjacent to the upper lip representation, and tends to be variable in its location. In fact, that study showed that, in three of five complete maps of crus IIa, no upper incisor representation was found.
Figure 1A quantifies the frequency of adjacency for the different receptive field types in the center of the folium for the 12 normal animals mapped as part of this study. After a repeated-measures ANOVA, a conservative post hoc Scheffé test did not indicate any significant difference among pairs of perioral structures (P > 0.05). In other words, the upper incisor representation was as frequently adjacent to the ipsilateral upper lip representation as other perioral structures. A comparison of the size of patches adjacent to the ipsilateral upper lip representation (Fig. 1B) also showed no significant post hoc comparisons among perioral structures (P > 0.05).
|
; Welker 1987), the largest representation in the granule cell layer of crus IIa is that of the upper lip and its related structures (furry buccal pad, anterior sinus hair, and vibrissae). For dominant responses, the extent of this representation differed significantly from all of the nonupper lip perioral structures (all post hoc pairs, P = 0.0001). Figure 2 also shows that all nonupper lip perioral structures were represented in equal proportion (P > 0.05, not significant), with the exception of the nose, which was found rarely.
|
; Shambes et al. 1978a,b; Welker 1987), we have previously shown that weaker, or "subdominant" multi-unit responses can often also be elicited, at least with the multicellular electrodes used in this and earlier studies, as well as with evoked potential analyses (Gonzalez et al. 1993; Morissette and Bower 1996). We explored both the dominant and weaker, subdominant responses in these experiments because previous work in somatosensory cortex has implicated subdominant responses in unmasking effects (Garraghty and Sur 1990). On average, 72% of the responsive penetrations had subdominant receptive fields. Of the penetrations with receptive fields innervated by the infraorbital branch of the trigeminal nerve (i.e., upper lip and related structures), 51% had non–infraorbital-related subdominant receptive fields. Could some of these subdominant responses be caused by passive electrical spread of granule cell activity? Forty-one percent of the subdominant receptive fields were immediately adjacent to locations with dominant responses of the same receptive field type and thus could very well have been recorded because of signal spread. The majority of subdominant responses (59%), however, seem to represent actual weak afferent projections.
Overall, no differences were found in the general pattern of distribution of the subdominant and dominant representations. As indicated by the gray bars in Fig. 2, the ipsilateral upper lip was the most common subdominant receptive field type, just as was found for the dominant representation. Similarly, the extent of the upper incisor subdominant representation did not differ significantly from that of any other perioral structure. All of the other perioral structures, including the upper incisor, were represented as subdominant receptive fields roughly in equal proportion (P > 0.05, not significant), with the exception of the nose which again was much rarer. The only pairwise statistical difference between any representation was between the ipsilateral upper lip representation and the nose (P = 0.02 using a repeated-measures ANOVA followed by a Scheffé test).
CEREBELLAR REORGANIZATION IMMEDIATELY AFTER DEAFFERENTATION. Having failed to find any difference between the upper incisor and the other non-upper lip receptive fields in normal maps, we next sought to determine whether the dominance of this structure in the reorganization in lesioned animals could be caused by an immediate unmasking of weak (subdominant) and/or previously silent inputs to crus IIa. To examine this question, we recorded responses before and immediately after deafferentation of the infraorbital branch of the trigeminal nerve (n = 3). We observed profound effects on the pattern of representation in the cerebellar tactile maps, as would be expected to result from removing one of the major tactile inputs to this region. While the pattern changed, it is important to note that the excitatory responses elicited after deafferentation (during the 6 h of recording) were generally of the same strength as those found before deafferentation, as measured qualitatively from auditory determination of response strength.
Figure 3 shows examples of the tactile maps obtained in crus IIa in these experiments. The maps in Fig. 3A are the standard crus IIa maps constructed from the perioral structures eliciting the strongest dominant response; the corresponding maps in Fig. 3B represent the strongest non–infraorbital-related subdominant responses. The maps in Fig. 3C represent the dominant responses remapped immediately after deafferentation. Shaded areas in Fig. 3C indicate the map regions in which no response could be recorded immediately after deafferentation. The thick black line in all maps indicates the area normally innervated by the infraorbital branch of the trigeminal nerve.
|
Are the new representations within the denervated area indicative of immediate unmasking of previously silent receptive fields or could they originate from surrounding receptive fields? Analysis of the representations adjacent to the 18 new responses in the middle track within the denervated area in the maps showed that the majority of the new responses (15/18; 83.3%) can be explained by their adjacency to subdominant or dominant receptive fields. Thus only 7.7% (3/39) of the analyzed responses within the denervated area are left as candidates for an immediate unmasking of previously silent representations.
The most important point is that the upper incisor representation did not occur with any greater frequency than the other nonupper lip perioral regions immediately after lesion. In fact, a comparison of receptive fields in the three animals recorded before and immediately after deafferentation indicates similar frequency distributions. Before lesion, the ipsilateral upper lip predominated (Fig. 4A, stippled bars; P = 0.0001, repeated-measures ANOVA), and all nonupper lip perioral structures were represented in roughly equal proportion, except for the nose. A Scheffé test indicated no significant differences among the nonupper lip structures. Immediately after lesion (black bars), the mean frequency of representation among the remaining perioral structures also was not significantly different, with the exception of a comparison between the lower incisor and the nose (P = 0.01, repeated-measures ANOVA). We also found no significant difference among nonperioral structures with a different analysis: computing the difference score for each receptive field before and after lesion and comparing the difference score among structures (data not shown, not significant). Comparison within any single receptive field type between the three animals before and immediately after lesion indicated that the only significant difference (beyond the obvious loss of the upper lip and related input) is a significant increase in the percentage of nonresponsive points, from 1 to 30%, as judged by a factorial ANOVA, followed by a Scheffé test (P = 0.0002).
CHANGE IN CEREBELLAR REPRESENTATION OF BODY PARTS OVER TIME.
Map pattern.
As shown in the summary histogram in Fig. 4B, the pattern of tactile activity in crus IIa immediately after deafferentation in adult rats does not predict the pattern found 2–3 mo later, as reflected in a repeated-measures ANOVA across receptive field types. Immediately after deafferentation (black bars), the upper incisor comprised just 17% of the map, similar to that of the other perioral structures (i.e., no significant difference among receptive field types). Thirty percent of the map was nonresponsive. In contrast, 2 mo later (striped bars), the upper incisor representation predominated, extending to 43% of the map, and fewer penetrations were nonresponsive (8%). For these recovered adults, the mean frequency of representation of the upper incisor differed significantly from all other receptive field types, as judged by a repeated-measures ANOVA (P = 0.0001), followed by a Scheffé test; also, the lower lip differed significantly from the nose representation. The same pattern of reorganization observed in recovered adults was found in developing animals (white bars, P = 0.0001), with the upper incisor representation differing significantly from all other receptive field types, and both the lower lip and contralateral upper lip representation differing significantly from the nose representation.
Comparison of any single receptive field type among the four different experimental groups represented in Fig. 4 showed that the more than two-fold increase in upper incisor representation between animals examined immediately after lesion and those several months later (both adults and neonates) was significant, as judged by a factorial ANOVA (P = 0.0001), followed by a Scheffé test. In addition, a significant difference was found in the extent of upper incisor representation between animals examined before lesion and recovered animals (both adults and neonates; P = 0.0001) and in the extent of lower lip representation between animals examined before and recovered adults (P = 0.037). The decrease in nonresponsiveness between animals examined immediately after lesion and recovered animals (both adults and neonates, in separate pairwise comparisons) also was significant (P = 0.0002). There was no significant change in areal extent of the other perioral structures among these groups.
Patch size.
Fractured cerebellar maps have multiple representations (i.e., patches) of different perioral structures. Within our recording areas, a given map usually had 7–15 patches. Figure 5 compares the representational area of these patches among the different experimental groups and normal animals, using the number of electrode penetrations necessary to define a patch as the areal measure.
In normal animals (Fig. 5A), the majority of patches detected as a dominant response (81%) were equal to or larger than three electrode penetrations. Only 9% of the patches comprised just one electrode penetration, and 10% comprised two. The subdominant receptive fields appeared to be organized in a finer-grained fractured pattern: 49% comprised three or more, whereas 32% of the subdominant responses comprised one electrode penetration and 19% had two penetrations. Statistical comparison indicated significant differences between dominant and subdominant patches for patch sizes comprising one penetration (P = 0.0001) as well as three or more (P = 0.0001).
Patch size varied between animals examined immediately after lesion and those examined several months later (Fig. 5B). In adult rats examined immediately after deafferentation (white bars), a diversity of patch sizes was found: 62% of the patches comprised three or more electrode penetrations, whereas 18% of the receptive fields comprised one electrode penetration, and 20% had two penetrations. In contrast, the pattern of patches in adult rats examined 2 mo later (black bars) more closely resembled the dominant receptive field maps of normal animals—that of a large patch surrounded by a few smaller patches (cf. Fig. 5, B and A). The majority (84%) of the responses comprised three or more electrode penetrations, and just 8 and 8% of the receptive fields comprised one and two penetrations, respectively. A similar result was found with developing animals examined 2 mo after lesion [gray bars: 1 electrode penetration (6%); 2 (6%); and 3 (88%)]. The difference between immediately after lesion and recovered animals (both adults and neonates) was significant for patch sizes comprising two penetrations (P = 0.0001) and three or more (P = 0.0001).
When all of this data are considered together, we conclude that there seems to be little immediate unmasking of silent upper incisor projections or expansion of previously weak projections in the cerebellum after a peripheral lesion. Most importantly, the pattern of activity observed immediately after lesion cannot explain the dominance of the upper incisor representation observed 2 mo later.
Do afferent structures contribute to cerebellar tactile map reorganization ?
REPRESENTATION OF THE UPPER INCISOR IN S1 OF NORMAL ANIMALS.
We have shown that the upper incisor representation, dominant in the reorganized cerebellar maps, is represented no more frequently than any other non-upper lip receptive field in normal maps, is no more likely to be adjacent to the upper lip region it comes to occupy, and is no more prevalent than any other representation immediately after deafferentation. We next sought to determine whether any special relationship existed between the upper incisor and the upper lip in structures afferent to the cerebellum. Given the abundant amount of information available on the topographic organization of the somatosensory (S1) cerebral cortex, we chose to examine the relationship between the upper incisor and upper lip representations before and after lesions in S1 maps. We first determined the areal extent and position of the S1 upper incisor representation in normal animals (for a mapping study of all perioral structures, see Remple et al. 2003; for field potential analysis of the incisor representation in rats, see Hayama et al. 1993; for mapping of incisors in other mammals, see Cusick et al. 1986; Jain et al. 2001; Ogawa et al. 1989; Tairak 1987).
Figure 6 presents examples of cortical maps of the upper incisor representation in normal animals. The mean area of this representation in normal animals was 11.5 ± 4.0 x 104 µm2 (n = 3). In each cortical map, the upper incisor was found to be adjacent to the ipsilateral upper lip representation, specifically the rostral/ventral surface of the upper lip (also see Remple et al. 2003). This is the region of the upper lip most heavily represented in crus IIa in the normal adult rat cerebellum (Bower and Kassel 1990).
|
|
|
How can we correlate the expansion in S1 cortex with changes in crus IIa? We used field potentials in crus IIa to relate the two structures. Normal field potentials consist of two components. The first, short-latency component (10–22 ms) originates from the trigeminal complex (Watson and Switzer 1978; Woolsten et al. 1981); the second, longer-latency component (21–37 ms) originates from S1 cortex (Morissette and Bower 1996). Morissette and Bower (1996) previously showed that there is a strong correlation between the latency of the tactilely evoked response in S1 cortex and the second component of the field potential observed in the granule cell layer of crus IIa. No such relationship was found between S1 and the first component of the field potential observed in crus IIa (Morissette and Bower 1996). These same authors also showed that lidocaine injection in S1, cortical ablation, and decerebration significantly affected only the second waveform, not the first. For examples of field potentials observed in lesioned neonates, see Gonzalez et al. 1993; for lesioned adults, see Shumway et al. 1999.
Because the only component of the field potential associated with S1 cortex is the second waveform, we compared the area of the upper incisor representation in S1 cortex with the percentage of the second waveform observed in crus IIa. Figure 8C shows that, as the area of S1 cortex devoted to the upper incisor expands, so does the frequency of penetrations reflecting S1 input only (R2 = 0.45, P = 0.0478, ANOVA). Thus the data suggest that there is a direct relationship between expansion of the upper incisor representation in the somatosensory cortex and upper incisor representation in the reorganized cerebellar maps.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
; Morissette 1996; Shumway et al. 1999). As shown in Fig. 9, the two primary sources of mossy fiber afferent input to crus IIa are the somatosensory cortex, by way of the pons (Bower et al. 1981; Leergaard et al. 2000) and the trigeminal sensory nucleus (Woolston et al. 1981). These afferent structures are topographically organized, but their projections to the cerebellum convey fractured information (Bower et al. 1981).
|
Our results suggest that the pattern of reorganization in crus IIa is not caused by mechanisms intrinsic to the cerebellum itself. First, in studies of lesion-induced map reorganization throughout the trigeminal neuraxis, adjacent representations usually expand into the denervated area (reviews: Garraghty et al. 1993; Kaas et al. 1995a,b). Because all somatosensory structures afferent to the cortex are topographic, there continues to be active debate as to whether this filling-in by adjacent representations is caused by mechanisms intrinsic to the cortex (reviews: Fox 2002; Glazewski et al. 1998; Jones 2000; Krupa et al. 1999), thalamus (Jones and Pons 1998; Shin et al. 1995), or prethalamic structures (Klein et al. 1998). In crus IIa, while there is some variability in the actual patterns of adjacency in the fractured maps across individual animals (Bower and Kassel 1990), our analysis of maps in multiple animals has shown that the upper incisor representation always fills in the deafferented upper lip. Unlike other studies of somatosensory reorganization, this representation is not preferentially adjacent to the upper lip area in normal maps. In fact, our detailed mapping study of the tactile maps in crus IIa in multiple animals (Bower and Kassel 1990) showed that the lower incisor, lower lip, and contralateral upper lip were 7–13 times more frequently adjacent to the upper lip than the upper incisor and 3–9 times more frequently represented in crus IIa.
One important limitation of the current mapping approach is that we did not systematically map the tactile representations in the folial walls of the cerebellum that also includes tactile representations. Therefore we cannot say with certainty that the upper lip representation is not bordered by the upper incisor deep in the folial walls. However, for such a deep representation to account for the reorganization of the crown of crus IIA, there would have to be a strong directional restriction on afferent regrowth. We know of no evidence for this kind of restriction in the cerebellum or any other somatosensory brain region.
Second, in S1 cortex, mapping studies before and immediately after peripheral lesions have shown a rapid unmasking of previously silent or subdominant responses (review: Kaas 1991; rats: Barbay et al. 1999; Byrne and Calford 1991; Doetsch et al. 1996). The unmasked responses (detected in 25% of the penetrations in monkey cortex: Garraghty and Muja 1996) correspond to those which eventually occupy the denervated area (Cusick et al. 1990; Silva et al. 1996; Schroeder et al. 1995, 1997). In rats, some unmasking also occurs at subcortical levels within an hour of infraorbital nerve lesion (Klein et al. 1998; but see Kis et al. 1999). In our study, <8% of the electrode penetrations revealed receptive fields not present before the denervation or that could not be explained by similar representations nearby. Also, the upper incisor was no more or less likely to drive unmasked receptive fields than any other intact representation.
Could the largely silent areas be masking an UI input? We found no evidence that excitability differed for the responsive penetrations, at least during the 6 h of recording after deafferentation. While our assessment of recording strength was qualitative (see METHODS), quantitative somatosensory studies in rats immediately after deafferentation also found no change in excitability (sciatic nerve crush; S1 cortex: Barbay et al. 1999; infraorbital nerve lesion; trigeminal nucleus: Klein et al. 1998). Thus the limited, rapid cerebellar unmasking does not predict the eventual dominance of the upper incisor observed after chronic alteration.
The lack of an immediate unmasking effect is consistent with what is known about afferent mossy fiber termination patterns in the cerebellum. In S1 cortex, unmasking effects are attributed to the spatial breadth of afferent thalamo-cortical projections, which are more broadly distributed than observed physiologically, and intracortical connections (Garraghty and Sur 1990; Hoeflinger et al. 1995; Rausell and Jones 1995). In the cerebellum, the spatial distribution of single mossy fiber afferents is less than the width of the patches in crus IIa determined physiologically (Li et al. 1997; Sultan 2001). Physiologically, unmasking in S1 has been attributed to shifts in the balance of excitatory and inhibitory connections over relatively large distances (hundreds of microns), influenced by neuromodulatory or neurotrophic substances (review: Dykes 1997; Hickmott and Merzenich 1998; Nishimura et al. 2002; Oyesiku et al. 1999). There is no evidence for similar large-scale excitatory and inhibitory interactions in the granule cell layer of the cerebellum. For example, Golgi cell inhibitory feedback is relatively restricted spatially (Brodal 1981). Accordingly, the cerebellum seems to lack an anatomical substrate for the type of immediate unmasking reported in SI cortex.
Third, along the somatosensory pathway, the time-course of reorganization seems to have two stages: unmasking of overlapping projections and use-dependent changes through additional unmasking and/or sprouting (Churchill et al. 1998). Sprouting has been observed in long-term studies after large peripheral injuries in monkeys, particularly at lower levels (review: Florence et al. 1993; Jain et al. 2000; but see Manger et al. 1996). Similarly, long-term studies in rats have shown sprouting in the spinal cord and brain stem after peripheral lesions (Fitzgerald 1985; Fitzgerald et al. 1990; Lane et al. 1995; Mannion et al. 1996; Sengelaub et al. 1997; Waite and de Permentier 1991), as well as sprouting in the hippocampus after central lesions (e.g., Kadish and Van Groen 2003). There is some evidence that sprouting occurs in the rat cerebellum, because, after deafferentation of mossy fiber input, one sees dendrodendritic sprouting of granule cells and axonal sprouting of Golgi neurons and Purkinje cells (Hamori and Somogyi 1982, 1983). However, as with unmasking, sprouting would have to be specific to the upper incisor, suppressing the expansion of the lower incisor and lower lip that are at least equally large and more likely to be adjacent to the upper lip representation in normal rats.
Cerebellar reorganization is associated with reorganization in afferent structures
Given the above, could the topographic details of reorganization in cerebellar maps instead be related to lesion-induced changes external to the cerebellum? We hypothesize that the predominance of the upper incisor in the reorganized cerebellar maps reflects the filling-in of the denervated region by the upper incisor in structures afferent to the cerebellum. The two primary structures providing mossy fiber afferent input to crus IIa are the S1 cortex by way of the pons (Bower et al. 1981; Leergaard et al. 2000) and the trigeminal sensory nucleus (Woolston et al. 1981). These afferent structures are topographically organized, but their projections to the cerebellum convey fractured information (Bower et al. 1981).
How might reorganization take place? A parsimonious explanation would be unmasking of the upper incisor area in S1 and/or subcortical areas (Faggin et al. 1997), with gradual expansion of the representation(s) over time because of selective use. We have previously shown with evoked-potential studies that the reorganization primarily involves expansion of the upper incisor in both cortical and subcortical afferents (Gonzalez et al. 1993; Shumway et al. 1999). However, there are developmental differences between structures, with the direct trigeminal input less plastic than the cortical input for certain developmental periods (physiology: Kis et al. 1999; Shumway et al. 1999; anatomy: Waite and de Permentier 1991). Recently, Land and Shamalla Hannah (2002) also reported a developmental difference. They found that experience had the greatest effect on zinc-containing S1 circuits in animals ranging from PND14 to PND28.
Two lines of evidence support our hypothesis of afferent control of reorganization: the concurrent physiological recordings in S1 cortex and crus IIa reported in this study and timing differences in the development of these two structures.
Lesion of the infraorbital trigeminal nerve should deafferent roughly 60% of S1, including the entire barrel field (Dawson and Killackey 1987; Kis et al. 1999; Rhoades et al. 1996). We have shown in this study that such lesions cause the S1 upper incisor representation to significantly expand, in parallel with the expansion seen in crus IIa. The expansion in S1 coincides with the increase in number of penetrations reflecting S1 input only. An earlier study of infraorbital nerve lesions in S1 (Waite 1984; Waite and Cragg 1981) reported expansion of the lower jaw/inside mouth representation, digits, and vibrissae over the eyes and ears. However, that study combined inside mouth representation with lower jaw and did not report an upper incisor representation. Thus comparison between results is difficult.
Timing differences between the development of the cerebellum and cerebral cortex also suggest a dominant role for afferent structures in cerebellar map reorganization. We have shown that increases in the expansion of the upper incisor in crus IIa occur with lesions made as early as PND1. However, mossy fiber afferents don't reach the cerebellum in rats until PND3-5 (Altman 1972), granule cells are not formed until PND5, with the bulk between PND8 and PND15 ( Arsénio-Nunes and Sotelo 1985; Tolbert et al. 1994), and mossy fiber terminals are rarely found before PND7 (Schoen et al. 1991). Given that rodent cerebellar circuitry does not exist at PND1 (Sanchez-Villagra and Sultan 2002), reorganization resulting from these early lesions cannot take place through intrinsic mechanisms. In S1, afferents seem to be organized into their adult topographic pattern by PND1 (Schlaggar and O'Leary 1994).
Predicted consequences of peripheral lesions in cerebellar-related circuits
The results presented in this paper suggest that lesion-induced changes in cerebellar maps are the consequence of changes in the maps of afferent structures. In crus IIa, the predominant afferent structures are the trigeminal nucleus and the trigeminal-thalamic-cortical pathway. We predict that the point-to-point pattern of afferent projections from the trigeminal nucleus to the cerebellum and from somatosensory cortex (through the pons) to the cerebellum will not change after peripheral lesions. Under control conditions, neurons and axons originating in the trigeminal nucleus or the somatosensory cortex conduct information about the upper lip. We predict that, after lesion and a shift in the topography of the maps in these afferent structures, these same neurons will now convey information about the ipsilateral upper incisor. This prediction is fully testable using anatomical tract-tracing procedures. Our recent anatomical tract-tracing studies of the normal projection pathway from S1 through the pons to the cerebellum (Trygve et al. 2000) were undertaken to lay the foundation for subsequent anatomical tests of our prediction.
Functional implications
Our data suggest that the cerebellar tactile map essentially mirrors shifts in S1 cortical maps after lesion-induced reorganization. In other words, the map itself does not change, reinforcing the likely functional importance of the detailed spatial structure of the cerebellar maps (Bower 1997a,b). Our earlier studies compared the normal and reorganized maps in great detail (Gonzalez et al. 1993; Shumway et al. 1999). These studies showed that the spatial structure does not change. Specifically, the fractured somatotopy is maintained, and the general size and distribution of the patches is similar. As in the normals, the reorganized maps tended to consist of a large patch surrounded by smaller patches. All tactile projections came from perioral structures. Examination of within-patch topography showed that the receptive fields were topographically arranged. There was no change in the spatial organization of those regions not affected by denervation, such as the boundary of the contralateral upper lip patch.
A lack of plasticity in mossy fiber projection patterns is consistent with the hypothesis of Bower and Kassel (1990) that the topography of afferent projections to the cerebellum is hard-wired and developmentally specified. It is also supported by previous experiments. The pattern of mossy fiber projections is established before afferents enter the cerebellum (Arsénio-Nunes et al. 1988; Grishkat and Eisenman 1994; but see Sotelo and Wassef 1991), and this initial pattern does not depend on the presence of the target, granule cells (Vogel and Prittie 1994). The terminal fields of mossy fiber cuneo-cerebellar afferents do not expand after neonatal lesion of spinocerebellar afferents (Ji and Hawkes 1995). The total number of dendrites and synaptic junctions in the reorganized mossy fiber/granule cell glomerular complex does not change after mossy fiber deafferentation (Hamori et al. 1997). The spatial expression pattern of the zebrin marker, which aligns with the boundaries between different tactile representations (Hallem et al. 1999), does not differ after infraorbital nerve lesion in neonates or adults (Leclerc et al. 1988). In summary, the invariant features of the fractured map and the lack of spatial plasticity after cerebellar lesions in rats suggest that the cerebellum's functions involve a high degree of temporal and spatial resolution linked to fundamental spatial relationships between sensory surfaces.
|
GRANTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Present addresses: C. Shumway, Department of Research, New England Aquarium, Central Wharf, Boston, MA 02110–3399; J. Morissette, Medtronic, Inc., 7000 Central Ave., N.E., B408, Minneapolis, MN 55432; J. Bower, Research Imaging Center, University of Texas Health Science Center at San Antonio, San Antonio, TX 78284-6240.
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: C. A. Shumway, Dept. of Research, New England Aquarium, Central Wharf, Boston, MA 02110-3399 (E-mail: cshumway{at}neaq.org)
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Arsénio-Nunes ML and Sotelo C. Development of the spinocerebellar system in the postnatal rat. J Comp Neurol 237: 265–271, 1985.
Arsénio-Nunes ML, Sotelo C, and Wehrlé R. Organization of spinocerebellar projection map in three types of agranular cerebellum: Purkinje cells vs. granule cells as organizer elements. J Comp Neurol 273: 120–136, 1988.[CrossRef][ISI][Medline]
Barbay S, Peden EK, Falchook G, and Nudo RJ. Sensitivity of neurons in somatosensory cortex (S1) to cutaneous stimulation of the hindlimb immediately following a sciatic nerve crush. Somatosens Mot Res 10: 103–114, 1999.
Bower JM. The cerebellum and the control of sensory data aquisition. In: The Cerebellum and Cognition, edited by Schmahmann J. San Diego: Academic Press, 1997a, p. 489–513.
Bower JM. Is the cerebellum sensory for motor's sake, or motor for sensory's sake? The view from the whiskers of a rat. Prog Brain Res 114: 483–516, 1997b.
Bower JM. The functional organization of cerebellar circuitry reconsidered. In: The Cerebellum
