Population Analysis of Single Neurons in Cat Somatosensory Cortex Richard A. Warren*.' and Robert W. Dykest?' *Department of Neurology and Neurosurgery, McGill Universiv, Montre'al, Que'bec, Canada; TDe'partement de Physiologie, Universite' de Montre'al, Montre'al, Que'bec, Canada
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
Abstract Single neurons in the somatosensorycortex are divisible into a population with receptive fields and a population without receptive fields. These two populations
display different laminar distributions,and their respective functions are unknown. We compared other physiological characteristics of these two neuronal populations in an attempt to understand why some neurons lack a receptive field. Only 23% of 465 neurons isolated in the somatosensory cortex of halothane-anesthetized cats displayed a cutaneous receptive field. The iontophoretic administration of glutamate uncovered input from the periphery in another 34% of the sample, leaving 43% of the neurons without evidence of peripheral input under these experimental conditions. Neurons with a receptive field were spontaneously active much more often than neurons lacking peripheral inputs, and their rates of discharge were higher. No differences were found between neurons having a receptive field uncovered with glutamate and those unaffected by glutamate. In all classes of neurons, those cells with spontaneous activity were excited by smaller amounts of glutamate than were silent neurons, but sensitivityto glutamate was not correlated with the presence or absence of a receptive field. We infer that some classes of somatosensory cortical neurons receive strong thalamocortical inputs, whereas others have only relatively weak or no thalamocortical connections. In other experiments we have shown also that those neurons lacking a receptive field and/ or spontaneous activity were more likely to be plastic than those with stronger inputs (see Warren and Dykes, 1993a,b), suggesting that neurons having weaker afferent inputs can be more readily modified under certain circumstances. Key words somatosensory cortex, cat, iontophoresis, cortical neurons,
receptive field
Although there has been a resurgence of interest in the details of somatotopy inprimary somatosensory cortex since the advent of the micromapping technique (Welker, 1971), there are few studies of the cellular composition of this region that is so important for normal tactile function and normal sensory gnosis (Randolf and Semmes, 1974). One intriguing aspect of this region is the fact that a large number of cells in primary somatosensory cortex lack receptive fields (Dykes and Lamour, 1988a). Often these cells are overlooked or 1. Present address: Department of Anatomy and Neurobiology, University of California at Irvine, Irvine, California 92717. 2. To whom all correspondence should be addressed, at Dkpartement de Physiologie, Faculte de Mtdecine, Universitt de Montreal, C.P. 6128, Succursale A, Montreal, Qutbec H3C 357, Canada.
remain undetected and are not studied because there are only a limited number of ways to activate them. One advantage offered by the multibarreled iontophoretic microelectrode is that one barrel routinely contains glutamate or another amino acid to excite and thus detect these cells. Although some of these cells may be part of a pathway from the contralateral cortex to the thalamus (Landry et al., 1984), those cells receiving callosal information can account for only a small fraction of the neurons lacking receptive fields. These cells are found in a number of different experimental conditions, including behaving animals, and it would be useful to know more about them. Routinely, somatosensory cortical neurons are also differentiated by their rates of spontaneous activity and by their receptive field
Somatosensory and Motor Research, Vol. 9, No. 4, 1992, pp. 297-312
Accepted June 12, 1992
297
WARREN AND DYKES
characteristics. The frequency with which each of these characteristics is encountered varies as a function of the laminar location of the cell being studied. As a baseline for other iontophoretic studies (Warren and Dykes 1993a,b), we compared the electrophysiological characteristics of a large sample of cortical neurons obtained with microiontophoretic pipettes containing glutamate to comparable samples described in the literature.
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
METHODS Experiments were performed on 46 adult mongrel cats of either sex. A detailed description of the animal preparation under halothane anesthesia and recording techniques has appeared elsewhere (Warren and Dykes, 1993a). Briefly, the animals were ventilated artificially with air containing 2% halothane during surgery and 1-1.5% during recording sessions. Mapping, Recording, and Classification
Following a craniotomy over the somatosensory cortex, brief mapping was performed; a carbon-fiber-in-glass microelectrode was used to locate areas responsive to stimulation of the skin of the forearm. That electrode was replaced by a combination microiontophoretic and recording electrode made with a seven-barrel glass pipette glued with light-cured dental adhesive to a single glass recording electrode. The tip of the recording electrode protruded 20-40 pm beyond the multibarrel electrode and was filled with 2% pontamine sky blue in NaC1. The multibarrel pipette contained DL-glutamate (0.15 M, pH 8.0) and NaCl (0.9%, pH 7.0) for current balancing, and different noradrenergic agonists and antagonists (Warren and Dykes, 1993a), but only the responses to glutamate are reported here. There are important limitations of the microiontophoretic technique, which have been discussed in depth by Hicks (1984) and others. The technique as practiced in this laboratory is described in Dykes et al. (1984), Metherate et al. (1988), and Warren and Dykes (1993a). Fortunately, glutamate is relatively easy to use because it is naturally present in the cortex and is removed rapidly by existing uptake mechanisms. The primary difficulty is that rapid administration of large amounts can result in depolarization block of the neurons near the ejection pipette. To avoid this problem, the experimenter must eject small, gradually increasing quantities intermittently as the electrode is advanced. Current or pH artifacts can be detected by ejection of NaC1. As the electrode was advanced, pulses of glutamate were delivered regularly to excite otherwise quiescent neurons. At the same time, the skin of the forearm was 298
stimulated with gentle somatic stimuli. Once a unit was isolated, its depth on the micrometer was noted, and the skin was searched to determine whether or not there was a receptive field. When present, the receptive field was classified as follows: 1. A receptive field was said to be located in the skin when light touch of the skin with a hand-held, firepolished glass probe and/or the flicking of a few hairs elicited a clear and reproducible response over the background activity. This criterion was used to define the boundary of the field. These cutaneous receptive fields were further characterized as being either slowly adapting (SSA) or rapidly adapting (SRA), according to whether they did or did not respond throughout the duration of a stimulus lasting several seconds. The shape and size of the receptive field were represented on a standard drawing of the cat forearm (Fig. 1). 2. A receptive field was classified as Deep when the neuron responded only to the stimulation of subcutaneous structures-that is, to the palpation of the underlying muscles or to joint movement. 3. In several cases receptive fields were classified as Tap when the unit required a light tapping of the skin to respond to every stimulus and the lower-velocity manipulations needed to differentiate between the cutaneous and deep categories were ineffective. For neurons without a receptive field, the search for a receptive field was repeated during the administration of subthreshold doses of glutamate. These neurons were classified as having no receptive field (No-RF) or as having a receptive field only in the presence of glutamate (Glut-RF).
FIGURE1. Examples of cutaneous receptive fields for single neurons isolated in this study. Receptive fields A , B, C, and D were cutaneous and rapidly adapting. The neurons were located in layer V (A), layer 111 (B), and layer IV (C). The neuron with receptive field D was not recovered in the histology, but according to the micrometer reading it was located in the middle of the cortex. Receptive field E was classified as Tap, and the neuron was not located.
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
NEURONAL POPULATIONS IN SOMATOSENSORY CORTEX
Neurons were also characterized by the presence or absence of spontaneous activity. Since spontaneous activity was usually very low, the neuron was left unstimulated for at least 1 min before a decision was made. In many cases when an ongoing discharge was present, the frequency of this activity was evaluated either by counting the action potentials with the window discriminator for at least 100 sec or by using data stored on the computer during the control period preceding a series of somatic stimuli or pulses of glutamate. All neurons displaying an ongoing discharge of less than 0.1 impulses/sec (imp/sec) were considered to be silent. For many neurons, the threshold quantity of glutamate (as measured in nanoamperes) necessary to produe a response was determined by first setting a current pulse (30 sec) that elicited a clear response, and then decreasing this current in small steps until no clear response was observed. Then the procedure was reversed: The current was increased in small steps until a just-noticeable response was detected over the background activity. The value of this current was recorded as the threshold. In some cases glutamate failed to elicit a response even with currents up to 500 nA, suggesting that these units lacked glutamate receptors and/or that they were axons. Histology When a significant number of neurons were isolated in one penetration, pontamine sky blue was ejected from the recording electrode in two locations along the electrode track by passing a current of negative polarity. Prior to perfusion of the animal, the halothane concentration in the inspired air was raised to 4% for 10 to 15 min. The animal was perfused through the ascending aorta with 0.9% saline followed by 10% buffered formalin. The brain was removed from the skull and placed in the same fixative for several days before being cryoprotected with 30% sucrose. Sections 80 pm thick were cut through the somatosensory cortex on a cryostat and mounted on gelatin-coated slides. The sections were stained for Nissl substance with cresyl violet according to standard procedures and coverslipped. The micrometer readings provided the depth of each cell as well as the depth of the pontamine sky blue ejections, allowing reconstruction of the electrode trajectory, with the knowledge of the distance between two dye spots giving the information to compensate for tissue shrinkage that occurred during the processing of the tissue. Camera lucida drawings of the sections containing the dye spots were made at X40, and each neuron was attributed to a cortical layer and cytoarchitectonic area according to the cytoarchitectonic cri-
teria of Hassler and Muhs-Clement (1964). In some cases the two dye spots could not be seen, but electrolytic lesions were present. Data Analysis Appropriate statistical tests were used whenever possible to compare the laminar distribution, the effect of glutamate, and the spontaneous activity of the different classes of neurons. The choice of the proper statistical test to be used was based on the theoretical arguments of Sokal and Rohlf (1981). To test for whether or not proportions in a sample differed from chance (i.e., a contingency table), the G statistic was used whenever the size of the sample was large enough. In some cases, classes displaying similar characteristics were pooled (e.g., laminar distribution, distribution of the frequencies of spontaneous activity and of threshold current of glutamate) to meet the prerequisite sample size of the test. In the case of 2 x 2 tables of small sample size, the Fisher’s exact test was applied according to the tables of Siege1 (1956). When the analysis of variance was used, the assumptions of that test were fulfilled. Statistical tests were complicated by the fact that the frequencies of spontaneous activity and the iontophoretic currents used were not normally distributed. Furthermore, the variances of the samples were seldom homogeneous. After a logarithmic transformation, both spontaneous activity and glutamate currents appeared to be normally distributed and to have homogeneous variances, so the analysis of variance was performed on the transformed data. The geometric means are reported instead of the usual arithmetic means, since they give a better representation of the central values of the data. The data for the activity evoked by cutaneous stimuli and the responses to glutamate were normally distributed, and their variances were homogeneous. In those cases the data did not need to be transformed; statistical tests were performed on the original data, and the arithmetic means were used. RESULTS Origin of the Sample A total of 533 single units were isolated in the somatosensory cortex. Of these, 68 were excluded from further analysis because they were not held long enough (n = 55) or they were found to be insensitive to glutamate ( n = 13), suggesting that they were fibers (although some could have been glutamate-insensitive neurons; Schneider and Perl, 1988). In the other 465 cases, the units were sensitive to glutamate and were held long 299
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
WARREN AND DYKES
enough to be tested for the presence of a receptive field (Table 1). These putative neurons were isolated along 66 electrode penetrations. In 17 Of 46 animals, the electrode tracks were reconstructed from the histological sections, and 189 neurons were located with respect to the dye spots or the lesions (Fig. 2). Only 23% of the neurons displayed a receptive field; 85 appeared to receive input from the skin, and of these, 82 were classified as SRA units, whereas only three had the characteristics of SSA units. Only 1 neuron appeared to receive input from muscle receptors, responding to the flexion of the fifth digit; it was classified as serving subcutaneous receptors (Deep). In 20 cases the modality appeared uncertain, because these neurons could have been classified as rapidly adapting neurons responding to high-threshold cutaneous stimuli or as low-threshold rapidly adapting receptors in subcutaneous tissues. Since they responded to light tapping of their receptive fields, they were classified as Tap. The receptive fields were elongated following the long axis of the limb (Fig. 1). They were also smaller on the paw than on the forearm or the arm. Overall, they were similar in shape and size to those described in previous studies (Sretavan and Dykes, 1983; Metherate et al., 1988; Temblay et al. 1990). In the remaining 77% of the sample, no evidence of peripheral input was found in the absence of drug treatments, but in 127 cases, subthreshold amounts of glutamate (ranging from 4 to 90 nA; geometric mean = 23.5 nA) were sufficient to uncover receptive fields with characteristics similar to those found in other cells in the absence of drugs. These neurons were classified as having a receptive field during glutamate iontophoresis (Glut-RF neurons). In the other 161 cases, glutamate treatment faild to uncover peripheral inputs, and these
neurons were classified as lacking any receptive field (No-RF neurons). In 71 cases, constituting 15% of the sample, no attempt was made to uncover a receptive field with glutamate. These were not considered a legitimate class, since these cases were likely to be both Glut-RF and No-RF neurons. Nevertheless, these neurons had to be accounted for in he computation of the proportions of the Glut-RF and No-RF classes by assuming that this group contained the same proportions of Glut-RF and No-RF neurons found in the sample tested with glutamate for receptive fields. Accordingly, they were distributed to the Glut-RF and No-RF groups; this yielded a proportion of 43% of the neurons without any evidence of peripheral input and 34% with receptive fields uncovered by glutamate. Consequently, a grand total of 57% of the sample appeared to receive some input from the periphery under these experimental conditions. Table 1 also compares the proportions of neurons of each class found in those penetrations located in the histology with the proportions found in those not located in the histology. There was no significant difference between these two subsets (G test, Gad, = 4.677, (df = 3, p > 0.1 ; SSA and Deep classes were excluded from the test), suggesting that the two samples belonged to the same population and that no bias was introduced by locating some neurons in the histology. As a result, the total sample represents the best estimate for the proportions of the different classes of neurons observed. Cytoarchitectonic Location of the Sample Between 5 and 19 neurons were isolated in each of the 17 reconstructed penetrations. Only 2 neurons appeared to be located in the white matter. These were discarded,
TABLE1 . Neurons Isolated in Cat Somatosensory Cortex and Tested for Sensitivity to Glutamate (n = 465) ~
Found in histology With a receptive field SRA
SSA Tap Deep Total Without a receptive field Glut-RF
No-RF Receptive field not tested with glutamate Total Note. See “Methods” for abbreviations.
300
42 (22%) 1 (0.5%) 7 (3.7%) 0 (0%)
50 (26%)
Not found in histology
Total
40 (14%) 2 (0.7%) 13 (4.7%) 1 (0.4%) 56 (20%)
82 (18%) 3 (0.6%) 20 (4.2%) 1 (0.2%) 106 (23%)
78 (35%)
49 (33%) 61 (41%)
loo (45%)
127 (34%) 161 (43%)
(29) 139 (74%)
(421 220 (80%)
(71) 359 (77%)
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
NEURONAL POPULATIONS IN SOMATOSENSORY CORTEX
Anterior -+
/
\ \
3b
lr
\
j
\ I
3a
4
III
I
1 mm FIGURE 2. (A) A cresyl-violet-stained sagittal section through the somatosensory cortex containing an electrode trajectory with two electrolytic lesions, and (B)the camera lucida reconstruction of that section. In this case, the pontamine sky blue deposit was not seen, but the current employed produced two electrolytic lesions. The known distance between the lesions was used to correct for shrinkage during tissue processing. The neurons were located and assigned to a layer by scaling the micrometer readings according to the distance between the lesions.
leaving a sample of 187 neurons. Ten of the penetrations were located in area 3b, comprising 5% of the sample, whereas 27% of the sample was located in area 1, leaving only 14% of the sample in areas 3a and 2 (Table 2). No sigmficant difference was found between the proportions of neurons with and without receptive fields among the different cortical areas (G,.dj = 4.757,df= 3,p > O.l), so the neurons from the different areas were pooled.
Laminar Distribution
The laminar distribution of the sample located in the histology is shown in Figure 3A. The sample in each layer came from 7 to 12 penetrations (mean = 9.0), suggesting that the sampling was quite uniform throughout the depth of the cortex. Indeed, 10 of the 17 penetrations covered the entire cortical depth; for 301
WARREN A N D DYKES
TABLE2. Cytoarchitectonic Locations of the Penetrations and of the Neurons Found in the Histology Cytoarchitectonic areas 3b Number of penetrations Neurons with RF
SRA Tap SSA Total with RF
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
Neurons without RF Glut-RF NO-RF Rf not tested with glutamate Total without RF Whole sample
1
10 (1) 31 3 1 35
9 2 0 11
29 31 15 75
12 20 8 40
110
51
3a 1
2 (3)
Total 17
1 0 0 1
42 7 1 50
4 6 1 11
3 3 5 11
49 61 29 137
14
12
187
Nore. Numbers in parentheses indicate the number of penetrations that entered more than one cytoarchitectonic area. RF, receptive field; see “Methods” for other abbreviations.
6 penetrations the sampling was located in the middle and lower layers, whereas 1 penetration sampled only the upper layers. One of the penetrations, located in the posterior bank of the posterior sigmoid gyrus, traversed the cortex obliquely, and 17 of the 19 neurons were located in the middle of layer VI (midVI), almost doubling the sample of neurons in this layer. The remaining two neurons from this penetration were located in lower layer VI (loVI). Because of this important bias, that penetration was removed from the laminar analysis, leaving a sample of 168 neurons with an average of 15.2 neurons sampled per layer. The smallest sample (10 neurons) was found in the upper part of layer I11 (upIII), and the biggest samples (18 neurons each) were located in layers IV and midVI (Fig. 3A). The laminar distribution of the different classes of neurons are shown in Figures 3B, 3C, and 3D. Neurons displaying a receptive field tended to be more numerous in the middle layers (Fig. 3B), and cells without a receptive field were more common in upper and lower layers (Fig. 3D). Classes of Neurons
The probabilities of finding neurons with a receptive field are shown in Figure 4A. Since the sample contained only 7 Tap neurons and 1 SSA neuron, these were pooled with the SRA neurons, yielding 47 neurons displaying a receptive field. Neurons displaying a receptive field were found more frequently in the middle layers and in the lower part of layer VI, whereas no neurons displaying a receptive field were found in the upper 302
part of layer 111. They were encountered infrequently in layers I1 and the middle of layer VI. Glut-RF neurons (Fig. 4B) were more often present in the upper part of layer I11 and the lower part of V, but less often in layers I, the middle part of 111, the upper part of VI, and the lower part of VI. In layer I and the upper part of VI, more than one-half of the neurons lacked evidence of peripheral input, whereas in the middle layers the absence of somatic input was less common (Fig. 4C). Figure 4D shows the probabilities of encountering neurons with a receptive field and Glut-RF neurons as a function of the layer in which they were located; the probabilities of finding neurons with evidence of peripheral input were much higher in the middle layers than in the upper and lower layers. Spontaneously Active Neurons
Of340 nerons tested for the presence of ongoing activity, 42% were found to be spontaneously active. Neurons having a receptive field were spontaneously active (74%; n = 7 of 96) much more often than those without a receptive field (29%; n = 70 of 244) (G test, Gad, = 53.837, df = 1, p < 0.001). The highest incidences of spontaneously active neurons were found in the middle layers; the lower part of layer I11 and layer IV displayed the highest probabilities, whereas in layers I and the upper part of VI neurons were often silent (Fig. 5B). For neurons displaying a receptive field, the lowest probability of being spontaneously active was found in the lower part of V and the upper part of VI, whereas almost all the
NEURONAL POPULATIONS IN SOMATOSENSORY CORTEX
A
B
I
-1
up111 mid111
10111
IV UPV IOV
upVI
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
midVI IOVI 7
0
2
4
6
8
10
12
14
16
18
20
0
2
4
6
8
10
12
14
16
18
20
16
18
20
NUMBER OF NEURONS
NUMBER OF NEURONS
D
C I
I
11
I1
up111
up111
mid111
mid111
10111
lo111
IV
IV
UPV
UPV
IOV
upVI
upVI
midVI
midVI
I0VI@ggJ 0
2
,
1,
,
,
,
,
,
,
,
4
6
8
10
12
14
16
18
20
NUMBER OF NEURONS
IOVI
.
.
0
2
4
6
8
10
12
14
NUMBER OF NEURONS
FIGURE 3. Laminar distribution of neurons located in the histology. The vertical lines represent the mean number of neurons across all layers. (A) All neurons located in the histology; (B) neurons displaying a receptive field; (C) Glut-RF neurons; and @) No-RF neurons.
neurons in the lower part of I11 and in IV were spontaneously active (Fig. 5C). The majority of neurons without a receptive field were silent. Recall that these No-RF cells were frequently found in the upper and lower layers (Fig. 3D). This fact is shown in Figure 5D: The mean probability was only 0.27 for spontaneous activity in the sample of No-RF neurons, but almost half of the No-RF cells were spontaneously active in the upper part of layer I11 and in the lower part of V. The upper part of layer VI and the lower part of layer VI contained only silent neurons, and only few spon-
taneously active neurons were found in layers I, lower 111, and IV (Fig. 5D). Discharge Frequencies of Spontaneously Active Neurons
The ongoing discharge rate was measured in 107 of 141 (76%) spontaneously active neurons. Frequencies ranged from 0.1 to 14 imp/sec. The probabilities of encountering cells having a given frequency are illustrated in Figure 6A. No significant difference was found between the 303
WARREN AND DYKES
B
A I I1
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
up111
1
0.0
I
0.2
0.4
0.6
PROBABILITIES
0.8
1.o
0.0
0.2
0.4
0.6
1.o
0.8
PROBABILITIES
(n = 47)
D
C
0 SAMPLE WITH RECEPTIVE FIELD 0 SAMPLE WITH RECEPTIVE FIELD UNCOVERED WITH GLUTAMATE (Glut-RF)
I I1
I
up111
I1
mid111
up111
10111
midIII
IV
10111
IOV
UPV
IV
upVI
IOV
midW
upVI
IOVI
0.0
1
midVI
0.2
0.4
0.6
PROBABILITIES
0.8
1.o
IOVI
I
0.0
1
0.2
0.6 PROBABILITIES 0.4
0.8
1.o
FIGURE4. Laminar distribution of the probabilities of finding a class of neurons in each layer. The vertical lines represent the average probability of finding a cell in all layers. (A) Neurons displaying a receptive field; (B) Glut-RF neurons; (C) NoRF neurons; and (D) neurons with evidence of somatic inputs (neurons displaying a receptive field and Glut-RF neurons).
304
NEURONAL POPULATIONS IN SOMATOSENSORY CORTEX
A
B
I I1 up111 mid111 IOIII IV UPV IOV upVI
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
midVI I
0
2
4
6
8
10
12
14 16
18 20
0.0
0.2
0.4
NUMBER OF NEURONS
0.6
0.8
1 .o
0.8
1.o
PROBABILITIES
C
D
I
I
I1
I1
midIII
mid111
Ion1
Ion1
IV
IV
UPV
UPV
up111
IOV
IOV
upV1
UOVI
midV1
I
i
I
midVI
o w l-
0.0
i
0.2
0.4
0.6
PROBABILITIES
,
,
0.8
1.o
low
0.0
0.2
0.4
0.6
PROBABILITIES
FIGURE 5. Laminar distributionof spontaneouslyactive neurons. (A) All spontaneously active neurons located in the histology. The vertical line represents the mean number of neurons in all layers. (B) Probabilities of finding spontaneouslyactive neurons in each layer. The vertical line represents the average probability of finding these across all layers. (C) Probabilities of finding spontaneously active neurons displaying a receptive field in each layer. The vertical line represents the average probability of finding these across all layers. (D) Probabilities of finding spontaneously active neurons lacking a receptive field in each layer. The vertical line represents the average probability of finding these across all layers.
305
WARREN AND DYKES
30 -
w
25 -
!a z
3 8
0 SAMPLE RECOVERED IN THE HISTOLOGY (n=33) SAMPLE NOT RECOVERED IN THE HISTOLOGY (n=74) SAMPLE (n=107)
20 1510-
5L
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
0
0.22 0.47 0.72 0.97 1.22 1.47 1.72 1.97
3.1
5.1 7.1
-
FREQUENCY (irnp/sec)
0 SAMPLE WITH A RECEPTIVE FIELD (n=47) 401 35
E4 SAMPLE LACKING RECEPTIVE FIELD (n=60)
h7.1
"
0.22 0.47 0.72 0.97 1.22 1.47 1.72 1.97
3.1
5.1
FREQUENCY (irnp/sec)
FIGURE 6. Distribution of the frequencies of spontaneous activity. The width of each class is 0.25 imp/sec for frequencies below 2.1 imp/sec and 2.0 imp/sec for values above 2.1 imp/sec. The number of neurons in each class was expressed as a percentage so that the distributions of the different samples could be compared readily. (A) Sample of neurons located in the histology, sample not located in the histology, and total sample. (There was no significant difference between the two subsets.) (B) Neurons displaying a receptive field and neurons lacking a receptive field.
frequency distribution for those found and those not found in the histology (G test, Gadj= 5.637, df = 4, p > 0.1). All three distributions were strongly skewed to the lowest frequencies; 62% of the total sample had frequencies lower than 1.O imp/sec, and only 20% had rates greater than 2.0 irnplsec. The geometric mean of the ongoing discharge for the entire sample was 0.77 imp/sec. Neurons displaying 306
a receptive field had significantly higher spontaneous rates than those lacking a receptive field (1.03 and 0.62 imp/sec, respectively; F = 5.046, df = 1 1 , p < 0.01). This difference appeared to be due to the fact that fewer neurons displaying a receptive field had very low spontaneous discharge frequencies and more had higher frequencies. The distributions were similar in the midrange of frequencies (Fig. 6B), and, overall, the shapes
NEURONAL POPULATIONS IN SOMATOSENSORY CORTEX
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
of the two distributions were not significantly different (G test, Gadj = 4.715, df = 3, p > 0.1). NO significant difference was found between SRA and Tap classes or between Glut-RF and No-RF classes, suggesting that the presence of a receptive field is a determining factor for the frequency of spontaneous activity but that the modality of the input is not. The rate of spontaneous activity was measured in 32 of 53 (60.4%) spontaneously active neurons recorded in the histology. Because of the small sample size, no measures were available for the upper part of layer V and the lower part of layer VI, and there was only one measure for layer I, the middle of layer 111, and the upper part of layer VI. For the remaining layers, averages were obtained from four to six cells, and the geometric means and 95% confidence limits are presented in Figure 7A. The geometric means for all the layers but two were within the confidence limits of the overall sample: The middle of layer I11 had an average above the upper limit, and the average of layer II was below the lower limit, suggesting that those two layers might have mean discharge frequencies different from those of the other layers of the sample. Sensitivity to Glutamate
A
I I I I
IOV
I
upVI
I I I
midVI IOVI
I
I I I
I
I
FREQUENCY (implsec)
B I
I1
The threshold current required to induce a discharge was determined for 301 neurons; the adequate currents ranged from 4 to 224 nA. The laminar locations of 132 of these neurons were plotted (Fig. 7B). The significant deviations from the mean value suggests that neurons in some layers are more readily activated by glutamate than those in other layers. Since no difference had been found previously between neurons with and without a receptive field, the data from the two groups were pooled. The neurons located in layer I, lower layer 111, and layer IV were the least sensitive to glutamate, and those located in layer 11, the middle of layer 111, and the upper part of layer V were the most sensitive. The mean values for the remaining layers were within the confidence limits of the sample. The distributions of effective currents were skewed to the left (Fig. 8A). In the total sample, 65% of the neurons were driven with less than 50 nA of glutamate, and only 6.6% needed 100 nA or more. The geometric means for the four classes of neurons ranged from 28.5 to 37.1 nA, but no significant differences were detected among them. To test the hypothesis that spontaneously active neurons had lower thresholds of activation, glutamate currents used to excite spontaneously active neurons were compared to those used to excite silent neurons. Major differences were found at both tails of the current distributions (Fig. 8B); 30% of spontaneously active neurons needed less than 20 nA and only 9.1% needed
up111 mid111 10111 IV
IOV
upVI midVI IOVI 1
0
10
20
30
40
50
CURRENT OF GLUTAMATE (nA)
FIGURE7.Laminar distribution of the geometric mean of the frequencies of spontaneous activity and of the threshold currents of glutamate. The vertical line in each histogram represents the geometric mean of the sample, and the dashed lines represent the 95% confidence limits. The numbers in the histogram represent the number of neurons found in each layer. (A) Frequencies of spontaneous activity as a function of laminar location for 32 neurons located in the histology; (B) threshold currents of glutamate for 132 neurons located in the histology.
307
L
WARREN AND DYKES
A 3c
0 SAMPLE RECOVERED IN THE HISTOLOGY (n-118) 0 SAMPLE NOT RECOVERED IN THE HISTOLOGY (n-183) W TOTAL SAMPLE (11-301)
25
2c a I ;f 15 -I W
B 8
10 5 0
Tw 4.5 14.5 24.5 34.5 44.5 54.5 64.5 74.5 84.5 94.5 104.5
-
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
CURRENT OF GLUTAMATE (nA)
B 30
0 SAMPLE WITH SPONTANEOUS ACTIVITY (n=87) Fa SAMPLE LACKING SPONTANEOUS ACTIVITY (n-175)
i n
25
22
;f
20 15
B 10 5
4.5
14.5 24.5 34.5 44.5 54.5 64.5 74.5 84.5 94.5 104.5
rn
CURRENT OF GLUTAMATE (nA)
C
'"1
1
0 SAMPLE WKH SPONTANEOUS ACTIVITY 0 SAMPLE LACKING SPONTANEOUS ACTIVITY
125
T
W
BI-
loo 75
T
$U " 3 0
25 0
SRA
TAP
GlutRF
NorRF
TOTAL
TYPE OF NEURON
FIGURE8. Distribution and the magnitude of the threshold currents of glutamate required to activate neurons. (A) Sample of neurons located in the histology, sample not located in the histology, and total sample. There was no significant difference (C test, Gadj= 0.348, df = 6 , p < 0.1). (B) Neurons displaying a receptive field and neurons lacking a receptive field. (C) Average threshold current of glutamate used on spontaneously active and silent SRA,Tap, Glut-RF, No-RF, and total sample. The currents of glutamate are in nanoamperes. The vertical bars represent the 95% confidence limits. 308
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
NEURONAL POPULATIONS IN SOMATOSENSORY CORTEX
more than 60 nA, whereas the picture was reversed for neurons having no ongoing activity (9.1% were excited with less than 20 nA and 29% with 60 nA or more). The difference between the two distributions was highly significant (Gtest, G,j = 27.673, df= 5, p < 0.001). The geometric mean of the current necessary to excite spontaneously active neurons did not vary much among the different classes of neurons (Fig. 8C), ranging from 23.4 to 29.8 nA with an overall geometric mean of 25.6 nA. On average, 13.8 nA more current was necessary to activate neurons that were not spontaneously active. The variations were larger among the different classes of silent neurons because of the small sample sizes for SRA (n = 6) and Tap (n = 5) neurons, but when these two classes were pooled, the mean was very close to those observed for the Glut-RF and NoRF classes. There was a highly significant difference between the spontaneouslyactive and the silent neurons, whereas no difference was found among the four functional classes of neurons. No significant correlation was found between the glutamate currents and the rates of spontaneous discharge (r = 0.231, p > 0.05, n = 72), suggesting that ongoing discharges were not mistaken for glutamate-induced activity. It is interesting to note that when spontaneous and silent neurons were pooled, neurons displaying a receptive field were slightly more sensitive to glutamate than those lacking a receptive field (-4.0 nA), whereas the situation was reversed for spontaneously active neurons and no differencewas observed for silent neurons. This suggested that the difference in the pooled data was attributable to the fact that neurons displaying a receptive field were more often spontaneouslyactive.
DISCUSSION The Sample A total of 465 single units were excited by glutamate, suggesting that they were neurons. Only 23% could be activated by somatic stimuli in the absence of glutamate. Ejections of small amounts of glutamate allowed another 34% to be activated by signals arising from stimuli applied by the experimenter, leaving 43% with no evidence of somatic input even when they were partially depolarized. Thus, under these experimental conditions three-quarters of the neurons in the somatosensory cortex lacked an overt receptive field, and nearly half displayed no inputs even during partial depolarization. These proportions are similar to the proportions reported in several other microiontophoretic studies in both cats (Dykes and Lamour, 1988a; Tremblay el al., 1988, 1990) and rats (Dykes and Lamour, 1988b; Lamour et al.,
1988), but in other studies where unresponsive cells were tabulated (Dykes et al., 1984; Metherate et al., 1988; Swadlow, 1989), more than half of the neurons studied displayed a receptive field. The reasons why the proportion of the sample displayinga receptive field differs so widely are not clear. Possible explanations are related to the kind of anesthetic used, the type of recording electrode used, and the presence of certain excitatory drugs in the iontophoretic pipette, rather than to fundamental differences in the cortex studied. For example, the presence of bicuculline in the microiontophoretic pipette has been shown to increase the probability of finding neurons displaying a receptive field (Lamour et al., 1988) and this might explain the higher proportions found in some studies (e.g., Dykes el al., 1984). The fact that Metherate et al. (1988) found almost twice as many neurons with a receptive field as we did in the present study can be attributed to the fact that they used carbon fiber electrodes (which may isolate the smaller cells found in the middle layers), rather than the glass pipettes used in the present study. Although anesthesia might be used as the major explanation for the small proportions of neurons displaying a receptive field in this and other reports on anesthetized animals, neurons without receptive fields are encountered frequently even in unanesthetized, paralyzed rabbits (Swadlow, 1989) and in cats (Dykes and Lamour, 1988a), and may account for as much as half of the neuronal population. The role of these neurons is not clear. Some may receive inputs from the corpus callosum (Landry and Dykes, 1985) and other nonsomatosensory sources; some may be efferent cells or may serve as a reservoir of neurons related to learning, memory, and adaptive functions. In this context, it is interesting to note that norepinephrine produced a longlasting enhancement of excitability in a much larger proportion of neurons lacking a receptive field than in neurons displaying a receptive field (Warren and Dykes, 1993b). Furthermore, the probability of observing a long-lasting enhancement was increased if neurons lacked both a receptive field and spontaneous activity. It is not immediately evident whether the differences between these results and the results of Smits et al. (1991) are attributable to a difference in selection of the sample studied or to a difference in the stimuli used. In a sample of 109 somatosensory cortical neurons, Smits et al. (1991) could influence each cell isolated when they stimulated a peripheral nerve electrically and 79% when they used natural stimulation of the skin surface. Their criterion for a response was the observation of excitatory postsynaptic potentials following the peripheral stimulus. Our extracellular recording position did not allow the detection of subthreshold events, and so our unresponsive neurons may have 309
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
WARREN AND DYKES
had unseen excitatory postsynaptic potentials elicited by somatic stimuli. Alternatively, the administration of glutamate in our study may have allowed us to detect unresponsive cells missed by Smits et al. (1991), who were unable to activate cells lacking afferent inputs with the techniques employed in their experiments. A significant and uncontrolled bias in this study arose because cells were isolated principally with the aid of iontophoretically administered glutamate. We believe that this method allows us to detect more silent cells than can be found with metal electrodes. Nevertheless, glutamate iontophoresis is not without problems; there is evidence in the spinal cord that some neurons do not respond to glutamate (Schneider and Perl, 1988) and some evidence that some neurons in cortex are much more sensitive than others to glutamate (Dykes et al., 1984). We were unable to find the same relationship between cortical layers and glutamate sensitivity reported by Dykes et al. (1984), but we did encounter large differences in glutamate sensitivity as a function of laminar position.
Spontaneous Activity and Sensitivity to Glutamate
Both the proportion of spontaneously active neurons and the rate of spontaneous activity found in the present study are in the range of those reported in other iontophoretic studies in cat somatosensory cortex from this laboratory, even though different anesthetics were employed (Metherate et al., 1988; Tremblay et al., 1988, 1990). In the present study, the probability of finding a spontaneously active neuron was increased to 74% if that neuron displayed a receptive field, whereas only 29% of neurons lacking a receptive field were spontaneously active. Furthermore, the rate of spontaneous activity was higher in neurons displaying a receptive field than in those lacking one. This suggests that inputs from the periphery might be an important factor in the generation of ongoing activity, and that neurons displaying a receptive field have a ratio of excitatory to inhibitory influences higher than that of neurons lacking one. This is supported by the fact that spontaneously active neurons displayed a lower threshold of activation by glutamate. In anesthetized cats, both classes of neurons appear to receive strong tonic inhibitory input from y-aminobutyric acid-ergic (GABA-ergic) neurons, which masks the expression of weaker peripheral excitatory inputs (Hicks and Dykes, 1983; Dykes et al., 1984). The GABA-ergic antagonist bicuculline produced larger receptive fields in cells with receptive fields and uncovered receptive fields in some cells previously lacking receptive fields. In rat somatosensory cortex,
310
bicuculline iontophoresis uncovered receptive fields more often than glutamate did (Lamour et al., 1988). These data suggest that at least some of the neurons lacking a receptive field receive an important input from the periphery, and that a large part of this excitatory input is shut down by GABA-ergic inhibition. Laminar Distribution of the Sample
In the present study, neurons were isolated rather uniformly from all cortical layers. In previous iontophoretic studies where glutamate has been used to excite otherwise quiescent neurons, the result has been a more uniform samplhg of neurons throughout the somatosensory cortex than was true for earlier studies done without glutamate (Mountcastle, 1957; Towe et al., 1964; Morse et al., 1965). Nevertheless, even those studies employing glutamate yielded more neurons in the middle layers than in the upper and lower layers in both cats (Dykes et al., 1984; Metherate et al., 1988; Tremblay et al., 1988) and rats (Dykes and Lamour, 1988b). In contrast, Tremblay et al. (1990), using the same anesthetic and same type of recording electrode as in the present study, obtained a distribution that more closely resembled the distribution reported here. Although differences in anesthetic might be responsible for reducing the number of cells found in the upper and lower layers, the electrode type might also be a determining factor. In the present experiments, as in the experiments of Tremblay et al. (1990), glass pipettes were used, whereas in previous iontophoretic studies carbon fiber electrodes were used and more neurons could be isolated in the middle layers; this suggests that carbon fiber electrodes more readily isolate the neurons found in the middle layers. Histological studies show that in the somatosensory cortex, the bottom of layer I11 and layer IV receive denser thalamic innervation than lower layers (Jones, 1975). Furthermore, the dendritic arbors of most pyramidal neurons of layer 111, V, and VI extend into the layers where the densest thalamic innervation is found, and thereby are in a position to receive numerous thalamic inputs (Hendry and Jones, 1983). In that context, it is interesting to note that a large proportion of NoRF neurons were found in layers I and VI, while GlutRF neurons were concentrated in layer 11, upper layer 111, and layer V; this suggests that the probability of finding demonstrable somatic input decreased as the neurons were located farther away from the middle layers. Higher proportions of spontaneously active neurons were found in layers containing high proportions of
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
NEURONAL POPULATIONS IN SOMATOSENSORY CORTEX
neurons displaying a receptive field. This distribution is similar to that found by Tremblay et al. (1990) in cats and by Dykes and Lamour (1988b) in rats. The most striking feature is the low proportion of spontaneously active neurons displaying a receptive field found in the lower part of layer I11 as compared to the middle of layer 111and layer IV. The reason for this is unknown, but neurons in layer I11 also displayed a higher rate of spontaneous activity. Neurons lacking a receptive field were more likely to be spontaneously active in upper and lower layers than in the middle layers. Swadlow (1989) reported that most efferent neurons (layers V and VI) had low spontaneous activity, and few had receptive fields. Although we have no information about the efferent targets of the cells studied, the low proportions of spontaneous activity neurons and of cells with receptive fields in these deeper layers where the efferent cells are located are consistent with that report. Layer IV and the bottom third of layer I11 contain the highest densities of GABA-immunoreactiveterminals (Hendry et al., 1987) suggesting that neurons located in these layers are more strongly inhibited than in other layers. Neurons located there were also less sensitive to glutamate than neurons found in the layers just above or below, perhaps because of this strong GABA-ergic inhibition. In conclusion, neurons with a receptive field display a higher probability of being spontaneously active than neurons lacking a receptive field. It is not clear whether the spontaneous activity is generated by the thalamocortical input or whether the ongoing activity is generated by another input. Spontaneously active neurons are more readily activated by glutamate than silent neurons, but sensitivity to glutamate is not dependent on the presence of a receptive field.
ACKNOWLEDGMENTS
We thank G. Filosi, G. Gauthier, and D. Cyr for the artwork; F. Cantin for the histological preparations; and C. Champagne for typing
the manuscript. This work was supported by the Medical Research Council of Canada. Richard A. Warren received a fellowship from the Medical Research Council of Canada.
REFERENCES
DYKES, R. W., and Y. LAMOUR (1988a) Neurons without demonstrable receptivefields outnumberneurons having receptive fields in samples from the somatosensory cortex of anesthetized or paralyzed cats and rats. Brain Res. 440: 133-143. DYKES, R. W., and Y. LAMOUR (1988b) An electrophysiologicalstudy of single somatosensory neurons in rat granular cortex serving the limbs: A laminar analysis. J. Neurophysiol. 60: 703-724.
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. 52: 1066-1093. HASSLER,R., and K. MUHS-CLEMENT (1964) Architektonischer aufbau des sensomotorischenund parietalen cortex der katze. J. Hirnforsch. 6: 377-420. HENDRY,S. H. C., and E. G. JONES (1983) The organization of pyramidal and non-pyramidal cell dendrites in relation to thalamic afferent terminations in the monkey somatic sensory cortex. J. Neurocytol. 12: 277-298. HENDRY, S. H. C., H. D. SCHWARK, E. G. JONES, and J. JAN (1987) Numbers and proportions of GABA-immunoreactive neurons in different areas of monkey cerebral cortex. J. Neurosci. 7 15031519. HICKS,T. P. (1984) The history and development of microiontophoresis in experimental neurobiology. Prog. Neurobiol22: 185-240. HICKS,T. P., and R. W. DYKES (1983) Receptive field size for certain neurons in primary somatosensory cortex is determined by GABAmediated intracortical inhibition. Brain Res. 274: 160-164. JONES, E. G. (1975) Lamination and differentialdistribution of thalamic afferents within the sensory-motor cortex of the squirrel monkey. J. Comp. Neurol. 160: 167-204. LAMOUR, Y., P. DUTAR,A. JOBERT, and R. W. DYKES(1988) An iontophoretic study of single somatosensory neurons in rat granular cortex serving the limbs: A laminar analysis of glutamate and acetylcholine effects on receptive field properties. J. Neurophysiol. 60:725-750. LANDRY, P., P. DIADORI, and R. W. DYKES(1984) Interhemispheric reciprocal interaction between ventropostemlateral thalamic nuclei involving cortical relay neurons. Brain Res. 323: 138-143. LANDRY, P., and R. W. DYKES (1985) Identification of two populations of corticothalamic neurons in cat primary somatosensory cortex. Exp. Brain Res. 60: 289-298. METHERATE, R., N. TREMBLAY, and R. W. DYKES (1988) The effects of acetylcholine on response properties of cat somatosensory cortical neurons. J. Neurophysiol. 59: 1231- 1252. MORSE,R. W., R. J. ADKINS,and A. L. TOWE(1%5) Population and modality characteristics of neurons in the coronal region of somatosensory area 1 of the cat. Exp. Neurol. 11: 419-440. MOUNTCASTLE, V. B. (1957) Modality and topographic properties of single neurons of cat’s somatic sensory cortex. J. Neurophysiol. 20: 408-434. RANDOLF,M., and J. SEMMES (1974) Behavioral consequences of selective subtotal ablations in the postcentral gyrus of Macaca mulatta. Brain Res. 70: 55-70. SCHNEIDER, S. P., and E. R. PERL (1988) Comparison of primary afferent and glutamate excitation of neurons in the mammalian spinal dorsal horn. J. Neurosci. 8: 2062-2073. SIEGEL,S. (1956) Nonparametric Statistics for the Behavioral Sciences, McGraw-Hill, New York. SMITS,E., D. C. GORDON, S. WIITE, D. D. RASMUSSON,and P. ZARZECKI (1991) Synaptic potentials evoked by convergent somatosensory and cortical inputs in raccoon somatosensory cortex: Substrates for plasticity. J. Neurophysiol. 66: 688-695. SOUL, R. R., and F. J. ROHLF(1981) Biometry, W. H. Freeman, San Francisco. SRETAVAN, D., and R. W. DYKES(1983) The organization of two cutaneous submodalities in the forearm region of area 3b of cat somatosensory cortex. J. Comp. Neurol. 213: 381-398. SWADMW, H. A. (1989) Efferent neurons and suspected interneurons in S-I vibrissa cortex of the awake rabbit: Receptive field and axonal properties. J. Neurophysiol. 62: 288-308.
311
WARREN AND DYKES TOWE,A. L., H.D. PATTON,and T. T. KENNEDY (1964) Response properties of neurons in the pericruciate cortex of the cat following electrical stimulation of the appendages. Exp. Neurol. 10: 325-
344.
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
TREMBLAY, N., R. A. 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 palanine. Neuroscience 26: 745-762. TREMBLAY, N., R. A. WARREN, and R. W. DYKES(1990) Electrophysiological studies of acetylcholine and the role of the basal forebrain in the somatosensory cortex of the cat: I. Cortical neurons excited by glutamate. J. Neurophysiol. 64: 1199- 1211.
312
WARREN, R. A,, and R. W.DYKES(1993a) Studies of norepinephrine with selective agonists and antagonists administered iontophoretically to somatosensory cortical neurons in halothane-anesthetized cats. J. Neurophysiol. In press. WARREN,R. A., and R. W. DYKES(1993b) Long-lasting effects of norepinephrine administered iontophoretically to somatosensory cortical neurons in halothane-anesthetized cats. J. Neurophysiol. In press. WELKER,C. (1971) Microelectrode delineation of fine grain somatotopic organization of SmI cerebral cortex in albino rat. Brain Res. 26: 259-276.
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
Abstract Single neurons in the somatosensorycortex are divisible into a population with receptive fields and a population without receptive fields. These two populations
display different laminar distributions,and their respective functions are unknown. We compared other physiological characteristics of these two neuronal populations in an attempt to understand why some neurons lack a receptive field. Only 23% of 465 neurons isolated in the somatosensory cortex of halothane-anesthetized cats displayed a cutaneous receptive field. The iontophoretic administration of glutamate uncovered input from the periphery in another 34% of the sample, leaving 43% of the neurons without evidence of peripheral input under these experimental conditions. Neurons with a receptive field were spontaneously active much more often than neurons lacking peripheral inputs, and their rates of discharge were higher. No differences were found between neurons having a receptive field uncovered with glutamate and those unaffected by glutamate. In all classes of neurons, those cells with spontaneous activity were excited by smaller amounts of glutamate than were silent neurons, but sensitivityto glutamate was not correlated with the presence or absence of a receptive field. We infer that some classes of somatosensory cortical neurons receive strong thalamocortical inputs, whereas others have only relatively weak or no thalamocortical connections. In other experiments we have shown also that those neurons lacking a receptive field and/ or spontaneous activity were more likely to be plastic than those with stronger inputs (see Warren and Dykes, 1993a,b), suggesting that neurons having weaker afferent inputs can be more readily modified under certain circumstances. Key words somatosensory cortex, cat, iontophoresis, cortical neurons,
receptive field
Although there has been a resurgence of interest in the details of somatotopy inprimary somatosensory cortex since the advent of the micromapping technique (Welker, 1971), there are few studies of the cellular composition of this region that is so important for normal tactile function and normal sensory gnosis (Randolf and Semmes, 1974). One intriguing aspect of this region is the fact that a large number of cells in primary somatosensory cortex lack receptive fields (Dykes and Lamour, 1988a). Often these cells are overlooked or 1. Present address: Department of Anatomy and Neurobiology, University of California at Irvine, Irvine, California 92717. 2. To whom all correspondence should be addressed, at Dkpartement de Physiologie, Faculte de Mtdecine, Universitt de Montreal, C.P. 6128, Succursale A, Montreal, Qutbec H3C 357, Canada.
remain undetected and are not studied because there are only a limited number of ways to activate them. One advantage offered by the multibarreled iontophoretic microelectrode is that one barrel routinely contains glutamate or another amino acid to excite and thus detect these cells. Although some of these cells may be part of a pathway from the contralateral cortex to the thalamus (Landry et al., 1984), those cells receiving callosal information can account for only a small fraction of the neurons lacking receptive fields. These cells are found in a number of different experimental conditions, including behaving animals, and it would be useful to know more about them. Routinely, somatosensory cortical neurons are also differentiated by their rates of spontaneous activity and by their receptive field
Somatosensory and Motor Research, Vol. 9, No. 4, 1992, pp. 297-312
Accepted June 12, 1992
297
WARREN AND DYKES
characteristics. The frequency with which each of these characteristics is encountered varies as a function of the laminar location of the cell being studied. As a baseline for other iontophoretic studies (Warren and Dykes 1993a,b), we compared the electrophysiological characteristics of a large sample of cortical neurons obtained with microiontophoretic pipettes containing glutamate to comparable samples described in the literature.
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
METHODS Experiments were performed on 46 adult mongrel cats of either sex. A detailed description of the animal preparation under halothane anesthesia and recording techniques has appeared elsewhere (Warren and Dykes, 1993a). Briefly, the animals were ventilated artificially with air containing 2% halothane during surgery and 1-1.5% during recording sessions. Mapping, Recording, and Classification
Following a craniotomy over the somatosensory cortex, brief mapping was performed; a carbon-fiber-in-glass microelectrode was used to locate areas responsive to stimulation of the skin of the forearm. That electrode was replaced by a combination microiontophoretic and recording electrode made with a seven-barrel glass pipette glued with light-cured dental adhesive to a single glass recording electrode. The tip of the recording electrode protruded 20-40 pm beyond the multibarrel electrode and was filled with 2% pontamine sky blue in NaC1. The multibarrel pipette contained DL-glutamate (0.15 M, pH 8.0) and NaCl (0.9%, pH 7.0) for current balancing, and different noradrenergic agonists and antagonists (Warren and Dykes, 1993a), but only the responses to glutamate are reported here. There are important limitations of the microiontophoretic technique, which have been discussed in depth by Hicks (1984) and others. The technique as practiced in this laboratory is described in Dykes et al. (1984), Metherate et al. (1988), and Warren and Dykes (1993a). Fortunately, glutamate is relatively easy to use because it is naturally present in the cortex and is removed rapidly by existing uptake mechanisms. The primary difficulty is that rapid administration of large amounts can result in depolarization block of the neurons near the ejection pipette. To avoid this problem, the experimenter must eject small, gradually increasing quantities intermittently as the electrode is advanced. Current or pH artifacts can be detected by ejection of NaC1. As the electrode was advanced, pulses of glutamate were delivered regularly to excite otherwise quiescent neurons. At the same time, the skin of the forearm was 298
stimulated with gentle somatic stimuli. Once a unit was isolated, its depth on the micrometer was noted, and the skin was searched to determine whether or not there was a receptive field. When present, the receptive field was classified as follows: 1. A receptive field was said to be located in the skin when light touch of the skin with a hand-held, firepolished glass probe and/or the flicking of a few hairs elicited a clear and reproducible response over the background activity. This criterion was used to define the boundary of the field. These cutaneous receptive fields were further characterized as being either slowly adapting (SSA) or rapidly adapting (SRA), according to whether they did or did not respond throughout the duration of a stimulus lasting several seconds. The shape and size of the receptive field were represented on a standard drawing of the cat forearm (Fig. 1). 2. A receptive field was classified as Deep when the neuron responded only to the stimulation of subcutaneous structures-that is, to the palpation of the underlying muscles or to joint movement. 3. In several cases receptive fields were classified as Tap when the unit required a light tapping of the skin to respond to every stimulus and the lower-velocity manipulations needed to differentiate between the cutaneous and deep categories were ineffective. For neurons without a receptive field, the search for a receptive field was repeated during the administration of subthreshold doses of glutamate. These neurons were classified as having no receptive field (No-RF) or as having a receptive field only in the presence of glutamate (Glut-RF).
FIGURE1. Examples of cutaneous receptive fields for single neurons isolated in this study. Receptive fields A , B, C, and D were cutaneous and rapidly adapting. The neurons were located in layer V (A), layer 111 (B), and layer IV (C). The neuron with receptive field D was not recovered in the histology, but according to the micrometer reading it was located in the middle of the cortex. Receptive field E was classified as Tap, and the neuron was not located.
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
NEURONAL POPULATIONS IN SOMATOSENSORY CORTEX
Neurons were also characterized by the presence or absence of spontaneous activity. Since spontaneous activity was usually very low, the neuron was left unstimulated for at least 1 min before a decision was made. In many cases when an ongoing discharge was present, the frequency of this activity was evaluated either by counting the action potentials with the window discriminator for at least 100 sec or by using data stored on the computer during the control period preceding a series of somatic stimuli or pulses of glutamate. All neurons displaying an ongoing discharge of less than 0.1 impulses/sec (imp/sec) were considered to be silent. For many neurons, the threshold quantity of glutamate (as measured in nanoamperes) necessary to produe a response was determined by first setting a current pulse (30 sec) that elicited a clear response, and then decreasing this current in small steps until no clear response was observed. Then the procedure was reversed: The current was increased in small steps until a just-noticeable response was detected over the background activity. The value of this current was recorded as the threshold. In some cases glutamate failed to elicit a response even with currents up to 500 nA, suggesting that these units lacked glutamate receptors and/or that they were axons. Histology When a significant number of neurons were isolated in one penetration, pontamine sky blue was ejected from the recording electrode in two locations along the electrode track by passing a current of negative polarity. Prior to perfusion of the animal, the halothane concentration in the inspired air was raised to 4% for 10 to 15 min. The animal was perfused through the ascending aorta with 0.9% saline followed by 10% buffered formalin. The brain was removed from the skull and placed in the same fixative for several days before being cryoprotected with 30% sucrose. Sections 80 pm thick were cut through the somatosensory cortex on a cryostat and mounted on gelatin-coated slides. The sections were stained for Nissl substance with cresyl violet according to standard procedures and coverslipped. The micrometer readings provided the depth of each cell as well as the depth of the pontamine sky blue ejections, allowing reconstruction of the electrode trajectory, with the knowledge of the distance between two dye spots giving the information to compensate for tissue shrinkage that occurred during the processing of the tissue. Camera lucida drawings of the sections containing the dye spots were made at X40, and each neuron was attributed to a cortical layer and cytoarchitectonic area according to the cytoarchitectonic cri-
teria of Hassler and Muhs-Clement (1964). In some cases the two dye spots could not be seen, but electrolytic lesions were present. Data Analysis Appropriate statistical tests were used whenever possible to compare the laminar distribution, the effect of glutamate, and the spontaneous activity of the different classes of neurons. The choice of the proper statistical test to be used was based on the theoretical arguments of Sokal and Rohlf (1981). To test for whether or not proportions in a sample differed from chance (i.e., a contingency table), the G statistic was used whenever the size of the sample was large enough. In some cases, classes displaying similar characteristics were pooled (e.g., laminar distribution, distribution of the frequencies of spontaneous activity and of threshold current of glutamate) to meet the prerequisite sample size of the test. In the case of 2 x 2 tables of small sample size, the Fisher’s exact test was applied according to the tables of Siege1 (1956). When the analysis of variance was used, the assumptions of that test were fulfilled. Statistical tests were complicated by the fact that the frequencies of spontaneous activity and the iontophoretic currents used were not normally distributed. Furthermore, the variances of the samples were seldom homogeneous. After a logarithmic transformation, both spontaneous activity and glutamate currents appeared to be normally distributed and to have homogeneous variances, so the analysis of variance was performed on the transformed data. The geometric means are reported instead of the usual arithmetic means, since they give a better representation of the central values of the data. The data for the activity evoked by cutaneous stimuli and the responses to glutamate were normally distributed, and their variances were homogeneous. In those cases the data did not need to be transformed; statistical tests were performed on the original data, and the arithmetic means were used. RESULTS Origin of the Sample A total of 533 single units were isolated in the somatosensory cortex. Of these, 68 were excluded from further analysis because they were not held long enough (n = 55) or they were found to be insensitive to glutamate ( n = 13), suggesting that they were fibers (although some could have been glutamate-insensitive neurons; Schneider and Perl, 1988). In the other 465 cases, the units were sensitive to glutamate and were held long 299
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
WARREN AND DYKES
enough to be tested for the presence of a receptive field (Table 1). These putative neurons were isolated along 66 electrode penetrations. In 17 Of 46 animals, the electrode tracks were reconstructed from the histological sections, and 189 neurons were located with respect to the dye spots or the lesions (Fig. 2). Only 23% of the neurons displayed a receptive field; 85 appeared to receive input from the skin, and of these, 82 were classified as SRA units, whereas only three had the characteristics of SSA units. Only 1 neuron appeared to receive input from muscle receptors, responding to the flexion of the fifth digit; it was classified as serving subcutaneous receptors (Deep). In 20 cases the modality appeared uncertain, because these neurons could have been classified as rapidly adapting neurons responding to high-threshold cutaneous stimuli or as low-threshold rapidly adapting receptors in subcutaneous tissues. Since they responded to light tapping of their receptive fields, they were classified as Tap. The receptive fields were elongated following the long axis of the limb (Fig. 1). They were also smaller on the paw than on the forearm or the arm. Overall, they were similar in shape and size to those described in previous studies (Sretavan and Dykes, 1983; Metherate et al., 1988; Temblay et al. 1990). In the remaining 77% of the sample, no evidence of peripheral input was found in the absence of drug treatments, but in 127 cases, subthreshold amounts of glutamate (ranging from 4 to 90 nA; geometric mean = 23.5 nA) were sufficient to uncover receptive fields with characteristics similar to those found in other cells in the absence of drugs. These neurons were classified as having a receptive field during glutamate iontophoresis (Glut-RF neurons). In the other 161 cases, glutamate treatment faild to uncover peripheral inputs, and these
neurons were classified as lacking any receptive field (No-RF neurons). In 71 cases, constituting 15% of the sample, no attempt was made to uncover a receptive field with glutamate. These were not considered a legitimate class, since these cases were likely to be both Glut-RF and No-RF neurons. Nevertheless, these neurons had to be accounted for in he computation of the proportions of the Glut-RF and No-RF classes by assuming that this group contained the same proportions of Glut-RF and No-RF neurons found in the sample tested with glutamate for receptive fields. Accordingly, they were distributed to the Glut-RF and No-RF groups; this yielded a proportion of 43% of the neurons without any evidence of peripheral input and 34% with receptive fields uncovered by glutamate. Consequently, a grand total of 57% of the sample appeared to receive some input from the periphery under these experimental conditions. Table 1 also compares the proportions of neurons of each class found in those penetrations located in the histology with the proportions found in those not located in the histology. There was no significant difference between these two subsets (G test, Gad, = 4.677, (df = 3, p > 0.1 ; SSA and Deep classes were excluded from the test), suggesting that the two samples belonged to the same population and that no bias was introduced by locating some neurons in the histology. As a result, the total sample represents the best estimate for the proportions of the different classes of neurons observed. Cytoarchitectonic Location of the Sample Between 5 and 19 neurons were isolated in each of the 17 reconstructed penetrations. Only 2 neurons appeared to be located in the white matter. These were discarded,
TABLE1 . Neurons Isolated in Cat Somatosensory Cortex and Tested for Sensitivity to Glutamate (n = 465) ~
Found in histology With a receptive field SRA
SSA Tap Deep Total Without a receptive field Glut-RF
No-RF Receptive field not tested with glutamate Total Note. See “Methods” for abbreviations.
300
42 (22%) 1 (0.5%) 7 (3.7%) 0 (0%)
50 (26%)
Not found in histology
Total
40 (14%) 2 (0.7%) 13 (4.7%) 1 (0.4%) 56 (20%)
82 (18%) 3 (0.6%) 20 (4.2%) 1 (0.2%) 106 (23%)
78 (35%)
49 (33%) 61 (41%)
loo (45%)
127 (34%) 161 (43%)
(29) 139 (74%)
(421 220 (80%)
(71) 359 (77%)
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
NEURONAL POPULATIONS IN SOMATOSENSORY CORTEX
Anterior -+
/
\ \
3b
lr
\
j
\ I
3a
4
III
I
1 mm FIGURE 2. (A) A cresyl-violet-stained sagittal section through the somatosensory cortex containing an electrode trajectory with two electrolytic lesions, and (B)the camera lucida reconstruction of that section. In this case, the pontamine sky blue deposit was not seen, but the current employed produced two electrolytic lesions. The known distance between the lesions was used to correct for shrinkage during tissue processing. The neurons were located and assigned to a layer by scaling the micrometer readings according to the distance between the lesions.
leaving a sample of 187 neurons. Ten of the penetrations were located in area 3b, comprising 5% of the sample, whereas 27% of the sample was located in area 1, leaving only 14% of the sample in areas 3a and 2 (Table 2). No sigmficant difference was found between the proportions of neurons with and without receptive fields among the different cortical areas (G,.dj = 4.757,df= 3,p > O.l), so the neurons from the different areas were pooled.
Laminar Distribution
The laminar distribution of the sample located in the histology is shown in Figure 3A. The sample in each layer came from 7 to 12 penetrations (mean = 9.0), suggesting that the sampling was quite uniform throughout the depth of the cortex. Indeed, 10 of the 17 penetrations covered the entire cortical depth; for 301
WARREN A N D DYKES
TABLE2. Cytoarchitectonic Locations of the Penetrations and of the Neurons Found in the Histology Cytoarchitectonic areas 3b Number of penetrations Neurons with RF
SRA Tap SSA Total with RF
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
Neurons without RF Glut-RF NO-RF Rf not tested with glutamate Total without RF Whole sample
1
10 (1) 31 3 1 35
9 2 0 11
29 31 15 75
12 20 8 40
110
51
3a 1
2 (3)
Total 17
1 0 0 1
42 7 1 50
4 6 1 11
3 3 5 11
49 61 29 137
14
12
187
Nore. Numbers in parentheses indicate the number of penetrations that entered more than one cytoarchitectonic area. RF, receptive field; see “Methods” for other abbreviations.
6 penetrations the sampling was located in the middle and lower layers, whereas 1 penetration sampled only the upper layers. One of the penetrations, located in the posterior bank of the posterior sigmoid gyrus, traversed the cortex obliquely, and 17 of the 19 neurons were located in the middle of layer VI (midVI), almost doubling the sample of neurons in this layer. The remaining two neurons from this penetration were located in lower layer VI (loVI). Because of this important bias, that penetration was removed from the laminar analysis, leaving a sample of 168 neurons with an average of 15.2 neurons sampled per layer. The smallest sample (10 neurons) was found in the upper part of layer I11 (upIII), and the biggest samples (18 neurons each) were located in layers IV and midVI (Fig. 3A). The laminar distribution of the different classes of neurons are shown in Figures 3B, 3C, and 3D. Neurons displaying a receptive field tended to be more numerous in the middle layers (Fig. 3B), and cells without a receptive field were more common in upper and lower layers (Fig. 3D). Classes of Neurons
The probabilities of finding neurons with a receptive field are shown in Figure 4A. Since the sample contained only 7 Tap neurons and 1 SSA neuron, these were pooled with the SRA neurons, yielding 47 neurons displaying a receptive field. Neurons displaying a receptive field were found more frequently in the middle layers and in the lower part of layer VI, whereas no neurons displaying a receptive field were found in the upper 302
part of layer 111. They were encountered infrequently in layers I1 and the middle of layer VI. Glut-RF neurons (Fig. 4B) were more often present in the upper part of layer I11 and the lower part of V, but less often in layers I, the middle part of 111, the upper part of VI, and the lower part of VI. In layer I and the upper part of VI, more than one-half of the neurons lacked evidence of peripheral input, whereas in the middle layers the absence of somatic input was less common (Fig. 4C). Figure 4D shows the probabilities of encountering neurons with a receptive field and Glut-RF neurons as a function of the layer in which they were located; the probabilities of finding neurons with evidence of peripheral input were much higher in the middle layers than in the upper and lower layers. Spontaneously Active Neurons
Of340 nerons tested for the presence of ongoing activity, 42% were found to be spontaneously active. Neurons having a receptive field were spontaneously active (74%; n = 7 of 96) much more often than those without a receptive field (29%; n = 70 of 244) (G test, Gad, = 53.837, df = 1, p < 0.001). The highest incidences of spontaneously active neurons were found in the middle layers; the lower part of layer I11 and layer IV displayed the highest probabilities, whereas in layers I and the upper part of VI neurons were often silent (Fig. 5B). For neurons displaying a receptive field, the lowest probability of being spontaneously active was found in the lower part of V and the upper part of VI, whereas almost all the
NEURONAL POPULATIONS IN SOMATOSENSORY CORTEX
A
B
I
-1
up111 mid111
10111
IV UPV IOV
upVI
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
midVI IOVI 7
0
2
4
6
8
10
12
14
16
18
20
0
2
4
6
8
10
12
14
16
18
20
16
18
20
NUMBER OF NEURONS
NUMBER OF NEURONS
D
C I
I
11
I1
up111
up111
mid111
mid111
10111
lo111
IV
IV
UPV
UPV
IOV
upVI
upVI
midVI
midVI
I0VI@ggJ 0
2
,
1,
,
,
,
,
,
,
,
4
6
8
10
12
14
16
18
20
NUMBER OF NEURONS
IOVI
.
.
0
2
4
6
8
10
12
14
NUMBER OF NEURONS
FIGURE 3. Laminar distribution of neurons located in the histology. The vertical lines represent the mean number of neurons across all layers. (A) All neurons located in the histology; (B) neurons displaying a receptive field; (C) Glut-RF neurons; and @) No-RF neurons.
neurons in the lower part of I11 and in IV were spontaneously active (Fig. 5C). The majority of neurons without a receptive field were silent. Recall that these No-RF cells were frequently found in the upper and lower layers (Fig. 3D). This fact is shown in Figure 5D: The mean probability was only 0.27 for spontaneous activity in the sample of No-RF neurons, but almost half of the No-RF cells were spontaneously active in the upper part of layer I11 and in the lower part of V. The upper part of layer VI and the lower part of layer VI contained only silent neurons, and only few spon-
taneously active neurons were found in layers I, lower 111, and IV (Fig. 5D). Discharge Frequencies of Spontaneously Active Neurons
The ongoing discharge rate was measured in 107 of 141 (76%) spontaneously active neurons. Frequencies ranged from 0.1 to 14 imp/sec. The probabilities of encountering cells having a given frequency are illustrated in Figure 6A. No significant difference was found between the 303
WARREN AND DYKES
B
A I I1
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
up111
1
0.0
I
0.2
0.4
0.6
PROBABILITIES
0.8
1.o
0.0
0.2
0.4
0.6
1.o
0.8
PROBABILITIES
(n = 47)
D
C
0 SAMPLE WITH RECEPTIVE FIELD 0 SAMPLE WITH RECEPTIVE FIELD UNCOVERED WITH GLUTAMATE (Glut-RF)
I I1
I
up111
I1
mid111
up111
10111
midIII
IV
10111
IOV
UPV
IV
upVI
IOV
midW
upVI
IOVI
0.0
1
midVI
0.2
0.4
0.6
PROBABILITIES
0.8
1.o
IOVI
I
0.0
1
0.2
0.6 PROBABILITIES 0.4
0.8
1.o
FIGURE4. Laminar distribution of the probabilities of finding a class of neurons in each layer. The vertical lines represent the average probability of finding a cell in all layers. (A) Neurons displaying a receptive field; (B) Glut-RF neurons; (C) NoRF neurons; and (D) neurons with evidence of somatic inputs (neurons displaying a receptive field and Glut-RF neurons).
304
NEURONAL POPULATIONS IN SOMATOSENSORY CORTEX
A
B
I I1 up111 mid111 IOIII IV UPV IOV upVI
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
midVI I
0
2
4
6
8
10
12
14 16
18 20
0.0
0.2
0.4
NUMBER OF NEURONS
0.6
0.8
1 .o
0.8
1.o
PROBABILITIES
C
D
I
I
I1
I1
midIII
mid111
Ion1
Ion1
IV
IV
UPV
UPV
up111
IOV
IOV
upV1
UOVI
midV1
I
i
I
midVI
o w l-
0.0
i
0.2
0.4
0.6
PROBABILITIES
,
,
0.8
1.o
low
0.0
0.2
0.4
0.6
PROBABILITIES
FIGURE 5. Laminar distributionof spontaneouslyactive neurons. (A) All spontaneously active neurons located in the histology. The vertical line represents the mean number of neurons in all layers. (B) Probabilities of finding spontaneouslyactive neurons in each layer. The vertical line represents the average probability of finding these across all layers. (C) Probabilities of finding spontaneously active neurons displaying a receptive field in each layer. The vertical line represents the average probability of finding these across all layers. (D) Probabilities of finding spontaneously active neurons lacking a receptive field in each layer. The vertical line represents the average probability of finding these across all layers.
305
WARREN AND DYKES
30 -
w
25 -
!a z
3 8
0 SAMPLE RECOVERED IN THE HISTOLOGY (n=33) SAMPLE NOT RECOVERED IN THE HISTOLOGY (n=74) SAMPLE (n=107)
20 1510-
5L
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
0
0.22 0.47 0.72 0.97 1.22 1.47 1.72 1.97
3.1
5.1 7.1
-
FREQUENCY (irnp/sec)
0 SAMPLE WITH A RECEPTIVE FIELD (n=47) 401 35
E4 SAMPLE LACKING RECEPTIVE FIELD (n=60)
h7.1
"
0.22 0.47 0.72 0.97 1.22 1.47 1.72 1.97
3.1
5.1
FREQUENCY (irnp/sec)
FIGURE 6. Distribution of the frequencies of spontaneous activity. The width of each class is 0.25 imp/sec for frequencies below 2.1 imp/sec and 2.0 imp/sec for values above 2.1 imp/sec. The number of neurons in each class was expressed as a percentage so that the distributions of the different samples could be compared readily. (A) Sample of neurons located in the histology, sample not located in the histology, and total sample. (There was no significant difference between the two subsets.) (B) Neurons displaying a receptive field and neurons lacking a receptive field.
frequency distribution for those found and those not found in the histology (G test, Gadj= 5.637, df = 4, p > 0.1). All three distributions were strongly skewed to the lowest frequencies; 62% of the total sample had frequencies lower than 1.O imp/sec, and only 20% had rates greater than 2.0 irnplsec. The geometric mean of the ongoing discharge for the entire sample was 0.77 imp/sec. Neurons displaying 306
a receptive field had significantly higher spontaneous rates than those lacking a receptive field (1.03 and 0.62 imp/sec, respectively; F = 5.046, df = 1 1 , p < 0.01). This difference appeared to be due to the fact that fewer neurons displaying a receptive field had very low spontaneous discharge frequencies and more had higher frequencies. The distributions were similar in the midrange of frequencies (Fig. 6B), and, overall, the shapes
NEURONAL POPULATIONS IN SOMATOSENSORY CORTEX
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
of the two distributions were not significantly different (G test, Gadj = 4.715, df = 3, p > 0.1). NO significant difference was found between SRA and Tap classes or between Glut-RF and No-RF classes, suggesting that the presence of a receptive field is a determining factor for the frequency of spontaneous activity but that the modality of the input is not. The rate of spontaneous activity was measured in 32 of 53 (60.4%) spontaneously active neurons recorded in the histology. Because of the small sample size, no measures were available for the upper part of layer V and the lower part of layer VI, and there was only one measure for layer I, the middle of layer 111, and the upper part of layer VI. For the remaining layers, averages were obtained from four to six cells, and the geometric means and 95% confidence limits are presented in Figure 7A. The geometric means for all the layers but two were within the confidence limits of the overall sample: The middle of layer I11 had an average above the upper limit, and the average of layer II was below the lower limit, suggesting that those two layers might have mean discharge frequencies different from those of the other layers of the sample. Sensitivity to Glutamate
A
I I I I
IOV
I
upVI
I I I
midVI IOVI
I
I I I
I
I
FREQUENCY (implsec)
B I
I1
The threshold current required to induce a discharge was determined for 301 neurons; the adequate currents ranged from 4 to 224 nA. The laminar locations of 132 of these neurons were plotted (Fig. 7B). The significant deviations from the mean value suggests that neurons in some layers are more readily activated by glutamate than those in other layers. Since no difference had been found previously between neurons with and without a receptive field, the data from the two groups were pooled. The neurons located in layer I, lower layer 111, and layer IV were the least sensitive to glutamate, and those located in layer 11, the middle of layer 111, and the upper part of layer V were the most sensitive. The mean values for the remaining layers were within the confidence limits of the sample. The distributions of effective currents were skewed to the left (Fig. 8A). In the total sample, 65% of the neurons were driven with less than 50 nA of glutamate, and only 6.6% needed 100 nA or more. The geometric means for the four classes of neurons ranged from 28.5 to 37.1 nA, but no significant differences were detected among them. To test the hypothesis that spontaneously active neurons had lower thresholds of activation, glutamate currents used to excite spontaneously active neurons were compared to those used to excite silent neurons. Major differences were found at both tails of the current distributions (Fig. 8B); 30% of spontaneously active neurons needed less than 20 nA and only 9.1% needed
up111 mid111 10111 IV
IOV
upVI midVI IOVI 1
0
10
20
30
40
50
CURRENT OF GLUTAMATE (nA)
FIGURE7.Laminar distribution of the geometric mean of the frequencies of spontaneous activity and of the threshold currents of glutamate. The vertical line in each histogram represents the geometric mean of the sample, and the dashed lines represent the 95% confidence limits. The numbers in the histogram represent the number of neurons found in each layer. (A) Frequencies of spontaneous activity as a function of laminar location for 32 neurons located in the histology; (B) threshold currents of glutamate for 132 neurons located in the histology.
307
L
WARREN AND DYKES
A 3c
0 SAMPLE RECOVERED IN THE HISTOLOGY (n-118) 0 SAMPLE NOT RECOVERED IN THE HISTOLOGY (n-183) W TOTAL SAMPLE (11-301)
25
2c a I ;f 15 -I W
B 8
10 5 0
Tw 4.5 14.5 24.5 34.5 44.5 54.5 64.5 74.5 84.5 94.5 104.5
-
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
CURRENT OF GLUTAMATE (nA)
B 30
0 SAMPLE WITH SPONTANEOUS ACTIVITY (n=87) Fa SAMPLE LACKING SPONTANEOUS ACTIVITY (n-175)
i n
25
22
;f
20 15
B 10 5
4.5
14.5 24.5 34.5 44.5 54.5 64.5 74.5 84.5 94.5 104.5
rn
CURRENT OF GLUTAMATE (nA)
C
'"1
1
0 SAMPLE WKH SPONTANEOUS ACTIVITY 0 SAMPLE LACKING SPONTANEOUS ACTIVITY
125
T
W
BI-
loo 75
T
$U " 3 0
25 0
SRA
TAP
GlutRF
NorRF
TOTAL
TYPE OF NEURON
FIGURE8. Distribution and the magnitude of the threshold currents of glutamate required to activate neurons. (A) Sample of neurons located in the histology, sample not located in the histology, and total sample. There was no significant difference (C test, Gadj= 0.348, df = 6 , p < 0.1). (B) Neurons displaying a receptive field and neurons lacking a receptive field. (C) Average threshold current of glutamate used on spontaneously active and silent SRA,Tap, Glut-RF, No-RF, and total sample. The currents of glutamate are in nanoamperes. The vertical bars represent the 95% confidence limits. 308
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
NEURONAL POPULATIONS IN SOMATOSENSORY CORTEX
more than 60 nA, whereas the picture was reversed for neurons having no ongoing activity (9.1% were excited with less than 20 nA and 29% with 60 nA or more). The difference between the two distributions was highly significant (Gtest, G,j = 27.673, df= 5, p < 0.001). The geometric mean of the current necessary to excite spontaneously active neurons did not vary much among the different classes of neurons (Fig. 8C), ranging from 23.4 to 29.8 nA with an overall geometric mean of 25.6 nA. On average, 13.8 nA more current was necessary to activate neurons that were not spontaneously active. The variations were larger among the different classes of silent neurons because of the small sample sizes for SRA (n = 6) and Tap (n = 5) neurons, but when these two classes were pooled, the mean was very close to those observed for the Glut-RF and NoRF classes. There was a highly significant difference between the spontaneouslyactive and the silent neurons, whereas no difference was found among the four functional classes of neurons. No significant correlation was found between the glutamate currents and the rates of spontaneous discharge (r = 0.231, p > 0.05, n = 72), suggesting that ongoing discharges were not mistaken for glutamate-induced activity. It is interesting to note that when spontaneous and silent neurons were pooled, neurons displaying a receptive field were slightly more sensitive to glutamate than those lacking a receptive field (-4.0 nA), whereas the situation was reversed for spontaneously active neurons and no differencewas observed for silent neurons. This suggested that the difference in the pooled data was attributable to the fact that neurons displaying a receptive field were more often spontaneouslyactive.
DISCUSSION The Sample A total of 465 single units were excited by glutamate, suggesting that they were neurons. Only 23% could be activated by somatic stimuli in the absence of glutamate. Ejections of small amounts of glutamate allowed another 34% to be activated by signals arising from stimuli applied by the experimenter, leaving 43% with no evidence of somatic input even when they were partially depolarized. Thus, under these experimental conditions three-quarters of the neurons in the somatosensory cortex lacked an overt receptive field, and nearly half displayed no inputs even during partial depolarization. These proportions are similar to the proportions reported in several other microiontophoretic studies in both cats (Dykes and Lamour, 1988a; Tremblay el al., 1988, 1990) and rats (Dykes and Lamour, 1988b; Lamour et al.,
1988), but in other studies where unresponsive cells were tabulated (Dykes et al., 1984; Metherate et al., 1988; Swadlow, 1989), more than half of the neurons studied displayed a receptive field. The reasons why the proportion of the sample displayinga receptive field differs so widely are not clear. Possible explanations are related to the kind of anesthetic used, the type of recording electrode used, and the presence of certain excitatory drugs in the iontophoretic pipette, rather than to fundamental differences in the cortex studied. For example, the presence of bicuculline in the microiontophoretic pipette has been shown to increase the probability of finding neurons displaying a receptive field (Lamour et al., 1988) and this might explain the higher proportions found in some studies (e.g., Dykes el al., 1984). The fact that Metherate et al. (1988) found almost twice as many neurons with a receptive field as we did in the present study can be attributed to the fact that they used carbon fiber electrodes (which may isolate the smaller cells found in the middle layers), rather than the glass pipettes used in the present study. Although anesthesia might be used as the major explanation for the small proportions of neurons displaying a receptive field in this and other reports on anesthetized animals, neurons without receptive fields are encountered frequently even in unanesthetized, paralyzed rabbits (Swadlow, 1989) and in cats (Dykes and Lamour, 1988a), and may account for as much as half of the neuronal population. The role of these neurons is not clear. Some may receive inputs from the corpus callosum (Landry and Dykes, 1985) and other nonsomatosensory sources; some may be efferent cells or may serve as a reservoir of neurons related to learning, memory, and adaptive functions. In this context, it is interesting to note that norepinephrine produced a longlasting enhancement of excitability in a much larger proportion of neurons lacking a receptive field than in neurons displaying a receptive field (Warren and Dykes, 1993b). Furthermore, the probability of observing a long-lasting enhancement was increased if neurons lacked both a receptive field and spontaneous activity. It is not immediately evident whether the differences between these results and the results of Smits et al. (1991) are attributable to a difference in selection of the sample studied or to a difference in the stimuli used. In a sample of 109 somatosensory cortical neurons, Smits et al. (1991) could influence each cell isolated when they stimulated a peripheral nerve electrically and 79% when they used natural stimulation of the skin surface. Their criterion for a response was the observation of excitatory postsynaptic potentials following the peripheral stimulus. Our extracellular recording position did not allow the detection of subthreshold events, and so our unresponsive neurons may have 309
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
WARREN AND DYKES
had unseen excitatory postsynaptic potentials elicited by somatic stimuli. Alternatively, the administration of glutamate in our study may have allowed us to detect unresponsive cells missed by Smits et al. (1991), who were unable to activate cells lacking afferent inputs with the techniques employed in their experiments. A significant and uncontrolled bias in this study arose because cells were isolated principally with the aid of iontophoretically administered glutamate. We believe that this method allows us to detect more silent cells than can be found with metal electrodes. Nevertheless, glutamate iontophoresis is not without problems; there is evidence in the spinal cord that some neurons do not respond to glutamate (Schneider and Perl, 1988) and some evidence that some neurons in cortex are much more sensitive than others to glutamate (Dykes et al., 1984). We were unable to find the same relationship between cortical layers and glutamate sensitivity reported by Dykes et al. (1984), but we did encounter large differences in glutamate sensitivity as a function of laminar position.
Spontaneous Activity and Sensitivity to Glutamate
Both the proportion of spontaneously active neurons and the rate of spontaneous activity found in the present study are in the range of those reported in other iontophoretic studies in cat somatosensory cortex from this laboratory, even though different anesthetics were employed (Metherate et al., 1988; Tremblay et al., 1988, 1990). In the present study, the probability of finding a spontaneously active neuron was increased to 74% if that neuron displayed a receptive field, whereas only 29% of neurons lacking a receptive field were spontaneously active. Furthermore, the rate of spontaneous activity was higher in neurons displaying a receptive field than in those lacking one. This suggests that inputs from the periphery might be an important factor in the generation of ongoing activity, and that neurons displaying a receptive field have a ratio of excitatory to inhibitory influences higher than that of neurons lacking one. This is supported by the fact that spontaneously active neurons displayed a lower threshold of activation by glutamate. In anesthetized cats, both classes of neurons appear to receive strong tonic inhibitory input from y-aminobutyric acid-ergic (GABA-ergic) neurons, which masks the expression of weaker peripheral excitatory inputs (Hicks and Dykes, 1983; Dykes et al., 1984). The GABA-ergic antagonist bicuculline produced larger receptive fields in cells with receptive fields and uncovered receptive fields in some cells previously lacking receptive fields. In rat somatosensory cortex,
310
bicuculline iontophoresis uncovered receptive fields more often than glutamate did (Lamour et al., 1988). These data suggest that at least some of the neurons lacking a receptive field receive an important input from the periphery, and that a large part of this excitatory input is shut down by GABA-ergic inhibition. Laminar Distribution of the Sample
In the present study, neurons were isolated rather uniformly from all cortical layers. In previous iontophoretic studies where glutamate has been used to excite otherwise quiescent neurons, the result has been a more uniform samplhg of neurons throughout the somatosensory cortex than was true for earlier studies done without glutamate (Mountcastle, 1957; Towe et al., 1964; Morse et al., 1965). Nevertheless, even those studies employing glutamate yielded more neurons in the middle layers than in the upper and lower layers in both cats (Dykes et al., 1984; Metherate et al., 1988; Tremblay et al., 1988) and rats (Dykes and Lamour, 1988b). In contrast, Tremblay et al. (1990), using the same anesthetic and same type of recording electrode as in the present study, obtained a distribution that more closely resembled the distribution reported here. Although differences in anesthetic might be responsible for reducing the number of cells found in the upper and lower layers, the electrode type might also be a determining factor. In the present experiments, as in the experiments of Tremblay et al. (1990), glass pipettes were used, whereas in previous iontophoretic studies carbon fiber electrodes were used and more neurons could be isolated in the middle layers; this suggests that carbon fiber electrodes more readily isolate the neurons found in the middle layers. Histological studies show that in the somatosensory cortex, the bottom of layer I11 and layer IV receive denser thalamic innervation than lower layers (Jones, 1975). Furthermore, the dendritic arbors of most pyramidal neurons of layer 111, V, and VI extend into the layers where the densest thalamic innervation is found, and thereby are in a position to receive numerous thalamic inputs (Hendry and Jones, 1983). In that context, it is interesting to note that a large proportion of NoRF neurons were found in layers I and VI, while GlutRF neurons were concentrated in layer 11, upper layer 111, and layer V; this suggests that the probability of finding demonstrable somatic input decreased as the neurons were located farther away from the middle layers. Higher proportions of spontaneously active neurons were found in layers containing high proportions of
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
NEURONAL POPULATIONS IN SOMATOSENSORY CORTEX
neurons displaying a receptive field. This distribution is similar to that found by Tremblay et al. (1990) in cats and by Dykes and Lamour (1988b) in rats. The most striking feature is the low proportion of spontaneously active neurons displaying a receptive field found in the lower part of layer I11 as compared to the middle of layer 111and layer IV. The reason for this is unknown, but neurons in layer I11 also displayed a higher rate of spontaneous activity. Neurons lacking a receptive field were more likely to be spontaneously active in upper and lower layers than in the middle layers. Swadlow (1989) reported that most efferent neurons (layers V and VI) had low spontaneous activity, and few had receptive fields. Although we have no information about the efferent targets of the cells studied, the low proportions of spontaneous activity neurons and of cells with receptive fields in these deeper layers where the efferent cells are located are consistent with that report. Layer IV and the bottom third of layer I11 contain the highest densities of GABA-immunoreactiveterminals (Hendry et al., 1987) suggesting that neurons located in these layers are more strongly inhibited than in other layers. Neurons located there were also less sensitive to glutamate than neurons found in the layers just above or below, perhaps because of this strong GABA-ergic inhibition. In conclusion, neurons with a receptive field display a higher probability of being spontaneously active than neurons lacking a receptive field. It is not clear whether the spontaneous activity is generated by the thalamocortical input or whether the ongoing activity is generated by another input. Spontaneously active neurons are more readily activated by glutamate than silent neurons, but sensitivity to glutamate is not dependent on the presence of a receptive field.
ACKNOWLEDGMENTS
We thank G. Filosi, G. Gauthier, and D. Cyr for the artwork; F. Cantin for the histological preparations; and C. Champagne for typing
the manuscript. This work was supported by the Medical Research Council of Canada. Richard A. Warren received a fellowship from the Medical Research Council of Canada.
REFERENCES
DYKES, R. W., and Y. LAMOUR (1988a) Neurons without demonstrable receptivefields outnumberneurons having receptive fields in samples from the somatosensory cortex of anesthetized or paralyzed cats and rats. Brain Res. 440: 133-143. DYKES, R. W., and Y. LAMOUR (1988b) An electrophysiologicalstudy of single somatosensory neurons in rat granular cortex serving the limbs: A laminar analysis. J. Neurophysiol. 60: 703-724.
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. 52: 1066-1093. HASSLER,R., and K. MUHS-CLEMENT (1964) Architektonischer aufbau des sensomotorischenund parietalen cortex der katze. J. Hirnforsch. 6: 377-420. HENDRY,S. H. C., and E. G. JONES (1983) The organization of pyramidal and non-pyramidal cell dendrites in relation to thalamic afferent terminations in the monkey somatic sensory cortex. J. Neurocytol. 12: 277-298. HENDRY, S. H. C., H. D. SCHWARK, E. G. JONES, and J. JAN (1987) Numbers and proportions of GABA-immunoreactive neurons in different areas of monkey cerebral cortex. J. Neurosci. 7 15031519. HICKS,T. P. (1984) The history and development of microiontophoresis in experimental neurobiology. Prog. Neurobiol22: 185-240. HICKS,T. P., and R. W. DYKES (1983) Receptive field size for certain neurons in primary somatosensory cortex is determined by GABAmediated intracortical inhibition. Brain Res. 274: 160-164. JONES, E. G. (1975) Lamination and differentialdistribution of thalamic afferents within the sensory-motor cortex of the squirrel monkey. J. Comp. Neurol. 160: 167-204. LAMOUR, Y., P. DUTAR,A. JOBERT, and R. W. DYKES(1988) An iontophoretic study of single somatosensory neurons in rat granular cortex serving the limbs: A laminar analysis of glutamate and acetylcholine effects on receptive field properties. J. Neurophysiol. 60:725-750. LANDRY, P., P. DIADORI, and R. W. DYKES(1984) Interhemispheric reciprocal interaction between ventropostemlateral thalamic nuclei involving cortical relay neurons. Brain Res. 323: 138-143. LANDRY, P., and R. W. DYKES (1985) Identification of two populations of corticothalamic neurons in cat primary somatosensory cortex. Exp. Brain Res. 60: 289-298. METHERATE, R., N. TREMBLAY, and R. W. DYKES (1988) The effects of acetylcholine on response properties of cat somatosensory cortical neurons. J. Neurophysiol. 59: 1231- 1252. MORSE,R. W., R. J. ADKINS,and A. L. TOWE(1%5) Population and modality characteristics of neurons in the coronal region of somatosensory area 1 of the cat. Exp. Neurol. 11: 419-440. MOUNTCASTLE, V. B. (1957) Modality and topographic properties of single neurons of cat’s somatic sensory cortex. J. Neurophysiol. 20: 408-434. RANDOLF,M., and J. SEMMES (1974) Behavioral consequences of selective subtotal ablations in the postcentral gyrus of Macaca mulatta. Brain Res. 70: 55-70. SCHNEIDER, S. P., and E. R. PERL (1988) Comparison of primary afferent and glutamate excitation of neurons in the mammalian spinal dorsal horn. J. Neurosci. 8: 2062-2073. SIEGEL,S. (1956) Nonparametric Statistics for the Behavioral Sciences, McGraw-Hill, New York. SMITS,E., D. C. GORDON, S. WIITE, D. D. RASMUSSON,and P. ZARZECKI (1991) Synaptic potentials evoked by convergent somatosensory and cortical inputs in raccoon somatosensory cortex: Substrates for plasticity. J. Neurophysiol. 66: 688-695. SOUL, R. R., and F. J. ROHLF(1981) Biometry, W. H. Freeman, San Francisco. SRETAVAN, D., and R. W. DYKES(1983) The organization of two cutaneous submodalities in the forearm region of area 3b of cat somatosensory cortex. J. Comp. Neurol. 213: 381-398. SWADMW, H. A. (1989) Efferent neurons and suspected interneurons in S-I vibrissa cortex of the awake rabbit: Receptive field and axonal properties. J. Neurophysiol. 62: 288-308.
311
WARREN AND DYKES TOWE,A. L., H.D. PATTON,and T. T. KENNEDY (1964) Response properties of neurons in the pericruciate cortex of the cat following electrical stimulation of the appendages. Exp. Neurol. 10: 325-
344.
Somatosens Mot Res Downloaded from informahealthcare.com by University of Otago on 01/04/15 For personal use only.
TREMBLAY, N., R. A. 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 palanine. Neuroscience 26: 745-762. TREMBLAY, N., R. A. WARREN, and R. W. DYKES(1990) Electrophysiological studies of acetylcholine and the role of the basal forebrain in the somatosensory cortex of the cat: I. Cortical neurons excited by glutamate. J. Neurophysiol. 64: 1199- 1211.
312
WARREN, R. A,, and R. W.DYKES(1993a) Studies of norepinephrine with selective agonists and antagonists administered iontophoretically to somatosensory cortical neurons in halothane-anesthetized cats. J. Neurophysiol. In press. WARREN,R. A., and R. W. DYKES(1993b) Long-lasting effects of norepinephrine administered iontophoretically to somatosensory cortical neurons in halothane-anesthetized cats. J. Neurophysiol. In press. WELKER,C. (1971) Microelectrode delineation of fine grain somatotopic organization of SmI cerebral cortex in albino rat. Brain Res. 26: 259-276.
