| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Physiology (2002), 539.2, pp. 511-522
© Copyright 2002 The Physiological Society
DOI: 10.1113/jphysiol.2001.013004
 |
ABSTRACT |
|---|
 TopAbstract IntroductionMethodsResultsDiscussionReferences |
|---|
Unique formation of rodent cortical barrels by layer 4 neurones attracts study of the sensory function of cortical input stage neurones (layer 4) compared with that of output stage neurones (layer 5). We have recorded extracellular responses from rat somatosensory cortical neurones to deflections of contralateral vibrissae. Thirty-two layer 4 barrel neurones and 29 layer 5b neurones were studied. Whisker stimulations were ramp-and-hold deflections with one of six different ramp velocities (100-2.5 mm s-1) and one of four different plateau amplitudes (2000-200 µm). Twenty-four (64) different stimulus forms were applied to the tip of a whisker trimmed to 10 mm in a predetermined order in stimulus cycles of 20-50 repetitions. Spike counts for a period of 2560 ms in 10 ms bins were summed to construct a matrix of 24 peristimulus histograms for each neurone. Twenty-four amplitude and 24 velocity values were computed from counts during the plateau and ramp phases, respectively. To determine the amplitude- and velocity dependence of a neurone, an amplitude F value (the ratio of variations among-/within-amplitude of the amplitude value) and a velocity F value (ratio of variations among-/within-velocity of the velocity value) were derived by analysis of variance. The amplitude F value of the layer 4 barrel neurones was greater than that of the layer 5b neurones (P < 0.0001). The velocity F value of the barrel neurones was smaller than that of the layer 5b neurones (P = 0.0226). The results suggests that barrel neurones and layer 5b neurones tend to detect amplitude and velocity components of whisker deflection, respectively.
(Resubmitted 17 July 2001; accepted after revision 22 November 2001)
Corresponding author M. Ito: Nijigaoka Garden 2-304, Uesono-cho 1-9-3, Meito-ku, Nagoya 465-0077, Japan. Email: pokorny{at}md.neweb.ne.jp
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
In the rodent somatosensory cortex there is a unique formation of granular cells, called barrels, which represent a brain map of the vibrissa layout on the face, this can be visualized using routine anatomical methods. The barrels appear in layer 4 where sensory signals from the periphery first arrive in the cortex and most likely function as the input stage of intracortical information processing.
Originally, like the relay neurones in the ascending lemniscal pathway (Zucker & Welker, 1969; Waite, 1973; Shipley, 1974), the neurones in a barrel were thought to respond to a single vibrissa only (Welker, 1976; Simons, 1978). Later studies, however, showed that in the thalamic ventrobasal relay nucleus (Ito, 1988; Simons & Carvell, 1989) and in the cortical barrels (Ito, 1985; Simons, 1985) a substantial proportion of neurones are multiwhisker responding. Simons and co-workers suggest a correlation between spiny versus non-spiny stellate cells and single versus multiwhisker barrel neurones in layer 4 (Simons & Carvell, 1989; Kyriazi & Simons, 1993; Brumberg et al. 1996; see also Woolsey et al. 1975; Simons & Woolsey 1984; McCormick et al. 1985). They hypothesized that the latter inhibit the former thus shaping sharp receptive fields (RFs) by lateral inhibition. Multiwhisker receptive fields of cortical output neurones in deep layers are again elaborated by the sequential process of transformation of signals from layer 4 through superficial to deep layers (Simons, 1978).
Clearly, spatial response features such as RF sizes can be more distinctly delineated in the rodent vibrissa, but there is a relative paucity of information regarding temporal response features including the frequency characteristics and synaptic transfer functions in the rodent vibrissa system. Our previous series of quantitative measurements of tuning curve slopes compared frequency domain characteristics between the barrel neurones (input neurones) and layer 5 pyramidal neurones (output neurones) and led us to suggest the mode of interaction between the two classes of neurones (Ito, 1981, 1985; Ito et al. 1981). It was shown that while layer 5 neurones tended to detect the velocity component of whisker deflection, layer 4 neurones were more dependent on the amplitude component. However, the calculation of the tuning curve slope was based on threshold determination which involved subjective judgements by the experimenter.
The present study aims at clarifying this issue using entirely objective measures and a robust statistical means. A new technique for characterizing these responses efficiently by the use of programmed whisker deflection and spike collection was developed to resolve the amplitude versus velocity aspects of responses on the basis of the analysis of variance (ANOVA). ANOVA is widely used in assessing statistical significance of differences between groups of data, but the present report is probably the first to describe the response properties of neurones as F values using ANOVA. Every neurone was identified histologically in relation to the barrels. The results from the present study lead us to suggest again that neurones residing in a barrel in layer 4 tend to encode the amplitude component of whisker deflection while layer 5b neurones are more dependent on the velocity component of the stimulus.
|
METHODS |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Animals and preparation
Adult white rats of Wistar strain were used. Under urethane anaesthesia (1.4 g (kg body weight)-1 I.V.) the rat was placed in a stereotaxic apparatus and a small burr hole was made over the left somatosensory barrel cortex. The dura mater was excised to expose the whisker area which was then covered with mineral oil to prevent it drying out. Experimental procedures were approved by the Ethics Committee on Animal Experiment of the Institute for Developmental Research.
Recordings
Extracellular single-unit recordings were made using glass microelectrodes pulled from borosilicate capillaries containing a filament (GC200F-15, Clark Electromedical Instruments, Reading, UK). The electrodes were back-filled with Methyl Blue saturated in 1 M potassium acetate. The tip of the electrode was broken to a diameter of ca 2-3 µm, the DC resistance was 3-5 M
. With this type of electrode, neuronal spikes were typically negative-positive configurations of 1-2 mV. The microelectrode potentials were amplified to a line level (in the order of 1 V) by the use of a Nihon Kohden type MZ-P1 and BMP-2 amplifier and displayed on a CRO and fed to a Schmitt trigger circuit (Bradley et al. 1967) the outputs of which, 5 V square pulses of 1 ms duration, were in turn fed to an 8255-based input-output interface of a computer (see below). The layer in which recording was made was always determined by deposit of the dye Methyl Blue, which was ejected iontophoretically by passing a negative DC of 10 µA for 3 min. Only one neurone was studied in each rat.
Whisker stimulation
Computer generated waveforms (see below) were fed to a DC power amplifier to drive a pen galvanometer that was fitted to the principal whisker (Simons, 1978), which was determined manually for each neurone by listening to the audiomonitor. As we already know the whereabouts of the central barrel cortex, the principal whisker was usually one of the central whiskers (B 1-3, C 1-3, D 1-3). The direction of whisker deflection was upwards (dorsal) or downwards (ventral). Neurones that were clearly more sensitive to directions other than dorsal or ventral were discarded (Ito, 1981). Quantitative stimulation was then applied dorsally or ventrally according to the preferred direction of the neurone.
Automated sequence of stimulus waveforms and data acquisition
A computer controlled the timing of stimulation and spike data collection. Ramp-and-hold deflections with six different ramp velocities (100, 50, 20, 10, 5 and 2.5 mm s-1) and four different plateau amplitudes (2000, 1000, 400 and 200 µm) were applied to the tip of a whisker trimmed to 10 mm. The four final amplitudes corresponded to 11.3, 5.71, 2.29 and 1.15 deg in angle. A stimulus trial consisted of a prestimulus period of 400 ms, a ramp-and-hold period of 1400 ms and a falling phase at a velocity of 10 mm s-1. Thus the largest deflection (2000 µm) with the slowest ramp (2.5 mm s-1) had the longest ramp phase of 800 ms (2000 µm divided by 2.5 mm s-1) and the shortest plateau phase of 600 ms (1400 minus 800 ms) among the 24 stimulus waveforms. Stimuli were given every 2560 ms and a stimulus cycle consisted of 24 (4 6) different stimuli, thus lasting for ~1 min (2.5 24 s). Between stimulus cycles there was a pause of ca 20 s during which 24 cumulative peristimulus histograms were refreshed and displayed on a monitor through the next stimulus cycle. The time courses of deflections are summarized in superposition of four different amplitudes for five faster velocities out of a total of six velocities used in Figs 1, 2, 3, 5 and 6. Only a period of 200 ms preceding the onset of whisker deflection and the first 800 ms thereafter are shown in these figures. The computer used was an 8086-based 16-bit machine with a clock-cycle of 6 MHz. Every millisecond the computer checked the high or low of the output level of the Schmitt trigger that was transforming a neuronal spike to a 1 ms pulse. Thus, no spikes were overlooked or counted twice. Twenty to fifty cycles were repeated and the number of spikes in a 10 ms bin was added to construct a peristimulus histogram of 256 bins for each of the 24 parameter combinations.
Statistics on individual neurones
The following statistical measures were defined to characterize the response features of individual neurones.
Twenty-four amplitude values. The amplitude value was defined for each of the 24 histograms. The cumulative spike numbers were further summed over the first 600 ms period (60 bins) beginning 20 ms after the hold (plateau) phase was reached and divided by 60 (see horizontal bars in Fig. 1 and others). The amplitude value was the mean accumulated spike number per 1 bin (10 ms). This delay of 20 ms avoided inclusion of phasic responses following the ramp onset. For the purpose of explanation, examples of amplitude values of a layer 4 neurone are given in Fig. 1.
 |
View larger version [in this window] [in a new window] |
|
|
Figure 1. Response profiles and amplitude values of a barrel neurone Upper panels, stimulus waveforms of four plateau amplitudes (2000, 1000, 400 and 200 µm) are shown for each of five velocities (100, 50, 20, 10 and 5 mm s-1). In this and subsequent figures the slowest deflections (2.5 mm s-1) are omitted but the data have been included in computation of the F value in each neurone. Only the prestimulus period of 200 ms and the first 800 ms period of the ramp-and-hold phase are depicted. Lower panels, corresponding histograms (in 4 | ||
Twenty-four velocity values. The velocity value was defined for each histogram by the peak of spike counts contained within a time interval between the ramp onset to 20 ms after the end of the ramp of the 2000 µm deflection for each velocity level. A velocity response was expected to occur during this time window. The additional 20 ms period (2 bins) made allowance for the latency of responses. The same time window was adopted for four deflection amplitudes at a given velocity (i.e. also for deflections of < 2000 µm; see horizontal bars in Fig. 2 and others), but this practice did not affect the result since a peak value was searched for during this time window. Velocity values of the layer 4 neurone are given in Fig. 2.
 |
View larger version [in this window] [in a new window] |
|
|
Figure 2. Same response profiles as shown in Fig. 1 this time with velocity values Horizontal bars indicate time windows during which maximum counts were searched for. The velocity value refers to the highest column during the time window. The velocity F value computed from 24 velocity values was only 1.88 for this neurone (see Table 2). | ||
Amplitude F value. An amplitude F value was computed using 24 amplitude values for each neurone. One-way ANOVA tests whether or not there are differences in the amplitude value, either among or within amplitudes. Here the amplitude corresponds to 'treatment' in ANOVA terminology. The rationale for this computation was that a pure amplitude-dependent neurone, at each of four amplitudes, should have the same amplitude value for six different velocities; no within-amplitude variations. In practice, the amplitude values were corrected by subtraction of the background activity calculated during a 200 ms period preceding the stimulus. Treatment (among amplitude) sum of squares, treatment mean square, error (within amplitude) sum of squares and error mean square were computed (Steel & Torrie, 1980). The F value was obtained by dividing the treatment mean square by the error mean square.
Velocity F value. Similarly, a velocity F value was computed from 24 velocity values for each neurone. One-way ANOVA computed the velocity F value by comparing variations of velocity values among velocities to those within velocities. Again, for this computation, the velocity values were corrected by subtraction of the background activity calculated during a 200 ms period preceding the stimulus. Treatment (among velocity) sum of squares, treatment mean square, error (within velocity) sum of squares and error mean square were computed to obtain the velocity F value.
Transient responses and F values. A time window comparable to that used by Pinto et al. (2000) is described and discussed together with the results in a subsection of Discussion.
Two-way ANOVA. For reasons explained later, a two-way ANOVA was also calculated from the 24 velocity values of each neurone to derive the velocity dependence of velocity values using the amplitude and velocity as two factors (velocity F2).
Significance of difference in the parameters between neurone groups. Since the distribution of measured F values was not normal, non-parametric statistics (Mann-Whitney's U test or Kruskal-Wallis test) were used to evaluate the significance of the data sampled from individual neurones. Correlation between parameters was also tested using non-parametric statistics (Spearmann rank correlation coefficients).
Grand totals of spike data and minor variations of statistical tests. Spike data of individual neurones were normalized to 50 repetitions of stimulus presentations and averaged for totals of 32 barrel and 29 layer 5b neurones. Averaged data were expressed in terms of mean number of spikes in every 10 ms. Amplitude F and velocity F values were computed for the two grand totals. In addition to the 'standard' amplitude values involving the first 600 ms period, spikes were also counted during the first 400 and 200 ms of the plateau phase to evaluate the amplitude dependence of early responses of the plateau phase. The F values were computed and were referred to as amplitude F-400 and amplitude F-200 values. Further, as a variant of velocity F value, the 3-point velocity F value was computed, based on a matrix of twenty-four 3-point velocity values, each being the average of three consecutive bins centred on the maximum spike count adopted originally as the velocity value.
Histology
In extracellular recording experiments, the animals were killed by an overdose of Nembutal (Abott, North Chicago, IL, USA), and the brain was removed from the skull. Frozen sections were cut at 100 µm in a plane tangential to the whisker field of the somatosensory cortex. The sections were collected in 0.1 M phosphate buffer (pH 7.4), which was then replaced by incubation solution for succinic acid dehydrogenase histochemistry (SDH; Nachlas et al. 1957; Kato et al. 2000). The sections then underwent fixation and were immersed briefly in 1 % gelatin and mounted on slides. Using a camera lucida, the contours of individual barrels were drawn on a sheet of paper at a magnification of 40. A barrel in which a dye spot was found was subdivided into 3 concentric zones of equal area according to Crandall et al. (1986). The positions of recording/staining could thus be divided into one of outer, intermediate and inner areal thirds of barrels.
|
RESULTS |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
In extracellular recording experiments, we isolated 61 neurones from the barrel cortex in 61 rats. All these neurones were identified histologically by dye injection for the site of recording. Since the neurones were not stained intracellularly, their anatomical cell types were not determined. Those neurones whose dye spots were found within a barrel were referred to as barrel neurones and those whose spots were below the barrels were termed non-barrel neurones. Neurones within the boundary of any barrel were considered layer 4 neurones. There was no way of knowing precisely whether a dye spot below the barrels was in layer 5 or 6 in the SDH-stained tangential sections. From the depth scale of the microelectrode manipulator (1000-1500 µm below pia mater) and the depth of the dye spot below the barrel level, however, we concluded that most of the non-barrel neurones were most probably in layer 5b. In addition, there was one case where the dye spot was found between barrels in layer 4, but this was discarded from our present sample. Of the 32 dye spots, six were found in barrel C2, 18 were in the neighbouring barrels. An impression was obtained that dye spots were deviated toward the periphery of a barrel, but the periphery has more chances of being hit by an electrode by a factor of the square of the distance from the gravity centre. Therefore, deviation of a dye spot from the barrel centre was calculated using a camera lucida according to Crandall et al. (1986), and each barrel neurone was categorized into one of inner, intermediate and outer equiareal concentric zones. The 32 barrel neurones could be divided into four inner, 10 intermediate and 18 outer neurones. Preferential location of barrel neurones in the periphery was significant (
2 = 9.25, d.f. = 2, P < 0.01), but there were no differences among these three areal groups of barrel neurones in either the amplitude F or velocity F value (Kruskal-Wallis test).
Response histograms of individual neurones
Figure 1 shows the result of a barrel neurone and illustrates the format of data display adopted in the present experiments. It shows, in each column of four peristimulus time histograms, the effect of changing the hold amplitude for a given ramp velocity. It also shows the amplitude value for each histogram and a time bar on which the amplitude value was based. The amplitude F value was 26.12, which was highly significant allowing this neurone to be called amplitude dependent (Table 1). Figure 2 repeats the histograms of the same neurone shown in Fig. 1, the only difference being that velocity values were computed from the histograms and the time windows for determining the velocity values are shown. Although there was a general tendency that the velocity value increased with increases in velocity, there were variations among amplitudes (within-velocity variations) thus cancelling out the among-velocity variations. The velocity F value was 1.88 which was not sufficient to be called velocity dependent (Table 2).
Figure 3 represents a typical layer 5b neurone which showed transient responses during the ramp phase, the magnitude of response being highly dependent on the velocity, i.e. there was a clearly discernible overall difference among the means of velocity values of the six velocity groups and within-velocity variations were minimal. The velocity F value was 33.33 (Table 3).
 |
View larger version [in this window] [in a new window] |
|
|
Figure 3. Response profiles from a typical layer 5b neurone which showed transient responses The velocity F value was 33.33 (see Table 3). | ||
Significance of ANOVA (Table 4)
Irrespective of whether a neurone was amplitude dependent or velocity dependent, each neurone had an amplitude F value and a velocity F value. The distributions of F values are plotted in Fig. 4. The amplitude F value of the barrel neurones ranged from 0.36 to 74.44 with a median of 2.67, whereas the amplitude F value of layer 5b neurones ranged from 0.12 to 7.55 with a median of 0.81. The Mann-Whitney U test revealed that the amplitude F values of the barrel neurones were significantly larger than those of layer 5b neurones (Z = 4.74, P < 0.0001, two-tailed). The velocity F value of the barrel neurones ranged from 0.21 to 6.81 with a median of 1.32. A range from 0.47 to 33.33 was obtained for layer 5b neurones with the median of 2.14. Overall, the velocity value was greater in layer 5b neurones (Mann-Whitney U test, Z = -2.28, P = 0.0226).
 |
View larger version [in this window] [in a new window] |
|
|
Figure 4. Amplitude F and velocity F values of layer 4 barrel and layer 5b neurones The amplitude F (upper half) and velocity F (lower half) values of each neurone are plotted on the horizontal axis. Dashed lines indicate F values of 3.10 and 2.77 which are critical for matrices of 4 amplitudes (treatment) | ||
For a matrix of 4 (amplitude) 6 (replication in ANOVA terminology) adopted in the present study, amplitude F values larger than 3.10 were significant at P < 0.05 for the amplitude dependence. According to this criterion, 12 of 32 barrel neurones but only two of 29 layer 5b neurones were significantly amplitude dependent. Conversely, for velocity dependence, the matrix of 6 (velocity) 4 (replication) the critical velocity F value was 2.77. Thus only five of the 32 barrel neurones but 12 of the 29 layer 5b neurones were significantly velocity dependent.
Absence of negative correlation between amplitude F and velocity F values
Among the 32 barrel neurones there were only three neurones that were significantly amplitude- and velocity dependent. Similarly, among the 29 layer 5b neurones only one was significantly amplitude and velocity dependent. It appeared as though there were negative correlations between the two measures. This suspicion also came from the 'trade-off' between the amplitude and velocity dependence inherent in tuning curve determination based on threshold measurement (Tuckett et al. 1978; Ito et al. 1981). However, Spearmann rank correlation coefficients between the two measures were 0.15, 0.26 and -0.03 for the barrel neurones, layer 5b neurones, and the both combined respectively. None of the coefficients were significant.
Spontaneous discharge rates and response characteristics
Another variable examined was the spontaneous discharge rate defined, for each neurone, by averaging spike numbers during the 200 ms prestimulus period of the 24 histograms and expressed in terms of spikes per second. The spontaneous discharge rate of barrel neurones ranged from 0 to 17.6 Hz with a median of 2.6 Hz and that of the layer 5b neurones from 0.2 to 17.5 Hz with a median of 4.7 Hz. This difference in the median value between these two groups of neurones was not significant (Mann-Whitney U test, Z = -1.23, P > 0.05). In either of two groups of neurones, the spontaneous discharge rate did not appear to be correlated with the amplitude F or velocity F value (Spearmann rank correlation coefficients; the largest Rs was 0.33 between the velocity F of layer 5b neurones and their background firing rates, and yet t (= 1.83) did not reach statistical significance (P = 0.05) (t = 2.05 for n = 29).
Grand totals of 32 barrel and 29 layer 5b neurones as population histograms
Amplitude F and velocity F values of population histograms. Figure 5A and B shows matrices of pooled histograms of the 32 barrel neurones and 29 layer 5b neurones. Both A and B show an abrupt increase in spike rate during ramp and a moderate but sustained increase (relative to prestimulus period) during plateau phase. However, a close inspection discloses an obviously increased sustained discharge with increasing the plateau amplitude in the former group. Indeed, the amplitude F values from the two populations were 17.59 and 2.27, respectively, remarkably greater than the medians calculated individually (2.67 and 0.81, respectively). The velocity F values calculated for the two populations were 5.41 and 9.13, respectively, respective medians of individual neurones were 1.32 and 2.14. Improvements in F values compared with those of individual neurones were undoubtedly due to the reduction of within-treatment variations.
 |
View larger version [in this window] [in a new window] |
|
|
Figure 5. Grand totals of the neurones studied A, grand totals of 32 layer 4 barrel neurones. Histograms of each neurone were normalized to 50 repetitions of stimulus cycles and averaged. Note that when data from 32 layer 4 barrel neurones were averaged, increases in stimulus amplitude resulted in increases in spike count during the plateau phase. Upper and lower horizontal bars beneath the histograms denote the time windows used for deriving velocity and amplitude (values not shown), respectively. B, grand totals of 29 layer 5b neurones. Histograms were prepared as in A. The population profile of layer 5b neurones is similar to that of layer 4 barrel neurones. However, there are differences in F values between the two groups of neurones, especially in the amplitude F value (see text). | ||
Amplitude F-400 and F-200 values. The choice of a 600 ms period for standard evaluation of the plateau activity was related to the plateau period of the 2000 µm deflection at the slowest ramp velocity. It took 800 ms (2000 µm in 2.5 mm) to reach the plateau which then lasted only 600 ms (1400 minus 800 ms). We were also interested in the amplitude dependence of the early responses of the plateau phase. Instead of the original amplitude values based on the 600 ms, spike counts during the first 400 and 200 ms of the plateau phase were used to derive amplitude F-400 and amplitude F-200 values (see Methods). The amplitude F-400 and amplitude F-200 values for the population histogram of barrel neurones were 16.97 and 16.09, respectively. The corresponding values of layer 5b neurones were 3.43 and 3.25, respectively. The measure of amplitude dependence was affected little by the choice of time window.
Three-point velocity F value. The velocity value so far was defined as the maximum spike count during the ramp period. However, the peak activity could be split in two neighbouring bins. To compensate for this possible underestimation of the peak spike count, 3-bin averages (see Methods) were calculated to replace the velocity values of the grand totals. 3-point velocity F values obtained thereof were 4.47 and 11.03 for barrel and layer 5b neurones, respectively. They were not substantially different from 'standard' velocity F values of 5.41 and 9.13, respectively, from the grand totals.
One-way versus two-way ANOVA as applied to the velocity data
The main objective of our study was to examine the effects of amplitude and velocity separately since a purely amplitude- (or velocity)-dependent neurone should in principle have no within-amplitude (or -velocity) variations. Using a one-way ANOVA, layer 5b neurones were more velocity dependent relative to barrel neurones. However, as Fig. 6 shows the transient response during the ramp period was often not only dependent on the velocity but also to a certain degree dependent on the amplitude. Figure 6 shows that within-velocity variations of velocity values (for example 24, 16, 7, 5 of 100 mm s-1 in the right-end column) were sufficiently large to counteract among-velocity variations of velocity values obtained for six different velocities. In this case the velocity F value of 2.29 fell short of the critical value of significance (F = 2.77) although the transient response and its dependence on the velocity were obvious. The velocity dependence of these neurones might have been underestimated (see also Fig. 3).
 |
View larger version [in this window] [in a new window] |
|
|
Figure 6. Interaction between amplitude and velocity Additive contribution of velocity and amplitude on the transient response of a layer 5b neurone. Note that depression following excitation was typical for layer 5b neurones. | ||
The velocity dependence under the combined effect of amplitude and velocity should alternatively be evaluated in a two-way ANOVA, and individual neurones underwent re-examination using the same velocity values to compute the velocity F2 values (see Methods). After all, the velocity F2 values in two-way ANOVA turned out to be similar to the velocity F values in one-way ANOVA mentioned earlier. The result again shows that layer 5b neurones, with velocity F2 values of 0.38-31.81 (median = 2.23), tended to be more oriented to velocity detection than the barrel neurones (0.18-15.22 with median of 1.57, Mann-Whitney U test, Z = 1.91, P = 0.0562). Such an additive contribution of the velocity and amplitude on the transient response has been reported in detail for the muscle spindle stretch responses in the cat (Houk et al. 1981; also see Shipley, 1974; Pinto et al. 2000).
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
ANOVA analysis of spike data
There are two general approaches for describing dynamic responses of vibrissa neurones studied by extracellular unit recordings. The first is to construct an ensemble of input-output curves with the frequency or ramp slope of vibrissa deflection on the abscissa and the magnitude of response on the ordinate at different stimulus intensities (Shipley, 1974; Simons, 1978). The second is a threshold tuning curve where the frequency (or time constant) of vibrissa deflection is plotted on the abscissa and the threshold stimulus intensity (amplitude) on the ordinate (Zucker & Welker, 1969; Ito, 1981, 1985; see also Tuckett et al. 1978 for cutaneous mechanoreceptors). The first analysis is usually based on a matrix of response histograms such as those presented also in our present report and data acquisition can be automated, but it has not been possible to express the response profile in one or two variables that describe the basic response properly so that different neurones can be compared on one or two common scales. The second analysis reduces the response feature to a value of tuning curve slope, but requires repeated presentation of stimuli with intensities near the threshold until the threshold value is determined (at every frequency tested) and yet this decision is made on a subjective basis (Ito, 1981). Moreover, with threshold tuning slope determination, there is an inherent trade-off between amplitude and velocity dependence. A given neurone cannot be highly amplitude-sensitive and velocity-sensitive at the same time. Therefore, such neurones, if any, could have escaped screening.
The present method incorporates the statistical technique of analysis of variance directly into evaluation of spike numbers. Based on the quantitative data acquisition, the response characteristics are expressed by only two parameters: amplitude F and velocity F values. With the present method, however, as different time windows (transient and plateau phases) after a stimulus are looked at, a neurone can theoretically have a high amplitude F and a high velocity F value. The present study shows that there are only a few neurones that are significantly amplitude and velocity dependent at the same time. Yet there was no significant negative (or positive) correlation between the amplitude F and velocity F value for barrel neurones, layer 5b neurones or both combined. These findings suggest that the present ANOVA method, examining a different aspect of responses, is a useful alternative to the previous approach and confirmed that barrel neurones and layer 5b neurones are relatively amplitude and velocity dependent, respectively.
Preferential location in periphery of barrel neurones responding to vibrissa deflection
Anatomical studies have shown in the mouse that neurones in layer 4 are more densely distributed in the side than in the hollow of a barrel or in the septum between the barrels, but in the rat such differences in the packing density between the barrel sides and both the hollows and septa are less pronounced (Welker & Woolsey, 1974). Functionally identified layer 4 and layer 5 neurones have been found in register with the periphery of a barrel in the mouse (Crandall et al. 1978; Lin et al. 1985). Although not stated as such, vibrissa responding neurones in the rat layer 4 and layer 5 appear to be associated with the periphery of a barrel (Simons & Land, 1987, Fig. 1; Ito, 1992, Fig. 8). In addition to confirming these previous findings of eccentric distribution of barrel neurones, the present result revealed different functions of the barrel neurones when compared with non-barrel neurones in deeper layers.
Differential laminar distribution of the amplitude- and velocity-dependent neurones in the barrel cortex
The main finding of the present study was that there was a remarkable difference in the amplitude dependence between layer 4 barrel neurones and layer 5b neurones. The barrel neurones had much higher amplitude F values that were defined by statistical analysis of variance method (P < 0.0001). In contrast, layer 5b neurones had higher velocity F values than the barrel neurones (P < 0.05). This conclusion is reminiscent of our previous result based on comparison of the tuning curve slopes of different cortical layers of the rat barrel cortex; layer 4 neurones were more oriented for amplitude detection and layer 5 neurones for velocity detection (Ito, 1985). However, before going into discussion of the result in the framework of amplitude/velocity detection, comparison of the present samples of neurones with the types of barrel cortex neurones in the literature would be in order. In the rat barrel cortex, Simons (1978) described regular-spike units (RSUs) and fast-spike units (FSUs) using extracellular in vivo recording, which was later confirmed by intracellular recording from slice specimens by Chagnac-Amitai & Connors (1989). The latter authors further subdivided RSUs into RS1 and RS2 based on the rate of adaptation of spike firing during direct intracellular stimulation. In the mouse barrel cortex, RS1 neurones show a relatively constant discharge rate after the initial transient phase and RS2 neurones undergo a continued adaptation of spike discharges (Agmon & Connors, 1992). Barrel neurones and layer 5b neurones in the present study may correspond to RS1 and RS2 of their mouse barrel cortex. In an in vivo study of the rat barrel cortex neurones comparable to the present study, Brumberg et al.(1996) reported that RS units had an averaged spontaneous rate of 1.8 Hz whereas FSUs had a much higher mean rate of 12.9 Hz (see also Armstrong-James et al. 1993). Our sample of barrel neurones covered a range of spontaneous discharge rates from 0 to 17.5 Hz with a median of 2.6 Hz. Considering that our layer 5b neurones had a median of 4.7 Hz and probably regular spikes from pyramidal neurones, most of the barrel neurones in the present study also were probably RSUs recorded from pyramidal neurones or spiny stellate neurones (Zhu & Connors, 1999).
Based on comparison between whisker responses of cortical layer 5b neurones of the normal rats with those of prenatally X-irradiated rats that are deprived of layer 4, we have previously suggested that normally there are two routes of flow of vibrissa information to layer 5b (output stage of cortex); a direct thalamo-layer 5b route may convey information regarding the velocity component of whisker deflection while an indirect thalamo-layer 4-layer 5b route may mediate the amplitude component. In fact, the remaining layer 5b neurones from the irradiated rats were purely velocity dependent. The present results are in line with the notion that the amplitude dependent response of layer 4 barrel neurones are differentiated with respect to time at the level of layer 5b neurones where the velocity dependent response of the direct route converges. An alternative explanation would be that the thalamic afferents convey information with a single class of sensitivity regarding velocity/amplitude detections but different laminar response characteristics are elaborated by differential intrinsic properties of target neurones (Zhu & Connors, 1999). Particularly relevant to this issue is the finding that the rat ventrobasal thalamic relay nucleus is composed of a single anatomical class of neurones (Harris, 1986, but see Armstrong-James & Callahan, 1991). If this was the case, cortical barrel neurones and layer 5b neurones in this study must possess different types of intrinsic properties (e.g. FSU versus RSU), hence different anatomical types (e.g. smooth stellate versus pyramidal neurones).
As regards the difference in the receptive field size between the neurones of the two layers, the extensive recurrent axonal collaterals (Chagnac-Amitai et al. 1990) and basal dendritic arborization (Ito, 1992) of layer 5b neurones should be taken into consideration.
Comparison with Pinto et al.'s experiments
In view of an apparent discrepancy with a recent demonstration by Pinto et al. (2000) that cortical barrel neurones are more oriented for velocity than for amplitude detection of whisker deflection, we reanalysed the data using a time window which is similar to that adopted by them. Here, we computed the integral of spike counts during a 30 ms (3 bins) period following the first 20 ms (allowance for response) of the deflection onset. This measure of the magnitude of transient response (not exactly the same as but comparable to Pinto's time window) was counted from the 24 histograms of individual neurones and referred to as 24 transient values. Further statistical procedures were same as in the main body of the present study, the only difference being that both the amplitude ANOVA and velocity ANOVA were based on these 24 transient values. Figure 7 shows the result in the same format as Fig. 4 and compares two groups of neurones. The amplitude F value of the barrel neurones ranged from 0.15 to 4.78 with a median of 1.30, whereas that of layer 5b neurones ranged from 0.07 to 6.66 with a median of 0.80 (Z = 2.25, P = 0.0244). The velocity F value of the barrel neurones ranged from 1.10 to 33.4 with a median of 2.46, whereas that of layer 5b neurones ranged from 0.35 to 22.2 with a median of 3.08 (Z = -0.90, P = 0.368).
 |
View larger version [in this window] [in a new window] |
|
|
Figure 7. F values in Pinto et al.'s mode Amplitude F values and velocity F values based on transient responses. The same format as in Fig. 4. An amplitude F value smaller than 0.1 is plotted as 0.1. n.s., not significant. | ||
Firstly, when the magnitude of amplitude dependence was compared with that of the velocity dependence in each group of neurones, the result of Pinto et al. (2000) was confirmed. Velocity F values were larger than amplitude F values in the barrel neurones (Z = -4.12, P < 0.0001), and the difference was even greater for the layer 5b neurones (Z = -5.03, P < 0.0001). Considering the difference in the significance level of F in the two parametres (3.10 versus 2.77, see Results), these estimates are on the safe side. Secondly, the difference in the amplitude dependence between the two groups of neurones was detected but found less remarkable compared with our conclusion (P = 0.0244 versus P < 0.0001, see Results) and the difference in the velocity dependence was not significant (P = 0.368). Thirdly, the apparent difference between Pinto et al.'s conclusion and our conclusion could probably be ascribed to the difference in the time window of spike sampling, and the fact that the amplitude dependence of barrel neurones is more enhanced when the sustained rather than the transient response during whisker deflection is analysed.
Functional implication
In our standard protocol, 12/32 layer 4 and 2/29 layer 5b neurones were significantly amplitude dependent while 5/32 layer 4 and 12/29 layer 5b neurones were velocity dependent. Considering that three layer 4 and one layer 5b neurones were significantly amplitude and significantly velocity dependent at the same time whilst the remaining 18 layer 4 and 16 layer 5b neurones were neither amplitude nor velocity dependent, one asks which physical component of whisker deflection is coded by these remaining neurones. They all clearly responded to whisker stimulation. Unexpectedly low F values are to be examined in two ways.
Velocity F values. Except for the three layer 4 and one layer 5b neurones mentioned above, if the velocity values are dependent not only on velocity but also on amplitude, the velocity F value is smaller than expected. This comes from a certain amount of 'within-velocity' variability (which is 0 in an ideally velocity dependent neurone) compared with 'among-velocity' variability (see Fig. 6). In this respect the present protocol is stringent. A possible contribution of acceleration is another problem, but it is difficult to study this aspect in an adequate way (under the present conditions, a stimulus form of constant acceleration would deflect the whisker beyond its physiological range).
Amplitude F value. Unlike with the conventional tuning curve construct in sensory physiology (Tuckett et al. 1978; Ito, 1981) where the threshold amplitude of a given frequency is judged to be reached when a discharge is evoked, a purely amplitude-sensitive neurone in the present analysis is assumed to maintain discharges over a period of 600 ms. A relatively small number of significantly amplitude-dependent neurones in layer 4 was unexpected in view of the previous conclusion (Ito, 1981). This discrepancy may be related to a wide variety of adaptation rates displayed by the neurones during the plateau phase of whisker deflection.
The same protocol was applied to all the whisker-responding neurones in layers 4 and 5b. There was a surprisingly large difference in the amplitude value and a significant difference in the velocity value between the two layers. Moreover, this method allows us to accumulate data from individual neurones to make population histograms. F values computed from these grand totals were dramatically increased as against the medians of F values of the individual neurones.
In summary, the present work is the first attempt to characterize each neurone's coding capacity using the F parameter of ANOVA and it was found that the rat cortical layer 4 barrel neurones are more oriented for amplitude coding and layer 5b neurones for velocity coding of whisker deflection when these two key stations of cortex (input and output stages) were compared. Using the same stimulus protocol, intracellular recording/staining should reveal the morphological substrate of the present data. Detailed description of somata and neurites (Chagnac-Amitai et al. 1990; Larkman & Mason, 1990; Mason & Larkman, 1990; Ito et al. 1998) is needed. Another important aspect is related to the genesis of burst discharges (Nakanishi & Kukita, 2000) in the neural element involved.
|
REFERENCES |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
| AGMON, A. & CONNORS, B. W. (1992). Correlation between intrinsic firing patterns and thalamocortical synaptic responses of neurons in mouse barrel cortex. Journal of Neuroscience 12, 319-329 | [Abstract] |
| ARMSTRONG-JAMES, M. & CALLAHAN, C. A. (1991). Thalamo-cortical processing of vibrissal information in the rat. II. Spatiotemporal convergence in the thalamic ventroposterior medial nucleus (VPm) and its relevance to generation of receptive fields of S1 cortical 'barrel' neurones. Journal of Comparative Neurology 303, 211-224 | [Medline] |
| ARMSTRONG-JAMES, M., WELKER, E. & CALLAHAN, C. A. (1993). The contribution of NMDA and non-NMDA receptors to fast and slow transmission of sensory information in the rat SI barrel cortex. Journal of Neuroscience 13, 2149-2180 | [Abstract] |
| BRADLEY, W., CONWAY, C., GLOVER, E. & MCCORMICK, S. (1967). Discriminator and integrator instrument for an on line frequency analysis of single unit discharges. Electroencephalography and Clinical Neurophysiology 22, 177-179 | [Medline] |
| BRUMBERG, J. C, PINTO, D. J. & SIMONS, D. (1996). Spatial gradients and inhibitory summation in the rat whisker barrel system. Journal of Neurophysiology 76, 130-140 | [Medline] |
| CHAGNAC-AMITAI, Y. & CONNORS, B. W. (1989). Synchronized excitation and inhibition driven by intrinsically bursting neurons in neocortex. Journal of Neurophysiology 62, 1149-1162 | [Medline] |
| CHAGNAC-AMITAI, Y., LUHMANN, H. J. & PRINCE, D. A. (1990). Burst generating and regular spiking layer 5 pyramidal neurons of rat neocortex have different morphological features. Journal of Comparative Neurology 296, 598-613 | [Medline] |
| CRANDALL, J. E., KORDE, M. & CAVINESS, V. S. JR (1986). Somata of layer V projection neurons in the mouse barrelfield cortex are in preferential register with the sides and septa of the barrels. Neuroscience Letters 67, 19-24 | [Medline] |
| HARRIS, R. M. (1986). Morphology of physiologically identified thalamocortical relay neurons in the rat ventrobasal thalamus. Journal of Comparative Neurology 251, 491-505 | [Medline] |
| HOUK, J. C., RYMER, W. Z. & CRAGO, P. E. (1981). Dependence of dynamic Response of spindle receptors on muscle length and velocity. Journal of Neurophysiology 46, 143-166 | [Medline] |
| ITO, M. (1981). Some quantitative aspects of vibrissa-driven neuronal responses in rat neocortex. Journal of Neurophysiology 46, 705-715 | [Medline] |
| ITO, M. (1985). Processing of vibrissa sensory information within the rat neocortex. Journal of Neurophysiology 54, 479-490 | [Medline] |
| ITO, M. (1988). Response properties and topography of vibrissa-sensitive VMP neurons in the rat. Journal of Neurophysiology 60, 1181-1197 | [Medline] |
| ITO, M. (1992). Simultaneous visualization of cortical barrels and horseradish peroxidase-injected layer 5b vibrissa neurones in the rat. Journal of Physiology 454, 247-265 | [Abstract] |
| ITO, M., KAWABATA, M. & SHOJI, R. (1981). Effect of prenatal X-irradiation on sensitivity of cortical neurons responding to vibrissa stimulation. Journal of Neurophysiology 46, 716-724 | [Medline] |
| ITO, M., KATO, M. & KAWABATA, M. (1998). Premature bifurcation of the apical dendritic trunk of vibrissa-responding pyramidal neurones of X-irradiated rat neocortex. Journal of Physiology 512, 543-553 | [Abstract/Full Text] |
| KATO, M., ITO, M. & SETO-OHSHIMA, A. (2000). Cortical somatosensory evoked potentials of seizure-sensitive and seizure-resistant gerbils. Epilepsy Research 40, 129-139 | [Medline] |
| KYRIAZI, H. T. & SIMONS, D. J. (1993). Thalamocortical response transformations in simulated whisker barrels. Journal of Neuroscience 13, 1601-1615 | [Abstract] |
| LARKMAN, A. & MASON, A. (1990). Correlations between morphology and electrophysiology of pyramidal neurons in slices of rat visual cortex. I. Establishment of cell classes. Journal of Neuroscience 10, 1407-1414 | [Abstract] |
| LIN, C.-S., LU, S. M. & SCHMECHEL, D. E. (1985). Glutamic acid decarboxylase immunoreactivity in layer IV of barrel cortex of rat and mouse. Journal of Neuroscience 5, 1934-1939 | [Abstract] |
| MCCORMICK, D. A., CONNORS, B. W., LIGHTHALL, J. W. & PRINCE, D. A. (1985). Comparative electrophysiology of pyramidal and sparsely spiny stellate neurons of the neocortex. Journal of Neurophysiology 54, 782-806 | [Medline] |
| MASON, A. & LARKMAN, A. (1990). Correlations between morphology and electrophysiology of pyramidal neurons in slices of rat visual cortex. II. Electrophysiology. Journal of Neuroscience 10, 1415-1428 | [Abstract] |
| NACHLAS, M. M., TSOU, K. C., DESOUZA, E., CHENG, C. S. & SELIGMAN, A. M. (1957). Cytochemical demonstration succinic dehydrogenase decarboxylase immunoreactivity in layer IV of barrel cortex of by the use of a new p-nitrophenyl substituted ditetrazole. Journal of Histochemistry and Cytochemistry 5, 420-436 | |
| NAKANISHI, K. & KUKITA, F. (2000). Intracellular [Cl-] modulates synchronous electrical activity in rat neocortical neurons in culture by way of GABAergic inputs. Brain Research 863, 192-204 | [Medline] |
| PINTO, D. J., BRUMBERG, J. C. & SIMONS, D. J. (2000). Circuit dynamics and coding strategies in rodent somatosensory cortex. Journal of Neurophysiology 83, 1158-1166 | [Abstract/Full Text] |
| SHIPLEY, M. T. (1974). Response characteristics of single units in the rat's trigeminal nuclei to vibrissa displacements. Journal of Neurophysiology 37, 73-90 | [Medline] |
| SIMONS, D. J. (1978). Response properties of vibrissa units in rat SI somatosensory neocortex. Journal of Neurophysiology 41, 798-820 | [Medline] |
| SIMONS, D. J. (1985). Temporal and spatial integration in the rat SI vibrissa cortex. Journal of Neurophysiology 54, 615-635 | [Medline] |
| SIMONS, D. J. & CARVELL, G. E. (1989). Thalamocortical response transformation in the rat vibrissa/barrel system. Journal of Neurophysiology 61, 311-330 | [Medline] |
| SIMONS, D. J. & LAND, P. W. (1987). Early experience of tactile stimulation influence organization of somatic sensory cortex. Nature 326, 694-697 | [Medline] |
| SIMONS, D. J. & WOOLSEY, T. A. (1984). Morphology of Golgi-Cox-impregnated barrel neurons in tat SmI cortex. Journal of Comparative Neurology 230, 119-132 | [Medline] |
| STEEL, R. G. D. & TORRIE, J. H. (1980). Principles and Procedures of Statistics, 2nd edn. McGraw-Hill Kogakusha, Tokyo, Japan | |
| TUCKETT, R. P., HORCH, K. W. & BURGESS, P. R. (1978). Response of cutaneous hair and field mechanoreceptors in cat to threshold stimuli. Journal of Neurophysiology 41, 138-149 | [Medline] |
| WAITE, P. M. E. (1973). Somatotopic organization of vibrissal responses in the ventrobasal complex of the rat thalamus. Journal of Physiology 28, 527-540 | |
| WELKER, C. (1976). Receptive fields of barrels in the somatosensory neocortex of the rat. Journal of Comparative Neurology 166, 173-189 | [Medline] |
| WELKER, C. & WOOLSEY, T. A. (1974). Structure of layer IV in the somatosensory neocortex of the rat: Description and comparison with the mouse. Journal of Comparative Neurology 158, 437-454 | [Medline] |
| WOOLSEY, T. A., DIERKER, M. I. & WANN, D. F. (1975). Mouse SmI cortex: qualitative and quantitative classification of Golgi-impregnated barrel neurons. Proceedings of the National Academy of Sciences of the USA 72, 2165-2169 | [Medline] |
| ZHU, J. J. & CONNORS, B. W. (1999). Intrinsic firing patterns and whisker-evoked synaptic responses of neurons in the rat barrel cortex. Journal of Neurophysiology 81, 1171-1183 | [Abstract/Full Text] |
| ZUCKER, E. & WELKER, W. I. (1969). Coding of somatic sensory input by vibrissae neurons in the rat's trigeminal ganglion. Brain Research 12, 138-156 | [Medline] |
This article has been cited by other articles:
|
|
 |
|
  C. P. J. de Kock, R. M. Bruno, H. Spors, and B. Sakmann Layer- and cell-type-specific suprathreshold stimulus representation in rat primary somatosensory cortex J. Physiol., May 15, 2007; 581(1): 139 - 154. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. Khatri and R. Ramos Rat Whisker Psychophysics J. Neurosci., November 29, 2006; 26(48): 12385 - 12386. [Full Text] [PDF]
|
|
|
|
 |
|
 L. M. Jones, D. A. Depireux, D. J. Simons, and A. Keller Robust Temporal Coding in the Trigeminal System Science, June 25, 2004; 304(5679): 1986 - 1989. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S.-H. Lee and D. J. Simons Angular Tuning and Velocity Sensitivity in Different Neuron Classes Within Layer 4 of Rat Barrel Cortex J Neurophysiol, January 1, 2004; 91(1): 223 - 229. [Abstract] [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
