INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The stage of corticothalamic maturation is an important determinant of the functional consequences of sensory deprivation (Antonini and Stryker 1993, Diamond et al. 1994; Jones 2000; Kaas 1995; Kaas et al. 1983; LeVay et al. 1980, McCasland et al. 1992, Nicolelis et al. 1998, Snow and Wilson 1991, Wallace and Fox 1999). Manipulations that alter normal sensory drive in neonatal animals produce substantial alterations in structure and function at multiple sensory pathway levels (Antonini and Stryker 1993; Juliano et al. 1994; Kaas et al. 1983; LeVay et al. 1980; Snow and Wilson 1991; Waters et al. 1990). Comparable procedures in adults result in physioanatomical changes unique to the mature animal (Buonomano and Merzenich 1998; Darian-Smith and Gilbert 1994; Florence et al. 1998; Kaas 1995; Kossut and Juliano 1999). This suggests that the consequences of altered sensory drive on neural function and structure will differ according to the maturation of the nervous system at the time of the manipulation.

In this study, we evaluated functional reorganization following peripheral nerve denervation at different stages of corticothalamic development. Because changes in the contribution from the radial nerve are predominant in studies using median or ulnar nerve lesions, we studied cortical response changes directed by the superficial radial nerve of ferret forepaw after an ulnar nerve lesion (Kaas 1995; Kaas et al. 1983). Radial nerve input to primary sensory cortex substantially overlaps the forepaw representations that receive input from the median and ulnar nerves (Schroeder et al. 1995). Following a loss of input, the typically subthreshold input from radial nerve to the deprived cortical territories becomes dominant (Schroeder et al. 1995, 1997; Wall et al. 1992a,b).

We characterized the spatiotemporal features of responses recorded over somatosensory cortex to input from superficial radial nerve in animals receiving ulnar nerve transections at different phases of cortical development. By using evoked responses at multiple sites at the cortical surface, we obtained an integrated map of electrical activity patterns across time and space in response to tactile digit stimulation. This method allowed us to view global responses in normal and deafferentated cortex, as well as obtain temporal features of the activity distribution. We demonstrate that the contributions to deafferentated cortex from intact superficial radial nerve will vary systematically according to maturation of corticothalamic relationships at the time of nerve transection. Ulnar nerve transections in young animals affect the spatiotemporal features of short-latency responses, while lesions in adult animals affect those of a longer latency component of the evoked response.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animal preparation

Experiments were conducted on six normal and eight experimental ferrets of either sex. During electrophysiological recording, they were anesthetized with halothane (2-3%), and expired CO₂ and body temperature (35-37°C) were monitored and maintained. Ophthalmic ointment was placed onto corneal lens surfaces, and heart rate was monitored periodically during the experiment. An intravenous catheter inserted in the right external jugular vein allowed continuous infusion of 5% dextrose in lactated Ringer's. The scalp and left forelimb were shaved and the animal's head secured in a stereotaxic device. A craniotomy was performed over the right somatosensory cortex and the dura mater removed to expose the pia. The brain was covered with warm mineral oil and photographed (including a scale; Fig. 1). An enlarged photograph was used to map recording sites. All procedures using animals were approved by the Uniformed Services University IACUC and followed the guidelines established by the NIH.

View larger version (26K): [in this window] [in a new window]   Fig. 1. Organization of ferret somatosensory cortex. Coronal sulcus (Cor S), cruciate sulcus (Cru S), posterior sigmoid gyrus (PSG), and coronal gyrus (CG) are labeled. A: digitized photomicrograph of the surface over ferret somatosensory cortex after craniotomy and removal of dura. The center of the postcruciate dimple is indicated by a crosshair. Numbers represent examples of recording sites marked on the photograph. The scale in the chamber is included in the photograph (top). Scale bar = 1 mm. B: topographic and submodality organization of ferret somatosensory cortex. The forepaw representation of ferret somatosensory cortex contains at least 3 representations of the digits that differ principally in terms of receptive field size and sensory modality. Caudally, neurons respond to a mixture of deep and cutaneous input (located comparably to areas 1 and 2 in cat somatosensory cortex and referred to here as area 1/2). More rostrally, neurons are found with small receptive fields representing cutaneous input from individual digits (area 3b). Anterior to area 3b is another cutaneous region containing neurons with somewhat wider receptive fields (caudal area 3a). The most rostral portion of the forepaw representation receives afferent projections from deep tissue, i.e., joints and muscles (rostral area 3a). Spatial distribution of somatosensory evoked potentials to tactile stimulation of forepaw digits (C and D). Responses to stimulation of digits 5 and 3 (C and D, respectively) were recorded at several sites overlying the cortical surface. *Location of the recording site with the earliest cortical activity. Circle, coincident early activity of opposite polarity that is due to activation of the digit 3 representation within the wall of the coronal sulcus (see Fig. 2 for further description). Triangles, short-latency, short-duration peaks recorded over the caudolateral crown of the posterior sigmoid gyrus (these presumably include synchronous activity in thalamic radiation); black diamonds, low-amplitude, delayed negativities in the rostral cutaneous field (presumably caudal area 3a); C: responses evoked to stimulation of digit 5 support its representation in the crown of the posterior sigmoid gyrus. D: spatial distribution of responses to stimulation of digit 3 is consistent with its representation partly in the crown and partly in the wall of the coronal sulcus (see Fig. 2 for further discussion). In C and D, stimulation occurs 6 ms after the beginning of each trace.

Data collection

Silver ball electrodes (200-400 µm diam) were used to record from the cortical surface (impedance, 0.5-1 MOhm at 1 kHz). Silver ball reference leads were placed on retracted muscle, lateral and adjacent to the craniotomy. Recordings were made sequentially at 20-25 sites that covered 12-16 mm² of cortex, which encompassed the somatosensory areas, including the region surrounding the postcentral dimple, the coronal (suprasylvian) gyrus, and nearby regions of motor cortex (Fig. 1A). A record at each site consisted of an average of 30 single-trial responses. A subset of individual recording sites were recorded at multiple times within the period of data collection as a means for assessing changes in the state of the animal over time; datasets containing waveforms that varied substantially during the session were excluded. Data collection began 20 ms before the onset of stimulation and terminated 180 ms later. Stimuli were presented at a rate of 1/s and consisted of light mechanical stimulation of the skin. Mechanical stimuli were 1-cm-diam Perspex spheres presented for 1 ms proximal to the dorsum of digits 2, 3, 4, or 5 (Juliano et al. 1981). Only the azimuth of the spheres contacted the skin during stimulation. To ensure stimulation of only the dorsal surface, individual digits were elevated and stabilized. The perceptual experience of the stimulus when presented to the dorsal surface of an immobilized digit in humans is a brief, suprathreshold tap. Paws and digits were shaved prior to mechanical stimulation.

Nerve transections

To examine the consequences of peripheral manipulation on the development of somatotopy in ferret somatosensory cortex, the ulnar nerve was lesioned in animals at different ages. We selected the ulnar nerve to determine the effect of its absence on radial nerve representations in sensory cortex and because a large proportion of its cortical projection territory lies in the crown of the posterior sigmoid gyrus and is easily accessible during recording. Ulnar nerve transection occurred at three different ages: between postnatal days 5 and 7 (P5-P7, n = 4), between postnatal days 20 and 21 (P20-P21, n = 2), or when the animals were adults (i.e., older than 6 mo, n = 2). Under sterile conditions, juvenile animals were anesthetized with ketamine (40 mg/kg) and rompun (4 mg/kg) and received ulnar nerve sections at the level of mid-forearm, deep to the flexor carpi ulnaris and palmaris longus. Following ligation, a 2- to 3-mm portion of the ulnar nerve was removed in each case, and the cut end of the nerve was restored to its relative location with respect to surrounding musculature. Adult animals receiving ulnar nerve transections were anesthetized with halothane (3%) in a mixture with oxygen (2%) and nitrous oxide (0.5%). Animals were monitored for full recovery from anesthetics before being returned to their home cages.

The electrophysiological experiments occurred 6-8 mo after the nerve lesion. After each recording session, the forelimbs of normal and lesioned animals were examined to confirm the anatomical identity of the transected nerve. All experimental animals included here correctly received transections of the ulnar nerve; re-innervation of the distal forepaw by ulnar nerve was not evident. After each recording session, animals were given an overdose of sodium pentobarbital (65 mg/kg, iv) and perfused transcardially with 0.9% saline followed by 4% sucrose in 0.1 M phosphate-buffered 4% paraformaldehyde.

Data analysis

Digitized data were submitted to Mathematica (Wolfram Research) for digital filtering and analysis. Using an inverse Fourier series smoothing approach, raw data for each dataset were filtered to remove components with frequencies higher than 200 Hz. Records containing artifacts were omitted from further processing. To monitor physiological stability, we sampled evoked potential waveforms at a subset of sites multiple times to confirm reproducibility across time. In cases where waveform components varied substantially across recordings, we omitted that dataset, presuming instability of the experimental preparation. Amplitude and latency measures of the response peaks of each waveform were made by obtaining the minimum and maximum values (amplitude) and their respective locations in time (latency).

Spatial maps of the response were obtained using an inverse square interpolation of the amplitude values. Coordinates for the map were measured from the locations of recording sites marked on the photograph, using the postcruciate dimple as the origin (Fig. 1A). Positions of each recording site were marked precisely on the photograph during each recording session. A color look-up table was applied to the amplitude values to provide a chromatic index of response intensity. Amplitude values for each dataset were normalized to the maximum value contained in the look-up table; all maps use the same maximum and minimum values. This approach allowed us to obtain a global view of the spatiotemporal features of neural activity in underlying somatosensory cortex (Nunez 1981, 1990).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Spatiotemporal distribution of cortical surface potentials in normal animals

In this study, we examine evoked responses recorded at the pial surface overlying ferret somatosensory cortex in normal and nerve-lesioned animals to stimulation of dorsal digits. The representations of digits 4 and 5 in area 3b lie on the crown of the posterior sigmoid gyrus (Fig. 1, A and B). The representation of digit 3 can lie partly on the crown and partly within the medial bank of the coronal sulcus (Fig. 1B) (McLaughlin et al. 1998). Because it can be represented by neurons located more deeply in cortex, responses to digit 3 stimulation can consist of two peaks of opposite polarity (Figs. 1D and 2, bottom). In area 1, representations of digits 3, 4, and 5 lie in the caudal crown of the posterior sigmoid gyrus and partly in the wall of the coronal sulcus (Fig. 1B), and activity originating in area 1 can presumably be detected as low-amplitude components in the surface-recorded waveform (although this was not confirmed in this study). After digit 5 stimulation, the most pronounced cortical response occurs near the postcruciate dimple, in agreement with the location of its representation (Fig. 1C) (McLaughlin et al. 1998). Early, long-duration peaks overlie central cutaneous representations in area 3b (Fig. 1C, asterisks) where receptive fields are smaller and more restricted to individual digits. Shorter-duration, smaller early negativities with onsets slightly advancing those marked by asterisks and occurring nearer to the coronal sulcus (see triangle in Fig. 1C) may be associated with the incoming thalamocortical volley, which would be detectable using our reference electrode placed on the lateral skull (Desmedt and Cheron 1980; McLaughlin and Kelly 1993). The long-duration, low-amplitude negativities that occur at rostral sites are presumed to represent the area 3a responses (see the black diamond in Fig. 1C). Based on our topographic mapping studies, cutaneous input will also drive neurons in the caudal portion of area 3a (Fig. 1B) (McLaughlin et al. 1998).

View larger version (14K): [in this window] [in a new window]   Fig. 2. Detailed features of somatosensory evoked potentials recorded at sites overlying the cutaneous representation of the forepaw. Evoked potentials were elicited by tactile stimulation of digit 3 in a normal animal and animals receiving ulnar nerve lesions at postnatal days 5 (P5) and 20 (P20) and in adulthood. Bottom: example of polarity inversions that occur in responses due to neural generators located within the wall of the coronal sulcus (Cor S). Inversions occur as a result of stimulation of skin sites that project specifically to somatosensory cortex occupying the medial wall of the coronal sulcus (i.e., digit 2 and occasionally digit 3). An activated region of cortex behaves as an equivalent dipole that generates a positivity rostromedially, i.e., overlying the posterior sigmoid gyrus (PSG) and a negativity caudolaterally, i.e., overlying the coronal gyrus (CG). The N1-P1-N2-P2 complex (top waveform) and its inverted counterpart (bottom right waveform) are mirror image evoked waveforms representing the same active neural ensembles.

In response to stimulation of digit 3, polarity inversions as described by Fig. 2 (bottom) are evident for initial activity (Fig. 1D, single asterisk and circle). The initial cortical response, represented by a long-duration negative peak, is present lateral to the coronal sulcus (see single asterisk). This component lateral to the sulcus is temporally coincident with an opposite-polarity peak overlying the crown of the posterior sigmoid gyrus (see circle in Fig. 1D). Numerous studies demonstrate that evoked potential configurations of this nature are consistent with an interpretation of the origin of the recorded activity as opposite poles of a deeply situated, active cortical source (e.g., Allison et al. 1990).

Time-domain components in normal cortex

All experiments were conducted on adult ferrets; those with nerve lesions earlier in their life were tested electrophysiologically between 6 and 8 mo of age. We obtained physiological responses at multiple locations on the cortical surface following localized light cutaneous digit stimulation. The most pronounced features of the waveforms are an early, long-duration negativity followed by a large, long-duration positivity (Fig. 2). The average peak latency of the initial cortical response, component N1, for digits 3, 4, and 5 is about 15-16 ms (Table 1). The evoked waveform to tactile stimulation is reproducibly followed by a single small positive and a single small negative component, and then by a larger positive component, P2 (Table 1).

The overall morphology of the evoked response in lesioned animals was similar to those recorded in normal animals. A statistically significant difference in the latency-to-peak measurement, however, occurred across treatment groups. Untreated animals, those lesioned as adults, and those lesioned at P20 exhibited N1 peak latencies in the 15- to 16-ms range, whereas peak latencies in animals lesioned as neonates at postnatal day 5 were significantly delayed in comparison (Fig. 3).

View larger version (13K): [in this window] [in a new window]   Fig. 3. Mean latency of the initial cortical response to tactile stimulation in normal animals and adult ferrets lesioned at P5 and 20 and as adults. The number of datasets (n) used for each category is shown. The peak latency of N1 in animals lesioned at P5-P7 is significantly different compared with N1 latencies measured for all other groups (P < 0.001, Student's t-test). **P < 0.001. Error bars represent SD.

Cortical surface activity maps

To obtain a view of the spatial extent of cortical activation to localized digit stimulation, we computed two-dimensional, color-indexed maps of evoked activity at a series of discrete time points. These images represent transformed activity values at discrete time points. All transformations were computed and normalized using the same color look-up table. The red and blue color indices reflect normalized values of 0.75 or higher. This approach confers uniformity across datasets and emphasizes larger amplitude response peaks. The largest evoked components are typically the early negativity N1, represented by red, and a later positivity P2, represented by blue. Digit 5 stimulation elicits an initial cortical response with an onset at about 12 ms, a duration approaching 15 ms, and a maximum near the postcruciate dimple, consistent with somatotopic organization of ferret somatosensory cortex (Fig. 1B). The distribution of this early activity is made more apparent in color-coded and amplitude-scaled surface maps (15 and 20 ms for digit 5 in Fig. 4). Digit 5 activation is associated with an activity focus near posterior cruciate dimple and overlying area 3b, and a smaller focus at 20 ms assigned to the cutaneous portion of area 3a. Following the initial cortical activity, responses at 25 and 30 ms overlie the caudal cutaneous field; this later activity is associated with a positivity and is coded blue (Fig. 4). Similar features are evident in the activity maps for stimulation of digits 3 and 4 (Fig. 4).

View larger version (67K): [in this window] [in a new window]   Fig. 4. Color-coded maps of the spatiotemporal distribution of evoked responses at discrete time points. Maps are shown for stimulation of digits 5, 4, and 3. In each map, red and blue represent a range of values within a dataset that code the largest negative and largest positive values, respectively; other colors code intensity ranges in between. Color look-up tables were applied to the amplitude values to provide a chromatic index of response intensity. Surface maps presented were all derived using the same look-up table. Amplitude values were normalized to the maximum value contained in the look-up table; color-coding reflects the relative distribution of response amplitudes for each dataset. Spatial maps of the responses were obtained using an inverse square interpolation of the amplitude values. Cartesian coordinates for the map were measured from the locations of recording sites marked on the photograph, using the postcruciate dimple as the origin (cross hair; also see Fig. 1). Coronal sulcus is represented by the heavy black line. Responses to stimulation of each digit conforms to expectation about somatotopy and submodality. Scale bar = 1 mm.

Peripheral distribution and activation pattern of radial nerve

To more confidently interpret the surface maps of normal and ulnar-nerve lesioned animals, we characterized and mapped the receptive fields of isolated nerves in the forearm of normal and lesioned animals. The ulnar nerve in ferret innervates ventral and dorsal portions of digits 4 and 5. The superficial radial nerve, however, also provides sensory innervation for dorsal digit 4 and the radial portion of dorsal digit 5 (Fig. 5, top inset). We found this innervation pattern reproducibly in normal (n = 4) and ulnar nerve-transected (n = 3) animals. The peripheral distribution of the radial nerve did not differ at the macroscopic level between normal and ulnar nerve lesioned animals, suggesting that the radial nerve did not alter its peripheral distribution after the ulnar lesion. To determine the region of cortex usually activated by this nerve, we stimulated the superficial radial nerve in three normal ferrets using mild, bipolar electrical stimulation near the wrist and recorded activity at the cortical surface. Stimulation of a distal branch of superficial radial nerve near ulnar paw evokes early activity over a circumscribed region of cortex, including initial activation of central area 3b and the caudal cutaneous representation, presumably areas 1 and 2. Succeeding activity also incorporates caudal (cutaneous) area 3a (Fig. 5; see asterisks at 15 and 20 ms). The evoked response peaks to stimulation of superficial radial nerve are similar to those evoked by tactile stimulation. The prominent, long-duration early negativity N1 and later P2 components are evident and have proportional amplitudes similar to their counterparts recorded to tactile digit stimulation. Component N1, represented by red in the color maps, occupies a sizeable amount of cortex on the crown of the posterior sigmoid gyrus, whereas the distribution of component P2 is not as uniformly extensive.

View larger version (43K): [in this window] [in a new window]   Fig. 5. Spatiotemporal distribution of responses to stimulation of superficial radial and ulnar nerves (A and C). A: surface map of activity after stimulation to a branch of superficial radial nerve that had its receptive field principally on dorsal digits 4 and 5. B: representative waveforms to stimulation of superficial radial nerve. Both responses were obtained over the central cutaneous field of somatosensory cortex and contain components similar to those identified in Fig. 3. C: surface map of activity to mild stimulation of the ulnar nerve at the level of distal forearm. The separate response foci on the cortical surface reflect distinct cortical projection targets for the fibers of the ulnar nerve. The receptive field of the nerve at this level of the forearm is shown, depicting the innervation zones (top inset). D: representative waveforms to stimulation of distal ulnar nerve. Responses were obtained over the cutaneous fields of somatosensory cortex and contain components similar to those in Fig. 3. The maps were derived as described in Fig. 5. Top inset: innervation territory of the superficial radial and ulnar nerves, both recorded at the level of distal forearm. Bottom inset: schematic depiction of the relative spatial locations of the representations of radial and ulnar nerves in somatosensory cortex. Scale bar = 1 mm.

Surface maps of responses to mild electrical stimulation of the ulnar nerve at distal forearm show separate cortical fields that represent the peripheral tissue innervated by ulnar nerve (Fig. 5). The caudomedial location near the postcruciate dimple reflects responses elicited by afferents innervating forepaw skin, while the location of the rostral focus (asterisks in Fig. 5C) is consistent with activation of neurons with wide receptive fields and receive projections from deeper tissues (e.g., muscle afferents). The representation of radial nerve-innervated territory can be roughly regarded as falling between and overlapping the two main cortical fields activated by ulnar nerve stimulation (Fig. 5, bottom inset). Evoked potentials to ulnar nerve stimulation contain peaks N1 and P2 and other familiar peaks, all in relative proportional and temporal relationships as previously shown (Fig. 5D).

Surface maps in denervated animals: input from digits 2 and 3

To facilitate comparison of the consequences of nerve transections, we regarded the dorsal input as deriving from two compartments. One compartment overlaps with normal ulnar nerve input (referred to as "digits 4-5") and the other does not (referred to as "digits 2-3").

After a lesion at P5-P7, the spatiotemporal surface maps reveal numerous distinct, noncontiguous activity foci in response to tactile stimulation. Following stimulation of digit 3, active regions did not conform to expected submodality distributions and were widespread, occurring over a broader cortical territory, especially at later latencies (Fig. 6A). Maps of the spatial distribution of the response also support the peak latency delay measurements for animals lesioned at P5-P7 (Figs. 6 and 7). The animals lesioned at P5-P7 display weak or diffuse responses at 15 ms, which later become more prominent.

View larger version (52K): [in this window] [in a new window]   Fig. 6. Spatiotemporal distribution of cortical activity evoked by tactile stimulation of digit 3 in animals receiving nerve transections. Maps derived as described in Fig. 4. A: after a lesion at P5, there is a delay to reach the peak of the initial cortical response (N1) and the activity is more widely and irregularly distributed. B: after a lesion at P20, initial cortical activity (N1) is expanded and slightly less focused compared with normal. C: adult nerve transections lead to changes that are more apparent later in the response and are evident as expansion of the P2 distribution. Scale bar = 1 mm.

View larger version (42K): [in this window] [in a new window]   Fig. 7. Spatiotemporal distribution of activity elicited by stimulation of digit 4 in animals receiving nerve transections. Maps derived as described in Fig. 4. A: after a P5 lesion initial cortical activity is greatly expanded compared with the normal pattern. B: lesion at P20 leads to a slightly expanded response distribution compared with normal. C: after nerve lesions in adult animals, later activity (P2) is more widely distributed. Scale bar = 1 mm.

We quantified the area of cortex activated by mild tactile digit stimulation in normal and ulnar nerve-lesioned animals by obtaining the areal measure (in pixels) of the distribution of N1 and P2 activity. Compared with the N1 distribution measured in normal animals, we found that the most substantial increases in the extent of activated cortex for the initial response (N1) occurred in animals receiving nerve lesions at postnatal days 5-7. When quantified, we found that the overall amount of cortex activated by tactile stimulation of digits 2 and 3 was statistically larger than in normal cortex (Fig. 8; Mann-Whitney U test, P < 0.01). After a lesion at P20, however, the cortical response pattern is less fractionated and appears more as a rostral expansion of the normal pattern, especially at the earliest latencies (Fig. 6B). Quantification of this expansion in P20-lesioned animals reveals that the changes are not significantly different from the extent of cortex active in normal animals to digit 2-3 stimulation (Fig. 8).

View larger version (38K): [in this window] [in a new window]   Fig. 8. Relative proportion of recorded cortical area occupied by components N1 and P2. Areal measurements of N1 and P2 were obtained and compared with normal values that were set to unit value for comparison across digit compartment and component (indicated as Index of Active Cortex on the y axis). Illustrated are mean values per experimental group relative to normal mean values. Yellow hues represent normalized values for component N1 and blue hues represent normalized values for P2. The amount of cortex initially responsive to stimulation of digits 4 and 5 in adults that received nerve lesions as juveniles at postnatal days 5-7 is significantly larger than the area activated by projections of digits 4 and 5 in unlesioned animals (P < 0.01, Mann-Whitney U test; 4.7 ± 1.1, n = 7 datasets for normal adults; 17.8 ± 1.7, n = 5 for P5-P7 lesioned animals). The initial cortical response to stimulation of digits 4 and 5 in animals with ulnar nerve transections at P20-P21 is also significantly different from measures of cortical activation in normal animals (P < 0.05, Mann-Whitney U test, 13.2 ± 2.6, n = 5 for P20-P21 lesioned animals). Nerve lesions in adults do not substantially alter the distribution of N1 to D4-D5 stimulation (3.1 ± 0.72, n = 4), whereas adult lesions lead to statistically significant changes in the distribution of P2 (P < 0.05, Mann-Whitney U test, 2.7 ± 1.2, n = 7 for normal adults; 5.6 ± 1.1, n = 4 for adult lesioned animals). The effects on P2 distributions were comparable for digits 4 and 5 and digits 2 and 3 (P < 0.05, Mann-Whitney U test, 3.6 ± 1.0, n = 11 for normal animals receiving stimulation to digits 2 or 3 compared with 6.8 ± 1.2, n = 5 for adult lesioned animals). Stimulation of digits 2 or 3 did not induce statistically significant changes in the distribution of P2 in animals lesioned at P5-P7 (1.8 ± 0.31, n = 5) or P20-P21 (3.5 ± 0.32, n = 4). Ulnar nerve transections at P5-P7 resulted in statistically significant changes in N1 to tactile stimulation of digits 2 and 3 (P < 0.05, Mann-Whitney U test; 4.4 ± 0.77, n = 11 for normal adults; 11.8 ± 0.8, n = 5 for P5-P7 lesioned animals). Changes were less pronounced when digits 2 and 3 were stimulated in animals that received manipulations at P20-P21 (5.8 ± 1.5, n = 4). **P < 0.01; *P < 0.05.

In animals receiving ulnar nerve lesions as adults, the predominant effect was a change in the distribution of the response at longer latencies, so that the distribution of P2 activity was markedly expanded in these animals (Fig. 6C). There is an inverse relationship between age at the time of lesion and the latency of the affected component (N1 or P2) of the response. As clearly evident in Fig. 8, animals lesioned as juveniles exhibit more widely distributed early responses (N1), whereas in adult animals, the largest expansions are seen later (P2). The spatial extent of component P2 is statistically significantly different for animals lesioned as adults compared with normals (Mann-Whitney U test, P < 0.05).

Surface maps in denervated animals: input from digits 4 and 5

When the dorsum of digit 4 or 5 is stimulated in adult animals 6-8 mo after an ulnar nerve lesion, the most widespread changes in activity occur after a P5-P7 lesion, which results in a clear rostromedial expansion (Fig. 7A). The activity is extensive and not easily assigned to a cytoarchitectonic region, but includes neural activity in cortex receiving projections of the superficial radial nerve (see the large foci in the 20- and 25-ms maps of P5 in Fig. 7) as well as in rostral area 3a. The spatial extent of N1 after lesions at P5-P7 is statistically significantly different from normal (Mann-Whitney U test, P < 0.01).

The spatial distribution of responses to digit 4 stimulation in animals lesioned at P20 also reflects a predominantly rostral spread of initial cortical activity (N1). The resulting cortical activation pattern following lesions at P20 is not as expansive as that evoked after lesions at P5-P7 (Fig. 7). The largest effect occurs for N1, where the representation of digits 4 and 5 increased nearly threefold relative to the extent in untreated animals (Fig 8). Responses in lesioned adults show a differential effect on component P2. The size of the distribution for P2 in animals with nerve lesions as adults is nearly double that for normal animals (Mann-Whitney U test, P < 0.05).

Figure 8 provides an index of the relative area of cortex activated by stimulation of digit 4 or 5 after ulnar nerve lesion. Transections at P5-P7 and P20-P21 result in expansions in the distribution of component N1 over the recorded area. These expansions are significantly larger than the area of cortex initially activated in normal animals (Fig. 8). Differences in the spatiotemporal distribution for digits 2-3 were most marked for lesions at P5-P7. The consequences of ulnar nerve transections in adults led to distributions of component N1 that were normal, while measures of areal activation on the later component P2 were statistically significantly different from values obtained for unlesioned animals (Mann-Whitney U test, P < 0.05). Changes in the spatial extent of P2 after nerve transections in adults were comparable for stimulation of digits 2-3 and 4-5.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Spatiotemporal activity patterns in ferret somatosensory cortex

The evoked responses recorded over ferret somatosensory cortex in response to tactile or electrical stimulation consist of a set of reproducible components. Using data obtained from multiple recording sites over a cortical area of about 16 mm², we quantified the spatial distribution of response components N1 and P2. In normal animals, the general spatiotemporal sequence can be described as initial activity in the central cutaneous field (area 3b), flanked in time by rostral (area 3a) and caudal (presumably area 1 or 2) activity (Fig. 5). Animals with perinatal lesions (P5-P7) exhibited pronounced enhancements in N1 distribution and a delay in its peak latency for all digits. The delay in peak latency is consistent with the additional time necessary to integrate activity over a larger volume of tissue that includes recruited deafferented cortex (Wallace and Fox 1999; Zarzecki et al. 1993). In older juveniles (P20-P21), response distributions changed for both the initial component N1 and the P2 component, particularly after digit 4-5 stimulation (see Fig. 8 for the intermediate behavior of N1 and P2 for nerve lesions in older juveniles). Additionally, ulnar nerve lesions in adult animals principally influenced the distribution of component P2. Nerve lesions performed perinatally altered initial cortical activation, a consequence consistent with modification of thalamocortical input. Transections in adults influenced later response components, implicative of changes in intracortical connections. Thus the findings support a direct (correlated) relationship between the age of lesion and the latency of the affected component of the response.

Our recordings of peripheral nerve are consistent with the role of central mechanisms in mediating these age-dependent effects. We found that the innervation territory of skin supplied by superficial radial nerve is maintained in lesioned animals. No detectable peripheral sprouting occurs after ulnar nerve lesions at any age, a finding consistent with results from Wall et al. (1992b), who found no physiological evidence for peripheral sprouting after nerve transections in postnatal primates.

Age-dependent consequences of peripheral nerve transection

Several types of cortical reorganization after deafferentation have been noted in previous studies, including expansions of intact representations (i.e., expansion), and the appearance of scattered, ectopic representations (i.e., fractionation) (Juliano et al. 1994; Kaas et al. 1983). The data presented here contain support for each of these configurations. Our results demonstrate that expansion and fractionation occur at different times during development, suggesting that these two types of cortical reorganization are governed by distinct mechanisms.

A number of mechanisms may underlie the emergence of responses to superficial radial nerve in the cortical representation of ventral forepaw after transection of ulnar nerve. Inputs from concentrically organized thalamocortical afferents make terminations into cortex that have a dense core of fibers ending on a central aggregate of neurons and a sparse surround of fibers ending on neurons fringing the central aggregate (Wall 1988). The central termination drives the target neurons of the central aggregate effectively, whereas the sparse fringe termination drives the target neurons to a lesser degree and may be under tonic inhibition from the cells in the central aggregate of neighboring cells (Alloway et al. 1989; Favorov and Whitsel 1988a,b; Garraghty et al. 1991, 1994; Juliano et al. 1989; Wall 1998). Fringe inputs may therefore be associated with ineffective synaptic connections or be under either pre- or post-synaptic inhibition mediated by cells in neighboring neurons (Zarzecki et al. 1993). In normal primate somatosensory cortex, functional input of the superficial radial nerve onto neurons in the median and ulnar nerve representations are subthreshold for eliciting action potentials, but are demonstrable using recordings of extracellular potentials (Schroeder et al. 1995). When the median nerve representation in primates is deafferented, there is an increase in the size of the response to radial nerve stimulation in deafferented cortex (Schroeder et al. 1997). Moreover, the sequence of activation reflects direct input from radial nerve-linked thalamocortical afferents to middle layers of the median nerve representation (Schroeder et al. 1997).

Several investigators have found evidence that after a median or ulnar nerve lesion, the radial nerve becomes a dominant nerve input and that its preexisting fringe inputs into both ulnar and median nerve cortical territory can effectively drive the neurons of this region, perhaps as a result of release from tonic inhibition (Alloway et al. 1989, Schroeder et al. 1997, Schwark 1996, Wall et al. 1992a). Our data are consistent with a dominance by superficial radial nerve after ulnar nerve lesions. In normal ferret forelimb, our peripheral nerve recordings clearly demonstrate that radial nerve innervates the dorsum of all the forepaw digits and overlaps the ulnar nerve innervation to digits 4 and 5. It is expected, therefore that the dorsum of digits 4 and 5 could gain dominance over the deafferented cortex.

Neural origin of components N1 and P2

The neural origin of somatosensory-evoked components has been examined in the somatosensory cortex of several species using intracortical recordings, cortical ablations, and pharmacological manipulations (Allison et al. 1990; Legatt et al. 1986a,b; McLaughlin and Kelly 1993). In human somatosensory cortex, the typical early response to median nerve stimulation consists of four reproducible components that originate in specific layers. An early negativity is linked to neural activity in the middle layers of primary somatosensory cortex followed in 10 ms by its upper layer counterpart (McLaughlin and Kelly 1993). A number of studies on cats, using a diversity of recording arrangements and stimulation conditions, found response latencies for initial cortical activation to be in the range of 10-15 ms (Allison and Hume 1981; Heninger and Sheard 1976; Iragui-Madoz and Wiederholt 1977a,b; Iwayama et al. 1987). The onset latencies of the earliest components recorded here in ferret are within the expected range based on extrapolation from the feline data. Our initial prominent, prolonged negativity N1, therefore likely reflects the onset of activity in the recipient layers of somatosensory cortex. Iwayama et al. (1987) noted that the initial response in cortex reversed in the middle layers and showed its highest amplitude near the level of the inversion. The ensuing component is largest in the superficial layers and has an opposite-polarity counterpart in the deep layers. This suggests that the initial cortical response (N1) is related to middle layer processing whereas the longer latency peak (P2) is associated with upper layer processing.

We interpret component N1 as a manifestation of direct, short-latency, excitatory activity in neurons in the middle layers of deafferented cortex to stimulation of superficial radial nerve. Nerve transections in younger animals lead to substantial changes in the distribution and peak latency of N1, whereas lesions in older juveniles and adults exert much less influence on initial cortical activity. Later activity, P2, is more heavily affected by lesions in adult animals, and presumably reflects processing subsequent to initial activation of middle cortical layers. Two possible mechanisms involved in the differential effects on components N1 and P2 could be the phase of thalamocortical afferent terminations at the time of the lesion and the capacity for expansion of the "fringe" (Wall 1988). It is well known that cell responsiveness in the extragranular layers of mature somatosensory cortex, particularly the upper layers, is more pliable than it is for middle layer neurons (Diamond et al. 1994; Kossut and Juliano 1999; McCasland et al. 1992, McLaughlin and Juliano 1997; McLaughlin and Kelly 1993; Wallace and Fox 1999). Plasticity in mature animals might also be served by normal intracortical connections that exist between the dorsal and ventral compartments of the paw (Doetsch et al. 1988).

Corticothalamic development at the time of ulnar nerve transection

Ferret somatosensory cortex is immature at P5-P7 and not completely mature at P20-P21 (Juliano et al. 1996; Noctor et al. 1997, 2001). Numerous studies of cortical development in rat and cat demonstrate that thalamic afferents refine anatomically and functionally over time in an activity-dependent manner (Antonini and Stryker 1993; Daw et al. 1992; Fregnac and Imbert 1984; LeVay et al. 1978, 1980; Stryker and Harris 1986; Waters et al. 1990). The initial termination pattern, however, is roughly topographic and develops independently of activity (Agmon et al. 1993; Molnar and Blakemore 1995; Trachtenberg et al. 2000). In ferret, projections from sensory thalamus are immature at birth. Although thalamic axons impinge on ferret somatosensory cortex by P1, they are not fully refined until the fourth week of life (Noctor et al. 2001).

The capacity for extensive elaboration is progressively diminished during ontogeny (Antonini and Stryker 1993). The P5-P7 ulnar lesioned animals may exhibit a higher degree of fractionated spatial activity due to unguided or misguided elaboration of thalamic axons, which is no longer possible at P20-P21. At the time of nerve transections in neonatal animals, the thalamic afferents have presumably made genetically determined terminations onto cortex and conferred a general sensory map. Following the neonatal lesion, the thalamocortical afferents likely undergo activity-dependent organizational changes that result in an abnormal distribution and changes in the spatiotemporal distribution of the N1 component of the response.

The consequences of lesions in adult animals occur after stabilization of thalamic projections and maturity of intracortical connections, whereas the outcome of manipulations at P20-P21 are likely to reflect a mixture of mature thalamocortical terminations and still maturing intracortical connections (Juliano et al. 1996; Sonty and Juliano 1997). Intermediate effects on components N1 and P2 for juveniles lesioned at P20-P21 is consistent with recent termination of thalamic afferents and immature intracortical connections. In adults, the consequences of lesions are strongly related to changes in component P2, which implicates intracortical processes, including intracolumnar and intercolumnar mechanisms (Diamond et al. 1994; Juliano et al. 1990; Sonty and Juliano 1997; Wallace and Fox 1999). Changes in the spatial distribution of component P2 are consistent with the potential for the upper layers to continue to display plasticity later in life (Diamond et al. 1994; Florence et al. 1998; McCasland et al. 1992, McLaughlin and Juliano 1997; McLaughlin and Kelly 1993; Wallace and Fox 1999).