EXPERIMENTAL

NEUROLOGY

115,2%-259

(19%)

Convergent Inputs to Single neurons in Two Dj~erent Subdivisions of Somatosensory Forepaw Digit Cortex of the Raccoon GERNOT S. DOETSCH,**~ ~~e~rt~ent

S. DAVID STONEY, JR.,~ AND DAVID H. HAUGE*

of Surgery (Section of N~~ro~~ge~~, and ~~epurt~nt of Fh~sio~~ Medical Colkge of Georgia,Augusta, Georgia 30912

The aim of this study was to compare the physiological properties of single neurons in the glabrous (G) and heterogeneous (H) subdivisions of primary somatosensory digit 3 cortex of adult raccoons. ~xtracellular recordings were obtained from 50 G neurons whose receptive Aelds (RFs) were confined to the glabrous skin of a digit, and 4 1 H neurons whose RFs were located on hairy skin, claws, or mixtures of skin types. Both electrical and mechanical stimulation of the digits were used to assessexcitatory neuronal responsiveness. The two sets of neurons, which had nearly identical depth distributions, differed considerably in their input convergence: (i) the percentage of neurons (%N) responding to electrical or mechanical stimulation of each offfocus digit and (ii) the number of digits from which individual cells could be driven were significantly greater for H neurons. Those G and H cells which could be excited by off-focus inputs were examined for probability of response (P), number of spikes per response (SIR), and latency of response (A) to digit stimulation. Surprisingly, for input from any one digit, there were no significant differences in these response properties between the two sets of neurons. However, inputs from different (on-focus versus off-focus) digits varied significantly and revealed patterns of response properties that were qualitatively similar for both G and H neurons. Specifically, %N and P decreased while L increased symmetrically with distance of each off-focus digit from the central on-focus digit 3, reflecting corresponding variations in the synaptic accessibility and conduction time of off-focus excitatory inputs. In contrast, SJR values were very similar for all digits, suggesting that the synaptic strength of off-focus inputs is regulated independently of accessibility. Finally, preliminary findings indicated that denervation of the third digit caused a decrease in off-focus response latencies, while the normal latency profile across digits was retained. This suggests that the previously existing pattern of off-focus inputs to G and II neurons provides a template for denervation-induced cortical reorganization, whereby the synaptic efllcacy of off-focus inputs is increased by disinhibition or facilitation. o ieez Aeademie PKWW, h. 1NTRODUCTION

Many studies have docum~nted~hanges in the physiological and topographic organization of primary soma250

0014~48&X6/92 $3.00 Copyright

0 1992 by Academic

Press,

Inc.

and E~r~no~~,

tosensory (SmI) cortex of adult mammals following damage to peripheral somatosensory structures (for reviews, see (l&18,22,34,35)). The dominant feature of such dene~ation-induced cerebral plasticity is a shift in the excitatory drive onto partially deafferented cortical neurons, such that they become responsive to stimulation of previously “ineffective” skin regions. There is some consensus that a principal mechanism involved in the appearance of novel receptive fields (RFs) is the unmasking of existing connections that originate from disparate regions and converge onto individual target neurons where they normally exert weak excitatory or subthreshold synaptic effects (5,18,19,21). The results of previous studies in our laboratory (3, ‘7, 19, 20) support this idea. Denervation of the third forepaw digit in young and adult raccoons was found to produce shifts in the responsiveness of clusters of neurons in digit 3 cortex from digit 3 to the adjacent digits or palmar pads (19, 20); similar results were obtained by Rasmusson (24) after denervation of the fifth digit. Furthermore, ablation of a portion of the digit 3 cortical zone caused neurons in adjoining cortical digit representations to develop larger than normal RFs (7). Both findings suggest that altered responsiveness could be due to ~sinhibition or facilitation of subthreshold convergent inputs onto neurons in the reorganized cortical areas. Anatomical tracer studies revealed that thalamic and cortical circuits necessary to support such a mechanism of reorganization does, in fact, exist in the raccoon brain (9, 10). The glabrous subdivision of each cortical digit zone receives highly specific thalamic projections from one subnucleus of VPL and sparse intracortical projections from adjoining cortical digit areas. In contrast, each heterogeneous subdivision receives convergent thaiamic projections from several VPL subnuclei as well as other thalamic regions, and extensive intracortical projections from neighboring heterogeneous cortical sectors. Unmasking of weak convergent inputs via thalamocortical or intracortical routes to a partially deafferented cortical digit zone-possibly through a relay in the heterogeneous subdivisions-could account for many aspects of denervation-induced cerebral plasticity. To determine whether this notion is correct requires a detailed understanding of the physiological properties

CONVERGENT

INPUTS

TO SmI CORTEX

of the glabrous and heterogeneous cortical subdivisions. The distinction between these two sectors was based originally on ~‘multiunit” electrophysiolo~~al recordings (16) and has been confirmed in several studies (15, 19, 20, 24). However, differences in the physiological characteristics of individual neurons located in these sectors have not been explored. The purpose of the present study was to address this issue by comparing the response properties of single cells in the glabrous and heterogeneous subdivisions of a single forepaw digit representation within SmI cortex of the raccoon. Preliminary experiments had shown that most neurons in the cortical areas for digit 3 and digit 4 received strong facilitatory or weak excitatory inputs from at least two off-focus digits adjoining the principal on-focus digit (8). Electrical stimulation of the digits using a ~on~tioning-testing paradigm revealed that the strongest facilitation generally occurred with condition-test intervals of l-5 ms and that off-focus excitatory responses typically had lower probabilities and longer latencies than on-focus excitatory responses. Neurons located in the heterogeneous sectors generally had larger RFs and received convergent facilitatory or excitatory inputs from more digits than neurons in the glabrous sectors. In the present investigation, these initial observations were extended to examine the characteristics of excitatory responses in greater detail. Two major features were studied: (i) the extent of convergence of excitatory inputs onto single neurons from different digits and (ii) differences in several properties of the neuronal responses inclu~ng response probability, number of spikes per response, and response latency. In addition, the physiological characteristics of digit 3 cortex of normal raccoons were compared with preliminary results from digit 3 cortex of raccoons which had previously undergone denervation of the third forepaw digit. METHODS

Electrophysiolo~caI data were obtained from nine normal adult raccoons (Procyon lotor) of either sex, weighing between 3.2 and 5.9 kg. Comparable data were obtained from two adult raccoons whose third digit of the left forepaw had been denervated by amputation about 9 months prior to study; details of the denervation procedure are given in a previous paper (19). Each animal was sedated with ketamine HCl(15 mg/kg, im), anesthetized with cY-chloralose (25-30 mg/kg, ip), and then intubated with a tracheal cannula and artificially respired with room air, The right femoral vein was cannulated to administer supplementary doses of chloralose (20-30 mg), dextrose-saline for replacing body fluids, and tubocurarine chloride (1.5-2.0 mg) for immobilizing the animal during electrical stimulation of the digits (see below). The femoral artery was cannulated to monitor arterial blood pressure. The animal’s head was supported in a stereotaxic frame, and the so-

OF RACCOON

251

matosensory forepaw cortex of the right hemisphere was surgically exposed and photographed. A mixture of Vaseline and mineral oil was spread over the exposed cortical surface to prevent dessication. Brain movement was minimized by draining the cisterna magna and performing a bilateral pneumothorax. Core body temperature was maintained at 37-38°C with a servoregulated DC heating pad placed beneath the animal. Evoked potentials were recorded from the surface of digit 3 cortex and adjoining cortical zones with silver ball electrodes to assess the responsiveness of these areas to peripheral stimulation. The responses of individual cortical neurons and, for certain measurements, clusters of neurons were recorded using tungsten microelectrodes with an impedance of 1.0-2.5 Mohms at 1 khz. The electrodes were advanced into the cortex, as normal to the surface as possible, with a depth-calibrated hydraulic microdrive. All penetrations were made to a depth of 2 mm and the recording sites were marked on enlarged photographs of the cortical surface. Neural signals were electronically amplified and monitored, recorded on magnetic tape along with voice commentary, and fed into a computerized digital signal storage and processing system. Spikes from single neurons or small clusters of neurons were led into a voltage window ~scriminator, and the output was used to generate poststimulus time histograms from which various response characteristics could be measured. Evoked potentials were averaged and analyzed for variations in configuration, amplitude, and latency. Electrical and mechanical stimulation of the forepaw digits were employed to examine the response properties of neurons in the glabrous and heterogeneous subdivisions of digit 3 cortex. Electrical shocks were delivered to each digit of the left forepaw via pairs of silver needle electrodes inserted through the skin of the tip of each digit; one needle was inserted into the medial aspect and the other into the lateral aspect of the digit tip, about 1 mm below the skin surface. Electrical stimuli consisted of 1 ma square wave pulses, 0.2 ms in duration, and delivered at a frequency of 1 Hz. For data analysis, 10 electrical shocks were given sequentially to each digit; the on-focus digit 3 was stimulated first, followed by each of the off-focus digits. Observations in addition to the 10 recorded data trials were made to ~stin~ish between any variable-latency spontaneous spikes and the fixed-latency evoked spikes, especially when the probability of firing was low. Subsequently, the natural cutaneous sensitivity of the neuron was determined by stimulation of the skin and claws with nylon or wooden probes. Conventional criteria were used to classify the cell as sensitive to skin touch, hair movement, or claw touch. The neuronal RF was mapped with a force-calibrated nylon probe at a standard suprathreshold stimulus intensity of 2 g force; the RF was then drawn on pictures of the forepaw. Finally, the neuron was categorized on the basis of its cutaneous sensitivity-glabrous

252

DOETSCH.

STONEY.

AND

HAUGE

Neuronal Depth Distributions

Cor

’

5mm

/

I

FIG. 1. Surface map of SmI cortex from one raccoon showing the location of electrode penetrations in the digit 3 area from which G and H neurons were recorded. Open circles, penetrations yielding only G neurons; solid circles, penetrations yielding only H neurons. Halffilled circle shows the location of a track with G neurons in the upper cortical layers and H neurons in deeper layers. Dotted line indicates the approximate border between glabrous and heterogeneous subdivisions of digit 3 cortex. 2,3,4, P, cortical representations of digits 2, 3,4, and palmar pads, respectively; COF, ascending branch of coronal sulcus; Tri, triradiate sulcus.

(G) if it responsed only to stimulation of glabrous skin and heterogeneous (H) if it responded exclusively or partly to stimulation of hairy skin or a claw. The following data were obtained for each neuron: (i) depth of isolation below the cortical surface, (ii) mean probability of response, (iii) mean number of spikes per response, (iv) two measures of mean latency of response, first spike latency and all spike latency, (v) submodality sensitivity, and (vi) RF location and size. For both electrical and mechanical stimulation, a convergence factor was determined based on the number of digits from which a particular neuron could be driven. This measure was used to compare the extent of input convergence for G and H neurons in digit 3 cortex. At the end of the experiment, the animal was sacrificed with an intravenous injection of pentobarbital sodium and perfused through the heart with physiological saline followed by 10% formalin.

RESULTS

A total of 91 single neurons, all recorded from SmI digit 3 cortex, were studied in detail. Fifty cells, found within the posterior portion of the digit 3 area, were classified as G neurons. The remaining 41 cells, located more anteriorly in the digit 3 zone, were classified as H neurons. Figure 1 shows a map of the cortical surface from one animal illustrating the locations of all electrode penetrations in which individual G or H neurons were recorded.

To examine the laminar distributions of G and H neurons, their locations based on the depth of recordings were compared (Fig. 2). The two distributions were quite similar; no statistically significant differences were found. Relatively few responsive cells were isolated in the upper, supragranular cortical layers (depth -=z500 pm). Most of the neurons were found in the middle layers, with a maximum concentration 1000-1250 km below the cortical surface; many other responsive cells were located in the deeper, infragranular layers. Thus, G and H neurons, although clearly segregated along the horizontal dimensions of the cortex, had very similar laminar distributions throughout the vertical dimension. Neuronal Responsiveness to Stimulation Digits

of Different

To assess the extent of input convergence for G and H cells, the percentage of neurons (%N) responding to each of the five digits was determined. The response criterion used was the occurrence of at least one evoked spike in 10 stimulus trials. The results obtained with electrical stimulation (Fig. 3A) revealed that a significantly greater proportion of H neurons responded to stimulation of off-focus digits than did G neurons (x2 tests,p < 0.001). By definition, all neurons in each of the two categories were excited by stimulation of the on-focus digit 3. However, 71-76% of H cells, but only 2022% of G cells, could be activated from digit 2 and digit 4. While 55% of H neurons received excitatory inputs from digit 1 and digit 5, only 4-6% of G neurons received such inputs. Clearly, excitatory input convergence was much greater for H neurons than for G neurons. Natural stimulation yielded qualitatively similar results (Fig. 3B). H neurons received significantly more convergent excitatory inputs following mechanical stim-

C Neurons 0

H Neurons

g 20 E 5

10

2 0

0

250

500

750 1000 I DEPTH (pm)

FIG. 2. Depth distributions of recorded G and H neurons in the digit 3 area of SmI cortex. The percentage of 50 G and 41 H neurons isolated within each 250 pm depth increment is shown.

CONVERGENT

INPUTS

TO

C Neurons [7

SmI

CORTEX

2 DIGIT

FIG.

3.

Percentage

of G and H neurons

3

4

H Neurons

5

responded

I 1

2

3

DIGIT

STIMULATED that

253

RACCOON

B

01

OF

to stimulation

ulation of off-focus digits than did G neurons (Fisher’s exact test, p < 0.04; x2 tests, p < 0.001). All G and H cells responded to mechanical as well as to electrical stimulation of digit 3. About 32-34% of H neurons, but only one G neuron, received excitatory inputs from digit 2 and digit 4. While lo-12% of H cells were activated from digit 1 and digit 5, no G cells were excited. The proportion of both sets of neurons activated from off-focus digits was considerably greater for electrical than for natural stimulation. Despite this quantitative difference, the profiles of responsiveness to stimulation of the five digits were qualitatively very similar for the two stimulus conditions. %N for each off-focus digit decreased symmetrically with distance of that digit from the central digit 3. The percentages for digit 2 and digit 4 were nearly identical and those for digit 1 and digit 5 were very similar but always lower than those for digits 2 and 4. This symmetrical pattern of responsiveness to the five digits was common to several of the response properties examined in the present study. Hence, it was found useful to apply the term “equivalent digits” to those pairs of digits that are equidistant from the on-focus digit 3-namely, digits 2 and 4 and digits 1 and 5. The term “nonequivalent digits” was used to refer to all other pairs of digits, e.g., digits 3 and 4 or digits 2 and 5. Convergence Factor To quantify the extent of convergent input to G and H neurons, a convergence factor (CF) was determined for each cell based on the total number of digits from which that cell could be excited. The percentages of G and H neurons responding to different numbers of digits, the CFs, were then compared for electrical and mechanical stimulation. The results were entirely compatible with those illustrated in Fig. 3. With electrical stimulation (Fig. 4A), H neurons showed much higher CFs than G neurons; 49% of H cells had a CF = 5, while 72% of G cells had a CF = 1. The CF distributions for the two sets of neurons were skewed in opposite directions, and

of each digit.

(A) Electrical

4

5

STIMULATED stimulation;

(B) mechanical

stimulation.

the differences between them were highly significant (Wilcoxon’s rank sum test, p < 0.001). The results obtained with natural stimulation (Fig. 4B) were qualitatively similar; H cells had higher CFs for mechanical stimulation than did G cells. About 42% of H neurons had CFs > 1, while only one G neuron had a CF greater than unity. These differences were statistically significant (Fisher’s exact test, p < 0.001). CFs of both G and H neurons were considerably greater for electrical than for natural stimulation. However, the results obtained under both stimulus conditions were consistent in demonstrating that H neurons received significantly more convergent excitatory inputs than G neurons.

Probability of Response Several different measures were used to examine the pattern of excitation of those G and H neurons which responded to stimulation of on-focus and off-focus digits. The synaptic accessibility of excitatory input from a digit to any one neuron was assessedby measuring the mean probability of response (P) to stimulation of each digit over 10 stimulus trials. The P values of G and H cells associated with each digit were compared as shown in Fig. 5A; a minimum P = 0.1 for any given digit was required to be included in this analysis. The two sets of neurons had very similar probabilities of response to stimulation of any particular digit. Although the P values of H cells were consistently greater than those of G cells, these differences were statistically significant only for digit 2 (t test, p < 0.03). In contrast, each set of neurons showed considerable differences in P between certain nonequivalent digits. Both H and G neurons had a P value for stimulation of the on-focus digit 3 of 0.9, greater than P for any off-focus digit. The response probabilities for the off-focus digits decreased systematically with distance of these digits from the central digit 3. The P values associated with digits 2 and

254

DOETSCH,

1

FIG. 4. stimulation

2 3 CONVERGENCE

4 FACTOR

STONEY,

AND

5

1

Extent of convergent input from the digits to G and H neurons. of 1,2,3, 4, or 5 digits. (A) Electrical stimulation; (B) mechanical

4 were 0.7 for H neurons and 0.4-0.5 for G neurons; the comparable values for digits 1 and 5 were 0.4-0.5 (H cells) and 0.4 (G cells). Neither set of neurons showed significant differences in p between any of these equivalent digits. However, the Pvalues for H cells were significantly different between all nonequivalent digits (t tests, p < 0.001) except digits 2 and 5. A similar trend was evident for G cells, which showed significant differences in P between digit 3 and both digits 2 and 4 (t tests, p < 0.006). No significant differences were observed for any digit pairs involving digit 1 or digit 5, presumably due to the low percentage of G neurons responding to these digits. These results indicated that the synaptic accessibility of excitatory off-focus inputs to H neurons declined symmetrically with distance of each digit from the central digit 3. G neurons exhibited a similar but steeper decline in accessibility to excitatory drive from digit 2 and digit 4. These findings were consistent with the sym-

The graphs stimulation.

2 3 CONVERGENCE show the percentage

4 FACTOR of neurons

metrical profiles of responsiveness shown in Fig. 3A. Number

5

that

responded

to different

of Spikes per Response

The strength of excitatory input to each G and H neuron responding to digit stimulation (P > 0.1) was estimated by measuring the mean number of spikes per response (SIR) produced in 10 stimulus trials. The SIR values of G and H cells were compared across digits and are shown in Fig. 5B. Surprisingly, the mean number of spikes per response of G and H neurons, ranging from 1.2 to 1.6, was nearly identical for all five digits. No significant differences in SIR were found within or between the two sets of neurons for any two digits. Thus, the symmetrical decline in P of G and H neurons to inputs from different digits (Fig. 5A) was independent of S/R (Fig. 5B). This indicated that differences in the synaptic accessibility of off-focus excitatory drive to

C Neurons 0

DlGlT Mean (+SEM) of each digit.

2.0

H Neurons

2

DIGIT

STIMULATED

probability of response (A) and mean N > 10 except for G neuron responses

to

digits

B

A

FIG. 6. stimulation

HAUGE

(&SEM) to digits

number of spikes per response 1 and 5 where N = 2-3.

STIMULATED (B) of G and H neurons

to electrical

CONVERGENT

INPUTS

TO

A 7t

CORTEX

OF

255

RACCOON

B 35

z

35

Ji 30 c 3“w 25

d

30 25

w sf ci

E * 9

20 ii Ez 0-J 15

20

6 !z k

15

3

10 1

2 DIGIT

FIG. 6. stimulation

SmI

3 4 STIMULATED

10 1

5

2 DIGIT

3 4 STIMULATED

Mean (?&EM) first spike latencies (A) and mean (?SEM) all spike latencies (B) of responses of each digit. N 2 10 except for G neuron responses to digits 1 and 5 where N = 2-3.

these cortical corresponding atory drive.

neurons were not closely differences in the strength

coupled with of that excit-

Latency of Response The relative timing of the spikes elicited in G and H neurons excited by digit stimulation (P 3 0.1) was determined with two different measures of mean latency of response (L) over 10 stimulus trials. The “first spike L,” was the latency of the first spike occurring in each response, averaged across all responsive trials; the “all spike L” was the mean latency of all spikes occurring in a response, averaged across all responsive trials. The two measures of L for the responses produced by stimulation of each digit are shown in Figs. 6A and 6B. The two sets of neurons had similar latencies of response, for each measure of L, to stimulation of any one digit. The latencies of on-focus digit 3 responses were slightly but consistently shorter for G cells than H cells, while those of off-focus responses were uniformly shorter for H cells. However, as with the other response properties examined, the differences in L between G and H neurons for any one digit were not statistically significant. Thus, the timing of the responses to excitatory input from a given digit was, on the average, indistin~ishable for the two sets of neurons. In contrast, significant differences in L were found between certain nonequivalent digits. H neurons had L values for stimulation of on-focus digit 3 that were considerably lower than the L values for any off-focus digit (t tests, p < 0.001). Furthermore, the response latencies associated with off-focus digits varied systematically with their distance from the central digit 3, exhibiting a pattern similar to that observed for P. L values associated with equivalent digits 2 and 4 as well as digits 1 and 5 did not differ significantly, while the two measures of L for nonequivalent digits were significantly different (t tests, p -K 0.04), with only two exceptions.

of G and H neurons

5

to electrical

For example, the all spike latencies for digits 2 and 4 were about 5 ms longer, and those for digits 1 and 5 were about 8 ms longer, than the latencies for digit 3. G neurons tended to show the same systematic profile of increasing L with distance of an off-focus digit from the central digit 3. The all spike latencies for digit 3 were significantly shorter than those for digits 2 and 4 by about 7 ms (t tests, p < 0.001). The L values for digit 3 were lower than those for digits 1 and 5 by about 11-14 ms, but these differences were not statistically significant. Finally, the all spike latencies associated with equivalent digits 1 and 5 were not significantly different from each other nor from any other latencies; this was partly due to the small percentage of G neurons responding to digits 1 and 5. These findings revealed a gradient of timing in the arrival of excitatory inputs to H and G neurons from different digits, with digit 3 having the shortest conduction time followed by equivalent digits 2 and 4 and then digits 1 and 5.

Naturat

Cutaneous Sensitivity

By definition, G and H neurons differed significantly in their sensitivity to various types of natural stimulation. All G cells responded to light touch of the glabrous skin of digit 3. The RFs of these cells were usually small and confined to the skin of digit 3, except for one neuron which also responded to stimulation of digits 2 and 4. The vast majority of H cells displayed mixtures of cutaneous sensitivities, i.e., touch on the hairy or glabrous surface of a digit, touch or tap of a claw, or movement of hairs on the dorsal surface of a digit. Only three H neurons were exclusively sensitive to only one submodality, namely claw touch. RFs of H cells typically were much larger than those of G cells, as expected from their higher CF; about 42% of the fields involved one or more off-focus digit as well as digit 3.

256 Variation

DOETSCH,

in Response Properties

STONEY,

with Cortical Depth

Statistical analyses were performed to examine whether the neuronal response properties varied systematically with depth below the cortical surface. With one exception, no clear pattern of laminaf differences emerged for either set of neurons. H cells showed significant differences in CF as a function of depth for electrical stimulation of the digits. The mean CF increased with cortical depth (r = 0.34, p < 0.04) and was greatest for those cells located in the infra~anular layers, 12002000 pm below the cortical surface. Comparison of Response Properties Deneruated Animals

in Normal

and

It was of interest to determine whether strengthening the normal pattern of off-focus inputs to digit 3 cortex could partly account for cortical reorganization following digit denervation (19,20). Hence, a comparison was made of the first spike and all spike latencies of responses obtained from small clusters of cortical neurons in two normal and denervated raccoons. No distinction was made between glabrous and heterogeneous subdivisions of the digit 3 zone because the boundary between these sectors is obliterated in reorganized cortex (19). Both measures of L suggested that off-focus response latencies in reorganized cortex had a symmetrical profile similar to that of off-focus response latencies in normal cortex. The all spike latencies for equivalent digits 2 and 4 were about l-4 ms shorter than those for digits 1 and 5. In comparing both measures of L for the denervated and normal animals, latencies of all off-focus responses in reorganized cortex were longer than latenties of on-focus digit 3 responses in normal cortex. After denervation, the all spike latencies for equivalent digits 2 and 4 were about l-2 ms longer, and those for digits 1 and 5 were about 3-5 ms longer, than L of normal digit 3 responses. In contrast, off-focus latencies for reorganized cortex clearly tended to be shorter than off-focus latencies for normal cortex. Following denervation, the all spike latencies for equivalent digits 2 and 4 were about 2-4 ms shorter, and those for digits 1 and 5 were about l-5 ms shorter, than the corresponding latencies for normal cortex. These preliminary findings indicated that the latenties of off-focus responses in the reorganized cortical zone were intermediate between those of on-focus digit 3 responses and off-focus responses in the normal cortical zone. These results were in agreement with data obtained from evoked potential recordings; the latencies of the positive peak of off-focus potentials in reorganized cortex were equal to or shorter than the corresponding latencies in normal cortex, but longer than the latencies for on-focus potentials in normal cortex. Thus, the pattern of off-focus latencies appeared to be similar in denervated and normal animals but shifted in time, suggesting that the synaptic efficacy of off-focus

AND HAUGE

excitatory vation.

inputs had increased after peripheral

dener-

DISCUSSION The results of the present study showed that single neurons in the heterogeneous subdivision of SmI digit 3 cortex of the raccoon received considerably more convergent inputs from the digits than did neurons in the glabrous subdivision of this cortical area. The extent of input convergence for both G and H neurons was significantly greater when measured with electrical than with natural stimulation, presumably reflecting the greater synaptic efficacy of synchronous afferent volleys produced by electrical stimulus pulses (6, 13, 29, 30, 33, 44,45). The pattern of convergent inputs from the digits to each of the two sets of cortical neurons was symmetrical and consistent, as reflected by the variations in (i) the percentage of neurons responding and the probability of responding, and (ii) the latency of responding to stimulation of different digits. The values of these response measures varied systematically with distance of each off-focus digit from digit 3, indicative of corresponding variations in the accessibility and conduction time of the off-focus synaptic drive onto G and H neurons. This unique profile of response properties-although characteristic of both sets of neurons-was more pronounced for H cells because a larger proportion of these cells responded to off-focus stimulation. Curiously, the strength of the inputs, as estimated by the number of spikes per response, did not vary accordingly, suggesting that accessibility may be regulated independently of strength of synaptic drive in the convergence pathway. A possible, albeit unlikely, explanation for this pattern of off-focus inputs to digit 3 cortex could be slower conduction in the pathways from each of the other digits to their respective cortical zones. “Lateral” spread of activity from these pathways to neurons of the digit 3 cortical area would then produce off-focus responses of longer latency. To rule out this possibility, on-focus response latencies were measured for clusters of neurons in each of the five cortical digit zones; as expected, the L values for these zones were virtually identical. Thus, the longer latency inputs from digits 1,2,4, and 5 to digit 3 cortex must be due to longer or more slowly conducting afferent pathways that are unique to the off-focus but not the on-focus inputs. Relatively little information is available on the origin of convergent excitatory or facilitatory inputs to SmI cortex. Strong evidence for extensive off-focus convergent effects has been obtained with peripheral electrical stimulation and con~tioning-testing interactions to detect weak or subthreshold influences on cortical neurons (6,29,30,44,45). Some of these inputs must originate in VPL nucleus of thalamus, since this region contains subsets of thalamocortical projection neurons

CONVERGENT

INPUTS

TO

with distinctly convergent properties (13,33). The idea that other convergent excitatory or facilitatory inputs may arrive via intracortical routes is consistent with anatomical (9) and electrophysiological data (‘23, 44) and also with pharmacological data on the effects of blocking local inhibitor influences. For example, iontophoretie administration of the glycine antagonist, strychnine (31), or the GABA antagonists, picrotoxin (2) and bicuculline (12), to SmI cortex of cats can enhance the responsiveness or enlarge the RFs of cortical neurons, particularly of rapidly adapting cells. In contrast, application of bicuculline to neurons in VPL seems to have little effect on RF size, suggesting a minor role for GABA-mediated RF control in this thalamic relay (14). Finally, Zarzecki, Rasmusson, and their colleagues (28, 42) found that many “glabrous” neurons in the forepaw digit 4 cortex of raccoons receive facilitatory or excitatory inputs via corticocortical projections from “heterogeneous” neurons-intracortical microstimulation of heterogeneous areas elicited monosynaptic EPSPs in at least 50% of neurons in the glabrous digit 4 area. The results of these studies are compatible with the view that intracortical projections could provide excitatory or facilitatory off-focus inputs to a particular cortical area. However, this does not preclude the possibility that some distant intracortical interactions are inhibitory. Evidence from recent studies suggests that neurons in normal SmI cortex of raccoons (7) and neurons in reorganized cortical areas of spinal cats (4) may be subject to tonic inhibitory influences that can be reduced or eliminated by restricted lesions of specific SmI cortical regions. The digit representations in SmI cortex of the raccoon show some interesting similarities to the vibrissa representations in the “barrel” cortex of rats and mice (1,27,38,43). Each barrel constitutes the cortical target of on-focus input from one specific vibrissa, but also receives inputs from neighboring off-focus vibrissae. Furthermore, the barrel is subdivided into two distinct regions, a central hollow and the surrounding wallfsepturn, which exhibit different physiological characteristics. According to Armstrong-James and Fox (l), input from the on-focus vibrissa projects strongly to neurons in both the hollow and septum, while inputs from off-focus vibrissae are preferentially directed to neurons in the septal regions; the off-focus responses are “overwhelmingly of long latency” compared with on-focus responses. The similarities between the two SmI cortical regions in the rodent and raccoon-barrel hollows to glabrous sectors and barrel septa to heterogeneous sectors-suggest that some common principles of organization apply. (For a discussion of other features shared by SmI cortex of the rat and raccoon, see p. 1897 of Ref. (9)). A major objective of the present study was to determine whether SmI cortical neurons in the raccoon normally receive off-focus inputs that might account for

SmI

CORTEX

OF

RACCOON

257

some aspects of denervation-induced cortical reorganization. The preliminary results obtained from denervated animals suggested that the ~u~~er~ of convergent inputs from off-focus digits was similar to that in normal animals, but that the latencies of off-focus responses in reorganized cortex were shorter than the corresponding latencies in normal cortex. Zarzecki and Rasmusson (personal communication) recently obtained comparable results from forepaw digit 4 cortex of the raccoon. About 6-10 months after denervation of the fourth digit, EPSPs elicited in “digit 4” cortical neurons by stimulation of off-focus digits 3 and 5 had shorter peak latencies than normal. These findings support the notion that the neural circuitry responsible for novel inputs to reorganized SmI cortex is present in the normal brain and may be stren~hened (unmasked) by denervation. The specific mechanisms of cortical reorganization are not known, but are likely to include d&inhibition and facilitation of existing input pathways to the affected neurons (5,18,19,21). Dykes and Metherate (11) have emphasized that partial deafferentation of SmI cortex increases the excitability of cortical neurons by release from inhibition and also enhances neuronal responsiveness to acetylcholine. They suggest that increased sensitivity to acetylcholine may permit strengthening of active (new) excitatory inputs and that other modulatory agents, such as norepinephrine, may decrease the synaptic efficacy of latent inputs. In the raccoon, neural circuits that could mediate such processes and also account for the pattern of on- and offfocus responses must include both G and H neurons. Afferent projections from the glabrous skin and dorsal hairy skin of each digit (32) are largely segregated at all levels of the raccoon somatosensory system including the cuneate nucleus (17,25), VPL nucleus of thalamus (10,26, 36, 37,39,41), and SmI cortex (10,15,16, 19, 20, 24, 40) where the glabrous and heterogeneous sectors of each digit area are large and distinct, as confirmed in the present study. Each glabrous subdivision receives discrete thalamic projections from VPL and has sparse intracortical connections, while the heterogeneous subdivision receives convergent thalamic projections and has extensive intracortical connections (9, 10). It is likely that each heterogeneous sector can be subdivided further into discrete regions representing claw, hairy skin, and mixtures of skin types (lo), but such parcellation was not attempted here. The two different projection patterns readily account for the physiological differences between G and H neurons, including the larger proportion of H cells responding to off-focus stimulation and their correspondingly higher CF and P values. The symmetrical variations in response latency of H neurons to off-focus inputs could be explained by systematic differences in the conduction distance of afferent pathways from the digits or of lateral connections from neighboring cortical digit areas; varia-

258

DOETSCH,

STONEY.

tions in conduction velocity, especially of horizonal or collateral fibers, may also be involved. It is likely that cortical G neurons receive some of their off-focus excitatory or facilitatory inputs from H neurons via intracortical routes (9, 28, 42). If there is such a relay in the heterogeneous sector, then the offfocus latencies of H cells should be shorter than those of G cells. This was, in fact, a consistent finding for inputs from all off-focus digits, but the differences in mean latencies between G and H neurons were not statistically significant. However, it is noteworthy that the probability of H neurons having latencies shorter than those of G neurons for all four off-focus digits was 0.07 (Sign test). A larger sample of cells or finer temporal resolution of latencies might well show the expected differences. Such a finding would not rule out the possibility that off-focus influences on G neurons also arise from other sets of cortical neurons as well as from thalamic sources. If the off-focus inputs to a glabrous sector are under tonic inhibitory control by certain “G interneurons,” then cortical reorganization produced by peripheral denervation could be explained as the release of those inputs from inhibition. Thus, removal of on-focus input to all G neurons in digit 3 cortex would disinhibit, and perhaps facilitate, the off-focus inputs originating from H neurons, causing the appearance of novel responses in “digit 3” G neurons. The functional properties of the reorganized cortical zone-loss of somatotopy, larger neuronal RFs, mixed responsiveness to glabrous and hairy skin stimulation-could be explained by an H neuron source for many of the new inputs. The profile of response latencies for digit stimulation is also consistent with this interpretation; the lower L values after denervation may be due to increased synaptic efficacy of previously existing, but weak H inputs. Recent evidence obtained by Zarzecki, Rasmusson, and their co-workers (28,42) indicates that denervation of a forepaw digit in the raccoon may increase the strength of corticocortical EPSPs produced in “glabrous” neurons by microstimulation of “heterogeneous” neurons. More studies are required to test specific aspects of this model. ACKNOWLEDGMENTS We express our appreciation to Drs. D. D. Rasmusson, R. H. Ray, A. L. Towe, and P. Zarzecki for their helpful comments on an early version of the manuscript, and to Mark Litaker for his assistance with statistical analysis of the data. We also thank Max Blanc0 and Ernestine Jones for their technical assistance during the experiments and Beulah Collins for typing the manuscript. This work was supported in part by NSF Grant BSN-8419035 to G.S.D.

REFERENCES 1.

ARMSTRONG-JAMES M., AND K. Fox. 1987. Spatiotemporal convergence and divergence in the rat Sl “barrel” cortex. J. Camp. Neural. 263: 265-281.

AND

HAUGE

2.

BATUEV, A. S., A. A. AL EXANDROV, AND N. A. SCHEYNIKOV. 1982. Picrotoxin action on the receptive fields of the cat sensorimotor cortex neurons. J. Neurosci. Res. 7: 49-55.

3.

CARSON, L. V., A. M. KELAHAN, R. H. RAY, C. E. MASSEY, AND G. S. DOETSCH. 1981. Effects of early peripheral lesions on the somatotopic organization of the cerebral cortex. In Clinical Neurosurgery (P. W. Carmel, Ed.), Vol. 28, Chap. 30, pp. 532-546. Williams & Wilkins, Baltimore.

4.

CASANOVA, C., P. A. MCKINLEY, AND S. MOLOTCHNIKOFF. 1991. Responsiveness of reorganized primary somatosensory (SI) cortex after local inactivation of normal SI cortex in chronic spinal cats. Somatosens. Motor Res. 8: 65-76.

5.

DEVOR, M. 1983. Plasticity of spinal cord somatotopy in adult mammals: Involvement of relatively ineffective synapses. In Nervous System Regeneration (B. Haber, J. R. Perez-Polo, G. A. Hashim, and A. M. G. Stella, Eds.), pp. 287-314. A. R. Liss, New York.

6.

DOETSCH, G. S., AND A. L. TOWE. 1976. Response distinct neuronal subsets in hindlimb sensorimotor tex of the domestic cat. Exp. Neural. 53: 520-547.

I.

DOETSCH, G. S., K. W. JOHNSTON, AND C. J. HANNAN, JR. 1990. Physiological changes in the somatosensory forepaw cerebral cortex of adult raccoons following lesions of a single cortical digit representation. Exp. Neural. 108: 162-175.

8.

DOETSCH, G. S., S. D. STONEY, JR., AND D. H. HAUGE. 1989. Convergent inputs to single neurons in SmI forepaw digit cortex of raccoons. Sot. Neurosci. Abstr. 15: 1053. DOETSCH, G. S., G. P. STANDAGE, K. W. JOHNSTON, AND C.-S. LIN. 1988. Intracortical connections of two functional subdivisions of the somatosensory forepaw cerebral cortex of the raccoon. J. Neurosci. 8: 1887-1900.

9.

10.

11.

12.

properties cerebral

of cor-

DOETSCH, G. S., G. P. STANDACE, K. W. JOHNSTON, AND C.-S. LIN. 1988. Thalamic connections of two functional subdivisions of the somatosensory forepaw cerebral cortex of the raccoon. J. Neurosci. 8: 1873-1886. DYES, R. W., AND R. METHERATE. 1988. Sensory cortical reorganization following peripheral nerve injury. In Brain Injury and Recovery (S. Finger, T. E. Levere, C. R. Almli, and D. G. Stein, Eds.), Chap. 15, pp. 215-234. Plenum, New York. DYKES, R. W., P. LANDRY, R. METHERATE, AND T. P. HICKS. 1984. Functional role of GABA in cat primary somatosensory cortex: Shaping receptive fields of cortical neurons. J. Neurophysiol. 62: 1066-1093.

13.

HARRIS, sensory

14.

HICKS, T. P., R. METHEF~ATE, P. LANDRY, AND R. W. DYKES. 1986. Bicuculline-induced alterations of response properties in functionally identified ventroposterior thalamic neurones. Exp. Brain Res. 63: 248-264. JOHNSON, J. I. 1985. Thalamocortical organization in the raccoon: Comparison with the primate. Exp. Brain Res., Suppl. 10: 294-312. JOHNSON, J. I., E.-M. OSTAPOFF, AND S. WARACH. 1982. The anterior border zones of primary somatic sensory (SI) neocortex and their relation to cerebral convolutions, shown by micromapping of peripheral projections to the region of the fourth forepaw digit representation in raccoons. Neuroscience 7: 915-936. JOHNSON, JR., J. I., W. I. WELKER, AND B. H. PUBOLS, JR. 1968. Somatotopic organization of raccoon dorsal column nuclei. J. Comp. Neural. 132: l-44. KAAS, J. H., M. M. MERZENICH, AND H. P. KILLACKEY. 1983. The reorganization of somatosensory cortex following peripheral nerve damage in adult and developing mammals. Annu. Rev. Neurosci. 6: 325-356. KELAHAN, A. M., AND G. S. DOETSCH. 1984. Time-dependent

15.

16.

17.

18.

19.

F. A. 1978. Functional subsets of neurons in somatothaiamus of the cat. Exp. Neurol. 58: 149-170.

CONVERGENT

20.

21.

22.

23.

24.

25.

26.

27. 28.

30.

31.

32.

TO SmI CORTEX

changes in the functional organization of somatosensory cerebral cortex following digit amputation in adult raccoons. Sonatosens. Res. 2: 49-81. KELAHAN, A. M., R. H. RAY, L. V. CARSON, C. E. MASSEY, AND G. S. DOETSCH. 1981. Functional reorganization of adult raccoon somatosensory eerbral cortex following neonatal digit amputation. Brain Res. 223: 151-159. MERRILL, E. G., AND P. D. WALL. 1978. Plasticity of connection in the adult nervous system. In Neuronal Plasticity (C. W. Cotman, Ed.), pp. 97-111. Raven, New York. MERZENICH, M. M., G. RECANZONE, W. M. JENKINS, T. T. ALLARD, AND R. J. N~o. 1988. Cortical representational plasticity. In ~eurob~o~gy of Neocor~e~ (P. Rakic and W. Singer, Eds.), pp. 41-67. Wiley, New York. MURRAY, H. G., AND C. J. HEATH. 1978. The spread of connections within cat primary somatic sensory cortex. Brain Res. 141:160-164. RASMUSSON, D. D. 1982. Reorganization of raccoon somatosensory cortex following removal of the fifth digit. J. Comp. Neural.

33.

34.

35.

36.

OF RACCOON

ZSY

vation of the raccoon forepaw: 3. Cutaneous domains and organization of nerves innervating the raccoon forepaw. Somatosens. Res. 5: 19-33. WALDRON, J. N. B., S. GHOSH, AND P. ZARZECKI. 1989. Multiple inputs to a population of thalamocortical neurons projecting to cat somatosensory cortex. Exp. Brain Res. 74: 105-115. WALL, J. T. 1988. Variable organization in cortical maps of the skin as an indication of the lifelong adaptive capacities of eircuits in the mammalian brain. Trends Neurosci. 11: 549-557. WALL, J. T., AND J. H. KA~s. 1985. Cortical reorganization and sensory recovery following nerve damage and regeneration. In Synaptic Plasticity (C. W. Cotman, Ed.), pp. 231-260. Guilford, New York. WARREN, S., AND B. H. PUBOLS, JR. 1984. Somatosensory thalamocortical connections in the raccoon: An HRP study. J. Comp. Neurol.

227:

597-606.

WARREN, S., A. M. KELAHAN, AND B. H. PUBOLS, JR. 1986. The somatosensory thalamus of the raccoon: Properties of single neurons responsive to light mechanical stimulation of the forepaw.

205:313-326.

J. Neurosci.

RASMUSSON, D. D. 1989. The projection pattern of forepaw nerves to the cuneate nucleus of the raccoon. Neurosci. Lett. 98: 129-134. RASMUSSON, D. D., AND D. M. NANCE. 1986. Non-overlapping thaiamocortical projections for separate forepaw digits before and after cortical reorg~ization in the raccoon, Brain Res. Bull. 16:399-406. SIMONS, D. J. 1978. Response properties of vibrissa units in rat SI somatosensory neocortex. J. Neurophysiol. 41: 798-820. SMITS, E., D. C. GORDON, P. ZARZECKI, AND D. D. RASMUSSON. 1989. Plasticity of raccoon somatosensory cortex: Roles of convergent sensory inputs and corticocortical EPSPs. Sot. Neuro-

WELKER, C. 1976. Receptive fields of barrels in the somatosensory neocortex of the rat. J. Comp. Neurol. 166: 173-190. WELKER, W. I., AND J. I. JOHNSON, JR. 1965. Correlation between nuclear morphology and somatotopic organization in ventro-basal complex of the raccoon’s thalamus. J. Anat. 99: 761790. WELKER, W. I., AND S. SEIDENSTEIN. 1959. Somatic sensory representation in the cerebral cortex of the raccoon (Procyon lotar). J. camp. Neural. 111: 469-501. WIENER, S. I., J. I. JOHNSON, AND E.-M. OSTAPOFF. 1987. Organization of postcranial kinesthetic projections to the ventrobasal thalamus in raccoons. J. Comp. Neural. 258: 496-508. WITTE, S., E, S&ITS, P. ZARZECKI, AND D. D. RASMUSSON. 1990. New somatosensory EPSPs in reorganized raccoon cortex without change in the incidence of corticocortical EPSPs. Sot. Neurosci. Abstr. 16: 44. WOOLSEY, T. A. 1978. Some anatomical bases of cortical somatotopic organization. Brain Behau. Evol. 15: 325-371. ZARZECKI, P. 1986. Functions of corticocortical neurons of somatosensory, motor, and parietal cortex. In Cerebral Cortex (E. G. Jones and A. Peters, Eds.), Vol. 5, Chap. 5, pp. 185-215. Plenum, New York. ZARZECKI, P., AND D. M. WIGGIN. 1982. Convergence of sensory inputs upon projection neurons of somatosensory cortex. Exp. Brain Res. 48: 28-42.

sci. Abstr. 29.

INPUTS

15: 1052.

Town, A. L., AND V. E. A~SSIAN. 1958. Patterns of activity in single cortical units following stimulation of the digits in monkeys. J. Neurophysiol. 21: 292-311. TOWE, A. L., 13. WHITEHORN, AND J. K. NYQUIST. 1968. Differential activity among wide-field neurons of the cat postcruciate cerebral cortex. Exp. Neural. 20: 497-521. TREMBLAE’, N., R. WARREN, AND R. W. DYKES. 1988. The effects of strychnine on neurons in cat somatosensory cortex and its interaction with the inhibitory amino acids, glycine, taurine and fl-alanine. Neuroscience 26: 745-762. TURNBULL, B. G., AND D. D. RASMUSSON. 1987. Sensory inner-

39.

42.

43. 44.

45.

6: 308-317.