JN Fuel your research with LabChart
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
 QUICK SEARCH:   [advanced]


     


J Neurophysiol 97: 2130-2147, 2007. First published November 22, 2006; doi:10.1152/jn.00338.2006 Free Article
0022-3077/07 $8.00
This Article
Free upon publication Free Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
97/3/2130    most recent
00338.2006v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via ISI Web of Science (1)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Quairiaux, C.
Right arrow Articles by Welker, E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Quairiaux, C.
Right arrow Articles by Welker, E.

Modified Sensory Processing in the Barrel Cortex of the Adult Mouse After Chronic Whisker Stimulation

Charles Quairiaux1, Michael Armstrong-James2 and Egbert Welker1

1Département de Biologie Cellulaire et de Morphologie, Université de Lausanne, Lausanne, Switzerland; and 2Department of Neuroscience, Queen Mary Westfield College, University of London, London, United Kingdom

Submitted 31 March 2006; accepted in final form 16 November 2006


    ABSTRACT
 TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 GRANTS
 ACKNOWLEDGMENTS
 REFERENCES
 
Chronic stimulation of a mystacial whisker follicle for 24 h induces structural and functional changes in layer IV of the corresponding barrel, with an insertion of new inhibitory synapses on spines and a depression of neuronal responses to the stimulated whisker. Under urethane anesthesia, we analyzed how sensory responses of single units are affected in layer IV and layers II & III of the stimulated barrel column as well as in adjacent columns. In the stimulated column, spatiotemporal characteristics of the activation evoked by the stimulated whisker are not altered, although spontaneous activity and response magnitude to the stimulated whisker are decreased. The sensitivity of neurons for the deflection of this whisker is not altered but the dynamic range of the response is reduced as tested by varying the amplitude and repetition rate of the deflection. Responses to deflection of nonstimulated whiskers remain unaltered with the exception of in-row whisker responses that are depressed in the column corresponding to the stimulated whisker. In adjacent nonstimulated columns, neuronal activity remains unaltered except for a diminished response of units in layer II/III to deflection of the stimulated whisker. From these results we propose that an increased inhibition within the stimulated barrel reduced the magnitude of its excitatory output and accordingly the flow of excitation toward layers II & III and the subsequent spread into adjacent columns. In addition, the period of uncorrelated activity between pathways from the stimulated and nonstimulated whiskers weakens synaptic inputs from in-row whiskers in the stimulated barrel column.


    INTRODUCTION
 TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 GRANTS
 ACKNOWLEDGMENTS
 REFERENCES
 
The whisker-to-barrel pathway of rodents has become a model system to investigate cortical plasticity. Mystacial whisker follicles map in a highly somatotopic manner onto segregated clusters of neurons termed barrels in layer IV of the somatosensory cortex, easily identifiable both structurally and functionally (Armstrong-James and Fox 1987Go; Simons 1978Go; Welker 1976Go; Woolsey and Van der Loos 1970Go). Neurons within a barrel primarily process the information from the anatomically related whisker, i.e., the principal whisker (PW), but possess receptive fields spanning several whiskers (Armstrong-James and Fox 1987Go; Welker et al. 1993Go). In both rats (Armstrong-James et al. 1994Go; Diamond et al. 1993Go; Glazewski et al. 1998Go; Wallace and Fox 1999bGo) and mice (Glazewski et al. 2000Go; Knott et al. 2002Go; Yang et al. 2002Go), those receptive fields were shown to exhibit activity-dependent plasticity during the entire life on various alterations in the pattern of whisker experience; however, the exact mechanisms of cortical plasticity are not well known.

It has been shown that a prolonged period of chronic whisker stimulation in a normally behaving adult mouse induces functional and morphological changes in the corresponding barrel (Knott et al. 2002Go; Rocamora et al. 1996Go; Welker et al. 1989Go, 1992Go). An electron microscopical analysis demonstrated that a 24-h period of chronic whisker stimulation causes a fourfold increase in the density of spineous GABAergic synapses in the stimulated barrel. Physiological recordings revealed that the firing of barrel neurons evoked by the stimulated whisker is depressed during the second epoch of the response, i.e., from 12 to 25 ms postwhisker deflection, whereas the earliest epoch from 3 to 12 ms is unaffected (Knott et al. 2002Go). Previous electrophysiological studies in rats showed that after an initial period of thalamocortical (TC) activation, sensory processing in layer IV is predominantly driven by cortico-cortical rather than by TC excitatory connections (Armstrong-James and Callahan 1991Go; Pinto et al. 2000Go; Simons and Carvell 1989Go). In analogy, the diminished firing during the second epoch of the response observed in mice should be attributable to experience-dependent alterations in barrel cortex (Welker et al. 1993Go). Taken together, these structural and functional observations suggest that a habituation evolves for the chronically stimulated whisker in the corresponding barrel and that this habituation could be mediated through the increase in the density of GABAergic synapses and consequently an enhanced evoked inhibition.

The purpose of the present work was to further characterize the functional consequences of the period of modified experience on the processing of sensory signals throughout the barrel cortex. In the barrel column corresponding to the stimulated whisker, is the depression of responses specific for the stimulated input or is the integration of responses to nonstimulated surround whiskers (SWs) also modified? Is the sensitivity of neurons to deflections of the stimulated whisker as well as their dynamic response properties altered? Is further intracortical processing of responses to the stimulated whisker modified, i.e., the transmission of the sensory signals from layer IV to layers II & III within the home barrel column as well as the transcolumnar propagation? To address these questions, we examined the effects of a 24-h period of chronic stimulation using current-source density (CSD) analyses and quantitative measurements of the firing of single units. Recordings at the end of a 24-h chronic whisker stimulation period were made in the barrel column corresponding to the stimulated whisker as well as in adjacent barrel columns, whose related principal whiskers had not been stimulated. Control recordings were made from nonstimulated mice, whereas long-term effects were analyzed in a group of mice recorded 4 days after the cessation of the stimulation.

Our results suggest that an experience-dependent mechanism modulates the magnitude of the responses generated by the direct TC afferents from the stimulated pathway (i.e., PW responses) and that this in turn leads to a weaker propagation of excitation within the corresponding column and through the transcolumnar pathways in layers II & III of the barrel cortex. Responses to nonstimulated whiskers were unaltered in their corresponding columns. In the column corresponding to the stimulated whisker, responses to SWs were not altered, with the exception of those evoked by surround in-row whiskers that were transiently depressed. We propose that following a model of Hebbian-like synaptic depression (Armstrong-James et al. 1994Go; Cruikshank and Weinberger 1996Go), the transcolumnar synaptic inputs from in-row SWs are weakened in the stimulated barrel column.


    METHODS
 TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 GRANTS
 ACKNOWLEDGMENTS
 REFERENCES
 
Recordings were made from a total of 160 adult female mice of the NOR strain derived from ICR stock (weight 25–35 g, between postnatal days 60 and 90; Van der Loos et al. 1986Go). All animal handling procedures used for this study were previously approved by the Office Veterinaire Cantonal (Lausanne, Switzerland) in accordance with Swiss Federal Laws.

Chronic whisker stimulation

Sixty-seven mice were submitted to a 24-h period of chronic stimulation of their left C2 whisker immediately before electrophysiological experiments (24h STIM). In an additional group, 18 mice were chronically stimulated for the same period of time but were left to recover 4 days in their home cage before the recording sessions (4-days-after-STIM). Seventy-five nonstimulated mice were used as control (NO STIM). Before the chronic stimulation procedure, mice were anesthetized with Nembutal [sodium pentobarbital, 60 mg/kg, administered intraperitoneally (ip)]; a small ferrous metal rod (ion/nickel, length 1.5 mm, diameter 0.2 mm, 0.6 mg) was then glued onto the C2 whisker of their left whiskerpad 3 mm away from the skin surface. After recovery from the anesthesia, animals were placed for a 24-h period in the Lausanne whisker stimulator (Melzer et al. 1985Go) in which they had access to food and water and behaved freely while being exposed to magnetic field bursts (repetition rate 9/s; burst duration 40 ms; intensity 7 x 103 A/m, rms value; frequency during burst 50 Hz, from the mains). At the end of the stimulation period, the animal was removed from the stimulator and the metal rod was detached from the whisker with acetone under visual control by means of a dissection microscope.

Anesthesia and surgery

Recordings were carried out under urethane anesthesia (10% solution in distilled water; 2 mg/g body weight, ip). The mice were mounted in a stereotaxic frame equipped with a homemade headholder (Kucera 1970Go), providing a continuous flow of oxygen in front of their nose, and received a subcutaneous dose of lidocaine (0.1 ml; 1% wt/vol) above the parietal bone. Body temperature was maintained at 37°C by a rectal thermistor-controlled heating pad (Homeothermic Blanket System, Harvard Apparatus, Eden, UK). A craniotomy was made over the right parietal cortex with a high-speed dental drill to expose a large part of the somatosensory cortex including the posteromedial barrel subfield region (PMBSF; Woolsey and Van der Loos 1970Go). The exposed region was covered with 1% agar dissolved in 0.9% saline. The depth of the anesthesia was continuously controlled using several criteria: absence of whisking or eye blink reflex and burst rate of layer V cortical neurons between two and four/s (Armstrong-James et al. 1985Go; Fox and Armstrong-James 1986Go). Supplementary urethane doses (10% of the original dose) were given when necessary.

Electrophysiological recordings

Extracellular neuronal activity was recorded using carbon-fiber microelectrodes of low impedance (1–5 M{Omega}), the tip of which had been sharpened by spark-etching (Armstrong-James and Millar 1979Go). The microelectrode was lowered in the cortex normal to the pial surface using a mechanical microdrive (Model S-11, 10-µm precision; Narashige Instruments, Tokyo, Japan) and a reference electrode was placed on the scalp. The differential voltage signal was recorded using a Neurolog modules arrangement (Neurolog System, Digitimer, Welwyn Garden City, UK). This signal was first amplified (2K) then filtered for extracellular spike activity and local field potentials (LFPs) through parallel band-pass filters of 0.8–5 kHz and 0–45 Hz, respectively. Single units were isolated by means of a waveform time window discriminator that provided digital output of the discriminated spike (DIS-1; BAK Electronics, Mt. Airy, MD). Digitized spike times and analog LFP waveforms were processed using the CED-Power 1401 interface [Cambridge Electronic Design (CED), Cambridge, UK].

SINGLE-UNIT ANALYSES. Neurons were recorded during the cortical active periods of slow-wave sleep. Spike times were recorded during a time period of 1,000 ms starting 250 ms before whisker deflection. They were collected into 1-ms bins in two separate channels, retrieving either all spike times to build peristimulus time histograms (PSTHs) or the time of only the first spike after the onset of the whisker deflection to build latency histograms (LHs). The response magnitude (RM) of a single unit was defined as the mean number of spikes evoked per deflection, corrected for the spontaneous activity. It was calculated using the PSTH data by counting the number of spikes recorded during the 3- to 100-ms interval after the onset of whisker deflection, from which we subtracted the number of spikes that occurred during a similar time interval before the whisker deflection. All neurons recorded during this study had significant responses to the deflection of the PW (RM >0.1). Epoch analyses of responses were carried out by dividing PSTHs in five consecutive poststimulus epochs that were chosen based on the temporal profile of the PSTHs of layer IV neuronal response to PW deflection (see RESULTS for determination of these epochs). Based on the LH, we calculated the mode (modal latencies; Armstrong-James and Fox 1987Go), the median, and the interquartile (IQ) range q75–q25 of the distribution of the spike latencies (median latencies and IQ range; Welker et al. 1993Go). Analyses were carried out using homemade scripts written in Spike2 software language (CED).

RECORDING SITES AND HISTOLOGY. Per animal, one to three electrode penetrations were made in barrel columns C2, C1, and/or D2. The penetrations were tangentially separated by a minimal distance of 100 µm. During a penetration, three to eight neurons were recorded separated from each other by a minimal distance of 50 µm along the depth. Radial position within a penetration was monitored from the micrometer setting of the stereotaxic manipulator and used to allocate units to cortical layers as previously determined for mice of the NOR strain (Welker et al. 1993Go): layers II & III = 150–350 µm from the pial surface, layer IV = 350–480 µm. At the end of the recording session, electrolytic microlesions (diameter 25 µm) were made in layer IV of each of the penetrations achieved. For this, the microelectrode was lowered at 400 µm from the pial surface and a 1.6-µA negative current was passed through the electrode for 7 s.

Animals then received an ip injection of 0.1 ml sodium pentobarbital (5 mg) and were transcardially perfused with 10% formalin in 0.9% NaCl. The brains were removed from the skull and postfixed for 1 to 3 wk. After 24 h of cryoprotection in 30% sucrose, the right hemispheres were oriented using a guillotine (Rice and Anders 1977Go) and sectioned tangentially to the pial surface overlying the barrel cortex on a cryotome (section thickness 40 µm). Sections were then stained for Nissl substance with cresyl-violet to visualize the position of the microlesions with respect to the barrel outlines.

LOCAL FIELD POTENTIAL AND CSD ANALYSES. To calculate CSDs, PW-evoked LFP waveforms were recorded in the C2 barrel column with one microelectrode successively placed at incremental intervals ({Delta}z = 50 µm) from the pial surface to 1,000 µm beneath, encompassing the entire thickness of the cortex. At each depth, the LFP waveform was averaged for a minimum of 50 whisker deflections processed with the standard protocol (1.43° upward, 0.5/s; see Whisker deflection). LFP waveforms were used to calculate the second spatial derivative of potentials along the depth, referred to as the one-dimensional CSD, and yields spatiotemporal profiles of extracellular current sinks and sources. Those CSD traces were calculated in one dimension (z), with the assumption that currents in the x and y directions were negligible and that intracortical conductivity was constant, using the following formula: CSD = –[V(z + {Delta}z, t) – 2V(z, t) + V(z{Delta}z, t)]/{Delta}z2, where V(z, t) is the measured voltage at a subpial depth z (t = time; {Delta}z = 50 µm) (Mitzdorf 1985Go; Nicholson and Freeman 1975Go). CSD calculations were carried out using homemade scripts written in Matlab software (The MathWorks, Natick, MA).

Whisker deflection

The whisker deflection consisted of an upward movement of 3-ms duration. To apply this deflection, the whisker (trimmed to a length {approx} 15 mm) was inserted into a thin (inner diameter 0.5 mm) borosilicate-glass tube attached to a piezoelectric bimorph slab. The upward movement of the probe was evoked by square-wave voltage pulses delivered by a stimulator with high-voltage output (stimulator D59A; Digitimer) gated by the CED–Power 1401 interface. Field potentials were carefully observed to ensure that deflections were applied only during appropriate stages of slow-wave sleep (Armstrong-James and Fox 1988Go). The following deflection amplitudes and repetition rates were used:

Standard protocol, which corresponds to the one used in previous studies in our laboratories (Armstrong-James and Fox 1987Go; Welker et al. 1993Go). The amplitude of the upward deflection was set to engage an angular displacement of the whisker of 1.43° from the horizontal plane. Evoked responses of barrel column neurons were all recorded toward their PW and six to eight immediate surround whiskers, one at a time. Each individual whisker was deflected 50 times, at a repetition rate of 0.5/s.

Varying amplitude protocol, applied to record the responses of layer IV neurons on deflections of different amplitudes (0.06, 0.18, 0.30, 0.68, 1.02, 1.43, and 2.1°) at a repetition rate of 0.5/s. Note that using our stimulator, both the amplitude and the velocity of the deflection were concomitantly decreased with decreasing voltage pulses, as illustrated in Armstrong-James and Fox (1987)Go. Here we did not measure the actual variations of the velocity.

Varying repetition rate protocol, in which the PW was deflected at various rates (0.5, 1, 2, 4, 8, and 16/s) at an angular displacement of 1.43° upward. Trains of 10 deflections were used for each repetition rate and applied with a 2-s deflection-free interval. Five runs of pseudorandomly interleaved trains were performed, resulting in a total of 50 deflections for each repetition rate.

Statistics

Single-unit data were grouped by cortical layers (II/III and IV), barrel columns (C2, C1, and D2), and experimental groups (NO STIM, 24h STIM, 4-days-after-STIM). Within each class and for each whisker deflected, distributions of the different response variables (RM, modal and median latencies, epoch's activity) were tested for normality (Shapiro–Wilk test). Normality was rejected in half of the cases; data were therefore normalized with a transform-ranking method before entering a multivariate hierarchical model [general linear model (GLM) procedure]. For each response variable, the hypothesis of no overall group effect on the cluster of dependent variables was tested with a MANOVA (applying Wilk's lambda test criteria). Where the null hypothesis was rejected, multiple comparisons of each variable between groups could then be performed with minimized risks of false positive (Type II error) using Tukey's Studentized range HSD (honestly significant difference) tests, which took the interanimal variability as a potential error source. For comparisons of the epoch's activity between groups, a procedure of repeated-measures ANOVA with contrast variables was used. For the series of recordings in which the PW was deflected at different amplitudes or rates, mean values of response variables were compared within groups using a MANOVA and Tukey's tests. We applied a GLM model of ANOVA with covariable to test whether the cortical responses in the two groups of mice (24h STIM and NO STIM) were modified in a similar manner by altering the deflection variables. All statistics were carried out using the SAS package for windows (SAS Institute, Cary, NC).


    RESULTS
 TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 GRANTS
 ACKNOWLEDGMENTS
 REFERENCES
 
Spatiotemporal distribution of evoked current flows

The sequential processing of sensory signals within a cortical column can be revealed by measuring the laminar LFPs evoked by whisker deflection and processing them through a one-dimensional CSD analysis that reveals the spatiotemporal pattern of synaptic activation (Mitzdorf 1985Go; Nicholson and Freeman 1975Go). We recorded the LFPs evoked by deflection of the C2 whisker according to the standard protocol (see METHODS) in the C2 barrel column of animals that were either nonstimulated (NO STIM; n = 7) or of those that had been stimulated 24 h for their C2 whisker (24h STIM; n = 7). Figure 1A shows a representative example of the laminar traces of the evoked LFPs in a nonstimulated mouse. The variations in LFP profiles across laminae indicate a heterogeneous pattern of current flows. During the first 40–50 ms of the cortical response, a biphasic extracellular field potential was evoked from 200- to 800-µm subpial depth that consisted of a short-latency negative-potential phase (5- to 7-ms onset in both groups) immediately followed by a positive phase. In all cases, the magnitude of voltage gradients was largest between lower layer III and upper layer V, with negative and positive peak amplitudes invariably present in layer IV. The initial negative phase terminated, respectively, 25.0 ± 4.1 and 24.2 ± 3.4 ms poststimulus in layer IV of NO STIM and 24h STIM groups (mean ± SD). In the most superficial (0–100 µm) and deepest lamina (850–1,000 µm), field potentials reversed polarity, with positive–negative phases variable in duration across mice. During the next 50 ms, a biphasic positive–negative potential sequence of small magnitude was observed in both groups that was spatially restricted between lower layer III and upper layer V. Subsequently, the average field potentials gradually returned to baseline level. The total duration of the evoked LFP response was 90–95 ms.


Figure 1
View larger version (26K):
[in this window]
[in a new window]

 
FIG. 1. A: average traces of local field potentials (LFPs, left) evoked by principal-whisker (PW) deflection along the depth of barrel column C2 in a nonstimulated mouse, together with the corresponding current-source density (CSD) traces (right). Red LFP trace illustrates the typical multiphasic LFP evoked in layer IV, with an early negative phase followed by a positive phase, then a positive/negative phase sequence of smaller magnitude. B: CSD contour plots linearly interpolated from the laminar CSD profiles evoked by the PW in the barrel column C2 from a nonstimulated (left) and a stimulated (right) mouse. Color coding shows current sources in blue, current sinks in red, and isopotentials in green.

 
The CSD analysis allows the localization of the current generators (Mitzdorf 1985Go; Nicholson and Freeman 1975Go). The CSD traces computed from the laminar LFP traces (Fig. 1A) point to a short-latency current sink (positive CSD phase) present in layers II & III to layer V by the PW deflection that was followed by a current source (negative CSD phase). These current events are more easily identifiable in Fig. 1B that shows two examples of color-coded contour plots, interpolated from the laminar CSD traces recorded in a nonstimulated (left) and in a stimulated mouse (right). These plots illustrate variations of the current sinks (warm colors) and sources (cold colors) as a function of cortical depth and poststimulus time. A short-latency current sink was evoked in layer IV and upper layer V at 5–7 ms poststimulus onset (20-ms duration) and it invaded layers II & III within the subsequent 1–3 ms. This early sink, which peaked in middle layer IV (400- to 450-µm depth), was balanced by a large current source located in layer VI and a smaller one in the superficial layers; the latter was more visible in CSD traces than in contour plots. A current source lasting from 25 to 50 ms poststimulus followed the early sink in layer IV and layers II & III. By 5–6 ms after the onset of the early layer IV sink, a current sink progressively invaded infragranular layers and peaked 34–45 ms poststimulus in middle layer VI, i.e., concomitant with the peak of the layer IV source. No differences were observed between 24h STIM and NO STIM animals in terms of either laminar location or extent, nor timing of current sinks and sources. Also the peak-to-peak magnitude of the initial waveform generated in layer IV was the same in both groups of mice: in NO STIM the mean value was 4.4 mV (SD = ±2.3) and in the 24h STIM group this was 4.3 mV (±1.7).

Modifications in single-unit activity in the "stimulated barrel column"

We quantitatively analyzed the effects of a 24-h period of chronic C2 whisker stimulation on the response properties of individual neurons in the C2 barrel column, as well as in neighboring barrel columns C1 and D2. In stimulated animals, the C2 barrel column will be referred to hereafter as the "stimulated barrel column" in contradistinction to neighboring "nonstimulated barrel columns." In all, 1,042 neurons were recorded in 68 animals from the NO STIM (469 neurons), 60 from the 24h STIM (n = 438), and 18 from the 4-days-after-STIM (n = 135) groups. Data from 224 of these neurons were already partially presented in a previous study (Knott et al. 2002Go).

TRANSIENT DECREASE IN SPONTANEOUS ACTIVITY. In the C2 barrel column of nonstimulated mice, mean spontaneous activity was 1.7 ± 0.8 spikes/s for neurons in layer IV (mean ± SD; n = 102; Fig. 2) and 1.3 ± 0.8 spikes/s in layers II & III (n = 73). After 24 h of whisker stimulation, mean spontaneous firing of neurons was significantly reduced by 47% in layer IV (n = 77) and 40% in layers II & III (n = 70) of the stimulated barrel column compared with values in nonstimulated animals (P < 0.01, Kruskal–Wallis). In mice recorded 4 days after the chronic stimulation, spontaneous activity of neurons (n = 82 and 53 in layer IV and layers II & III, respectively) was similar to values in nonstimulated mice. This transient decrease in spontaneous activity was restricted to barrel column C2 because in barrel columns C1 and D2 it was identical in mice of the 24h STIM group and of the NO STIM group.


Figure 2
View larger version (14K):
[in this window]
[in a new window]

 
FIG. 2. Mean spontaneous activity of neurons in layer IV and layers II & III of the C2, C1, and D2 barrel columns from the 3 groups of mice studied. Numbers in the bars indicate the number of neurons recorded. Asterisks and brackets indicate significant differences between groups at P < 0.01 (Kruskal–Wallis). After 24 h of C2 whisker stimulation, spontaneous activity was significantly lowered in the corresponding C2 barrel column, whereas it remained unaltered in neighboring barrel columns C1 and D2. Spontaneous activity returned to control levels 4 days after the chronic stimulation was halted.

 
DECREASE IN PW RESPONSES. Figure 3A shows the distributions of the latency of PW responses as a function of cortical depth for all three groups of mice analyzed. Responses of single units in the three groups were evoked using the standard protocol of whisker deflection. In a comparison of the three distributions, it does not seem that chronic whisker stimulation affects the latency of the responses in any of the cortical layers. This was confirmed (Fig. 3B) by the statistical analysis of the median latency of the responses of neurons of layer IV (NO STIM, n = 102, latency = 11.4 ± 3.8 ms; 24h STIM, n = 77, latency = 10.9 ± 3.4 ms; 4-days-after-STIM, n = 82, latency = 10.7 ± 2.7 ms) and layers II & III (NO STIM, n = 73, latency = 14.7 ± 5.8 ms; 24h STIM, n = 70, latency = 14.8 ± 8.8 ms; 4-days-after-STIM, n = 53, latency = 13.9 ± 3.4 ms).


Figure 3
View larger version (15K):
[in this window]
[in a new window]

 
FIG. 3. PW-C2 response latencies in the C2 barrel column. A: raster plots showing the distribution of the PW-response modal latencies of single units recorded in barrel column C2 as a function of cortical depth. Dashed horizontal lines represent the borders between cortical layers. B: box plots of median latencies computed for neurons in layer IV and layers II & III. Interquartile ranges as well as 10–90% percentiles are presented for each group. Chronic stimulation did not alter the median latency of responses.

 
Analysis of the response magnitude (RM) of these sets of neurons showed striking effects of the chronic whisker stimulation. Figure 4A shows the frequency distributions of response magnitude of neurons in barrel column C2 to deflections of the C2 whisker. In layer IV of NO STIM mice the distribution was centered at 0.8–1.0 spike per deflection; <25% of neurons showed responses <0.6 spike per deflection. The RM distribution for the 24h STIM group became clearly asymmetrical, with 50% of the neurons showing responses <0.6 spike/deflection. This distribution is significantly different from that of the NO STIM group (P < 0.01; chi-square). The RM distribution of the 4-days-after-STIM group was significantly different from the NO STIM and 24h STIM groups. Compared with the distribution in the NO STIM group, the distribution in the 4-days-after-STIM group shows an increase in highly responsive neurons (>1.4 spikes/deflection). Figure 4B displays the mean RM of the neurons in layer IV in the three groups of mice and indicates the significant depression of RM of neurons in the C2 barrel after 24 h of chronic whisker stimulation (mean ± SD = 0.97 ± 0.53 in NO STIM vs. 0.71 ± 0.40 in 24h STIM; P < 0.01, MANOVA and Tukey's HSD). However, the RM increased significantly (15%) at 4 days after stimulation compared with NO STIM (4-days-after-STIM: 1.14 ± 0.52; P = 0.03). A similar analysis was performed on the RM of neurons in layers II & III. Figure 4A shows the frequency distribution of the RM of these neurons in the three groups of mice studied. After 24 h of whisker stimulation the RM distribution was significantly shifted to smaller values compared with the NO STIM (P < 0.01). Four days after the stimulation, the distribution was no longer significantly different compared with NO STIM. Statistical analysis of the mean RM values showed significant depression of RM of neurons in layers II & III of the 24h STIM group (mean ± SD = 0.74 ± 0.36 in NO STIM; 0.47 ± 0.36 in 24h STIM; P < 0.01). After 4 days of recovery, the mean RM of neurons was not significantly different from neurons in nonstimulated mice (4-days-after-STIM: 0.67 ± 0.32).


Figure 4
View larger version (23K):
[in this window]
[in a new window]

 
FIG. 4. Altered response magnitude (RM) to the stimulated whisker in the C2 barrel column. A: distributions of PW-evoked RM for neurons in layer IV and layers II & III of the C2 barrel column (increment of classes: 0.2 spike/stim; n = numbers of neurons analyzed). Asterisks indicate significant differences (chi-square; P < 0.01) between RM distributions of neurons in the experimental (24h STIM, 4-days-after-STIM) groups and controls (NO STIM). B: mean RM ± SE for neurons in layer IV and layers II & III of barrel columns C2, C1, and D2 to the deflection of their PW. Chronic stimulation decreased RM in both layer IV and layers II & III of the C2 barrel column [general linear model (GLM) procedure; Tukey's HSD; P < 0.01] but did not modify PW responses in neighboring barrel columns (C1 and D2). Four days after the cessation of chronic stimulation, depression of RM in the C2 barrel column was no longer present; on the contrary, mean RM to C2 whisker deflection was significantly higher for layer IV neurons of the 4-days-after-STIM mice compared with those of nonstimulated mice (P < 0.05).

 
Thus a 24-h period of chronic stimulation induced a decrease in RM toward the deflection of the stimulated whisker both in layer IV and in layers II & III. This depression of PW-evoked responses was transient because the mean RM of neurons recorded 4 days after the stimulation period was no longer smaller than that in nonstimulated animals. The data of the 4-days-after-STIM group show that in layers II & III, the RM returns to control level, whereas in layer IV we detected a small but significant increase in RM compared with the value of the NO STIM group.

Furthermore, this part of the analysis was complemented with a comparison of the PW response between neurons of layers II & III and layer IV. In the 24h STIM group the mean RM of neurons in layers II & III was decreased by 36% relative to the control group, whereas in layer IV the decrease was 26%. To test this difference for statistical significance, we calculated for each neuron of the 24h STIM group an index representing the ratio between its RM and the mean RM of neurons in the same layer of the NO STIM group [RM24h STIM(RM24h STIM + mean RMNO STIM)]. This comparison revealed that the mean value of this index for neurons in layers II & III was significantly smaller than that for neurons in layer IV, indicating that the effect of stimulation on RM depression was stronger in layers II & III than that in layer IV (mean ± SD: 0.40 ± 0.12 in layer IV and 0.32 ± 0.12 in layers II & III).

QUANTITATIVE ANALYSIS OF PW RESPONSES IN NONSTIMULATED BARREL COLUMNS. To test whether the effect of the C2 stimulation is restricted to PW responses inside the C2 barrel column, we also analyzed neuronal responses in neighboring barrel columns C1 and D2 for deflection of the corresponding whisker using the standard protocol (C1 column: layer IV NO STIM, n = 54; 24h STIM, n = 49; layers II & III NO STIM, n = 53; 24h STIM, n = 50; D2 column: layer IV NO STIM, n = 65; 24h STIM, n = 58; layers II & III NO STIM, n = 48; 24h STIM, n = 68). No statistically significant differences were revealed at P < 0.05 in mean RM between stimulated and nonstimulated animals in either layer IV or layers II & III (Fig. 4B), indicating that the decrease in PW-evoked response was restricted to the barrel column corresponding to the stimulated whisker. C2 whisker stimulation does not affect PW responses in neighboring barrel columns.

EPOCH ANALYSIS OF PW RESPONSES. To determine whether the decrease in the magnitude of PW responses in the stimulated column reflected a general suppression of evoked firing or affected only parts of the response, we performed a quantitative poststimulus epoch analysis of the responses to C2 deflection according to the standard protocol. This analysis included the responses of all neurons in layer IV and layers II & III of the C2 barrel column in the three groups of mice (layer IV: NO STIM, n = 102; 24h STIM, n = 77; 4-days-after-STIM, n = 82; layers II & III: NO STIM, n = 73; 24h STIM, n = 70; 4-days-after-STIM, n = 53; same neurons as in Fig. 4).

Layer IV.    The temporal profile of the mean population PSTHs in layer IV of nonstimulated animals was used to determine five successive epochs (Fig. 5A). Neurons fired action potentials (APs) in response to whisker deflection from 6 ms poststimulus, reached a peak in spiking probability at 11 ms (poststimulus epoch T1: 3–11 ms), then gradually decreased to near-spontaneous levels about 25 ms poststimulus (T2: 12–25 ms). During the next 25 ms, probabilities did not exceed the spontaneous level (T3: 26–50 ms). This profile of evoked spiking precisely matches the timing of the current sink-source sequence in layer IV: a period of increased firing in layer IV temporally matched the early sink, whereas firing is almost absent during the following source. A rebound of activity took place during the fourth epoch (T4: 51–100 ms), after which spiking probabilities returned to the spontaneous level.


Figure 5
View larger version (36K):
[in this window]
[in a new window]

 
FIG. 5. Epoch analyses. A: population PSTHs at 1-ms binning for neurons responding to PW deflection in layer IV and layers II & III of the C2 barrel column. Epochs are labeled above the first histogram and indicated using graded colors. B: results of epoch analyses expressed as mean ± SE population spike rates per millisecond for the 5 epochs. Asterisks and brackets indicate significant differences between groups at P < 0.05 (GLM procedure; ANOVA for repeated measures). After chronic stimulation, neuronal responses were significantly decreased during T2 and T4 in layer IV, and during T2, T3, and T4 in layers II & III, whereas spike rates remained unaltered during the earliest epoch. Note that after a stimulus-free period of 4 days the response of layer IV neurons was significantly higher during the 4th epoch compared with the NO STIM group.

 
The quantitative epoch analysis (Fig. 5B) showed that PW response is not affected by chronic stimulation during T1, whereas it is significantly smaller during T2 (P < 0.01, ANOVA with repeated measures). During T3 neuronal activity remains at the spontaneous level in all groups. The characteristic rebound in neuronal activity evoked by deflection of the PW during T4 was strongly reduced in 24h STIM mice relative to NO STIM mice (P < 0.01). Four days after the stimulation period, however, evoked spiking was significantly enhanced (P = 0.02) during this late poststimulus epoch compared with the activity of neurons of the NO STIM animals.

Layer II/III.    The same epochs were used for analysis of the responses of cells in layers II & III. Evoked activity began at 6 ms poststimulus (Fig. 5A). The peak of spiking probability was reached 2–3 ms later than in layer IV, at 13–14 ms poststimulus. Spiking probability then gradually decreased to reach a plateau of low probability of spiking above the spontaneous level around 25 ms poststimulus, until 100 ms poststimulus. There is no difference in duration and temporal profile of the evoked population spiking in the three groups.

As illustrated in Fig. 5B, chronic stimulation did not alter mean spike rate during the first epoch (P = 0.7) but significantly decreased evoked firing during the following time epochs T2, T3, and T4 (P < 0.01). After the initial epoch T1, during which layer IV neurons reached their maximal firing, layers II & III firing strongly increased until a peak at 13–14 ms in mice of the NO STIM group, whereas in the 24h STIM group mean spiking probability did not exceed the level attained at 11 ms poststimulus and decreased rapidly after 13 ms poststimulus. Accordingly, the mean spiking rate was higher during T2 compared with that during T1 in the NO STIM group (P < 0.01), whereas it was not in 24h STIM animals. Mean spike rates per epoch were not significantly different in the 4-days-after-STIM and the NO STIM groups.

DECREASED DYNAMIC RANGE FOR DEFLECTION-AMPLITUDE–DEPENDENT RESPONSES IN THE STIMULATED BARREL. To test whether the decrease in response magnitude of layer IV neurons is paralleled by a change in sensitivity of cortical neurons to whisker deflection, we applied the varying amplitude deflection protocol (see METHODS) and quantified the responses of neurons in layer IV of four mice of the 24h STIM group (n = 31) and five mice of the NO STIM group (n = 35). As illustrated in the example of Fig. 6A, LFPs and single-unit spikes exhibited similar input–output relationships: the magnitude of field potentials and of neuronal firing increased with incremented deflection amplitude. Concomitantly, both the onset of evoked LFP waveforms and the mode of single-unit spiking distributions were delayed with decremented amplitude. The GLM procedure of ANOVA with covariable showed that mean modal latency decreased with deflection amplitude (Fig. 6B), whereas the mean RM similarly increased with deflection amplitude (Fig. 6C) in both the 24h STIM and the NO STIM groups at P < 0.05.


Figure 6
View larger version (10K):
[in this window]
[in a new window]

 
FIG. 6. Neuronal response in layer IV of the C2 barrel as a function of whisker deflection amplitude. A: examples of LFPs and PSTHs in layer IV of a nonstimulated mouse using 3 different amplitudes of PW deflection. Both parameters show an increase in magnitude with incremented deflection amplitude. B: mean modal latencies (±SD) of single unit's responses on incremental deflection amplitudes in nonstimulated and stimulated animals. C: mean RM (± SE) of single-unit responses increased with deflection amplitude. Asterisks indicate differences between RM of nonstimulated and stimulated animals at P < 0.01 (GLM procedure; MANOVA and Tukey's HSD). Amplitude of whisker deflection at 1 to 7: 0.06, 0.18, 0.30, 0.68, 1.02, 1.43, and 2.10°, respectively.

 
All neurons in both groups of mice showed a strong correlation between deflection amplitude and RM (P < 0.0001, Spearman correlation test). The maximal values of RM were reached at a deflection amplitude of 1.43° in both groups of mice. The lowest deflection amplitude assessed still evoked significant responses above spontaneous activity in two thirds of the neurons recorded in both groups (24/35 in NO STIM, 25/31 in 24h STIM animals), whereas all cells were significantly responsive after two increments in deflection amplitude. Mean RM was smaller in stimulated animals than that in nonstimulated animals at all deflection amplitudes. These differences reach a significant level for responses to PW deflections >0.68° (P < 0.05; MANOVA and Tukey's test). Thus although chronic stimulation did not alter the sensitivity of layer IV neurons to whisker deflections of small amplitude, it reduced the dynamic range of neuronal response for deflection amplitude.

ADAPTATION TO INCREASED REPETITION RATE OF WHISKER DEFLECTION. We further explored the effect of chronic whisker stimulation on response properties of cortical neurons by varying the repetition rate of the whisker deflection. The following repetition rates were used: 0.5, 1, 2, 4, 8, and 16 deflections/s while recording neurons in the C2 barrel column from five nonstimulated (layer IV, n = 19; layers II & III, n = 20) and four stimulated mice (layer IV, n = 16; layers II & III, n = 18). At 0.5 deflection/s neuronal firing does not vary during a train of deflections, whereas at higher repetition rates the response declines rapidly at subsequent deflections, reaching a steady level of firing after the third or fourth deflection (Fig. 7A). The number of deflections at which this steady level is reached is not different between control and 24h STIM.


Figure 7
View larger version (24K):
[in this window]
[in a new window]

 
FIG. 7. Neuronal response in layer IV (left) and layers II & III (right) of the C2 barrel column as a function of deflection repetition rate. A: mean RM (±SE) for each stimulus within trains of 10 deflections applied at rates of 0.5 and 8/s. Responses to subsequent deflections were kept separately and mean values calculated from 5 runs. B: mean RM (± SE) for trains of PW deflections applied at repetition rates of 0.5, 4, and 8 deflections/s in nonstimulated (NO STIM) vs. stimulated (24h STIM) mice. Here, mean values were calculated by averaging the number of spikes in 3- to 100-ms poststimulus windows for all deflections within a train. P values calculated for the differences in mean RM between groups are indicated above each repetition rate of deflection (GLM procedure; MANOVA and Tukey's HSD). Note that the response range for neurons in the stimulated barrel columns is drastically reduced compared with NO STIM.

 
For each single unit we determined the RM as the average evoked firing per stimulus during five trains of 10 deflections, taking into account the response to all deflections within a train. Figure 7B gives the mean values of RM per repetition rate tested for the two groups of mice per layer analyzed. The RM of all neurons in layer IV and layers II & III decreased with incremental repetition rate between 0.5 and 8 deflections/s in both NO STIM and 24h STIM groups (P < 0.0001, Spearman correlation test).

In layer IV, chronic stimulation significantly reduced mean RM of the neuronal population (P < 0.05; MANOVA). The GLM procedure of ANOVA with covariable confirmed this effect of chronic whisker stimulation and showed that the mean RM is dramatically affected by repetition rate in a similar manner in both stimulated and control mice. In both groups of mice, neurons adapt to incremental repetition rate in the same manner, graphically expressed by the two, almost parallel lines in Fig. 7B. In this respect for both groups, we observed a similar twofold reduction of RM for responses recorded at 8 deflections/s compared with those at 0.5/s.

In layers II & III, RM was depressed after chronic stimulation for responses at 0.5 and 4/s at P < 0.05 (MANOVA and Tukey's test); however, the responses to 8 deflections/s were not significantly different in both groups (P = 0.2709). Between 0.5 and 8 deflections/s, we observed a 4.4-fold reduction of RM in nonstimulated animals and a threefold reduction in stimulated animals. This difference in the effect of the repetition rate between the NO STIM and the 24h STIM groups was significant, as revealed by the GLM procedure of ANOVA with covariable (P = 0.0025). In agreement, the two regression lines between repetition rate and RM (Fig. 7C) for the neurons in this layer are not in parallel and the mean slope is significantly smaller in 24h STIM compared with that in the NO STIM group (mean by,x ± SD: 0.27 ± 0.1 vs. 0.12 ± 0.7. P = 0.0004, two-tailed t-test, using unitary values of increment for the different repetition rates). This may be explained by the fact that the first deflection within a train evokes a very small response of layers II & III neurons in chronically stimulated animals.

Altered surround receptive field in the "stimulated barrel column"

For all neurons recorded in layer IV and layers II & III of the C2 barrel column, the PW deflection invariably evoked a stronger response than did any SW (number of neurons analyzed given in third paragraph of RESULTS). Figure 8 presents the mean RM of those neurons on the deflection of the PW and all SWs using the standard deflection protocol. In layer IV and layers II & III, <10% of neurons responded to the PW only, the majority responding to five SWs. In layer IV, 5% of neurons responded to all eight SWs tested, whereas in layers II & III this characterizes 10% of the neurons. The number of SWs eliciting significant responses per neuron was not altered by chronic stimulation, suggesting that the size of the surround receptive field was not diminished. Also the median latencies to SW deflection were not altered by the chronic stimulation.


Figure 8
View larger version (25K):
[in this window]
[in a new window]

 
FIG. 8. Modifications in principal vs. surround whisker (SW) RM in barrel column C2. A: mean RM (±SE) for layer IV and layers II & III neurons on PW and SW deflections. Responses to SWs that did not belong to row C were pooled (see left diagram): Arc-1 (B1 + D1), Arc-2 (B2 + D2), and Arc-3 (B3 + D3). Asterisks indicate values that differ significantly between groups at P < 0.05 (GLM procedure; MANOVA and Tukey's HSD). After chronic stimulation of whisker C2, the responses to C1 deflection were decreased in layer IV and layers II & III. Responses to deflection of other SWs were not significantly altered by the chronic stimulation. B: mean population PSTHs (2-ms bins) toward C1 whisker deflections. Asterisks indicate epochs during which mean firing rate is significantly lower in the 24h STIM group compared with the NO STIM group at P < 0.01 (Kruskal–Wallis).

 
In agreement with a previous study in the same mouse strain (Welker et al. 1993Go), neuronal responses vary as a function of the position of the deflected SW on the whiskerpad relative to the PW; usually the SW positioned within the same row as the PW elicited the strongest and fastest response. In layer IV and layers II & III of barrel column C2, the deflection of in-row whiskers C1 and C3 evoked faster responses than the deflection of any other SWs (P < 0.05; Kruskal–Wallis). Whisker C1 evoked the strongest response followed by whisker C3 (P < 0.05). With respect to responses evoked by the deflection of whiskers in rows B and D, the caudal whiskers (Arc-1) evoked stronger responses than the rostral whiskers (Arc-3); however, these differences did not reach significant levels. Because of the anisotropy between responses to SWs, these were grouped as follows: responses to caudal whisker C1 and rostral whisker C3 were kept separated, whereas responses to whiskers from Arc-1 (B1, D1) were pooled, along with responses to whiskers from Arc-2 (B2, D2) and Arc-3 (B3, C3).

Multivariate analyses revealed a significant effect of chronic stimulation on RM to SW deflection in both layer IV and layers II & III at P < 0.05 (MANOVA). For layer IV neurons, multiple comparisons (Tukey's test) identified that the RM to the deflection of the caudal SW C1 was significantly reduced (42%) in 24h STIM compared with the NO STIM group (mean ± SD: 0.29 ± 0.25 in NO STIM vs. 0.17 ± 0.23 in 24h STIM.; P < 0.01; Fig. 8A). The response magnitude to rostral SW C3 was also reduced after stimulation; however, this decrease did not reach a significant level. Responses evoked by SWs on Arc-1 to Arc-3 were not affected. Four days after the stimulation was stopped, a tendency toward lower values of mean RM was still observed to the deflection of the C1 and C3 whiskers compared with nonstimulated mice, but these differences were not statistically significant. Figure 8B shows the mean population PSTHs for the C1 whisker for the three groups of mice. Evoked activity above spontaneous firing starts at 9–10 ms poststimulus, the peak of activity was reached around 17–18 ms poststimulus, then firing gradually decreased until a long-lasting plateau of low firing took place. The temporal response profile remained unaltered after the 24-h period of chronic stimulation and modal latency of neurons to the deflection of SW C1 was not altered by chronic stimulation (20.7 ± 17.5 ms in NO STIM, 19.3 ± 18.2 ms in 24h STIM, 24.0 ± 14.4 ms in 4-days-after-STIM; mean ± SD). However, evoked spiking probability of C1 responses was significantly lowered during almost the entire duration of the response, i.e., from T2 to T4, in the 24h STIM group of animals compared with the NO STIM and the 4-days-after-STIM groups.

In layers II & III, chronic sensory stimulation led to effects similar to those in layer IV: besides a decrease in PW-evoked mean RM, we observed a 52% decrease in mean RM to deflection of the C1 whisker (0.32 ± 0.30 in NO STIM vs. 0.15 ± 0.23 in 24h STIM; P < 0.01; Fig. 8A). Mean RMs to deflection of all the other SWs were not significantly different between groups. Figure 8B presents the mean population PSTHs of neurons to deflection of SW C1. Onset of evoked activity above the spontaneous level occurred at 9–10 ms poststimulus, rapidly reached maximal firing probabilities, then decreased to a long-lasting plateau of low spiking probability around 25 ms poststimulus in nonstimulated mice. After stimulation, a drastic reduction in firing affected the entire duration of the response. Modal latency to whisker C1 deflection was not altered by chronic stimulation (23.3 ± 15.3 ms in NO STIM, 19.7 ± 7.8 ms in 24h STIM, 23.3 ± 15.0 ms in 4-days-after-STIM; mean ± SD).

Reduction of evoked responses in adjacent barrel columns

Neurons recorded in the C1 and D2 barrel columns that were used for the quantitative analysis of PW responses described above were also tested for deflection of their SWs using the standard protocol. The MANOVA did not detect a global effect of the group on RM either in the C1 or in the D2 barrel columns. However, this model could have underestimated the potential effect of the stimulation period on RM of neurons to individual whiskers because it was designed to be highly robust, minimizing type II error (false positive). Therefore despite the negative MANOVA overall test, we decided to compare separately the RMs of neurons for each individual whisker deflection between the two groups of animals using nonparametric statistics. Four or seven SWs were tested in D2 and C1 columns, respectively. Therefore the level of significance of the nonparametric tests was lowered to P = 0.005 to test whether stimulation alters RM (standard Bonferroni correction for multiple comparisons). In layer IV of barrels C1 and D2, RMs were not altered after chronic stimulation on deflection of any SW (all P >> 0.05; Fig. 9, A and B). In layers II & III, however, mean RM to the deflection of the C2 whisker decreased with high P values in barrel columns C1 (P = 0.0007) and D2 (P = 0.0038), whereas P values for all other SWs were >0.05. The specific decrease on the deflection of the C2 whisker and the deliberate robustness of our multivariate model could explain why the MANOVA test did not reveal an overall group effect.


Figure 9
View larger version (22K):
[in this window]
[in a new window]

 
FIG. 9. Effects of chronic C2 whisker stimulation onto RM in barrel columns C1 and D2. Mean RM ± SE for neurons in barrel column C1 (A) and D2 (B) on PW and SW deflections. Asterisks indicate RM values that differ between groups at P < 0.005 (Kruskal–Wallis). Note that in both barrel columns, the PW response is unaltered for cells in layer IV and layer III/III, whereas the response to the C2 whisker was significantly decreased in layers II & III but not in layer IV. RM on deflection of the other SWs was not significantly modified.

 

    DISCUSSION
 TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 GRANTS
 ACKNOWLEDGMENTS
 REFERENCES
 
Unaltered spatiotemporal characteristics of barrel column activation

The CSD analysis showed a stereotype sequence of activation throughout the barrel column after deflection of the corresponding whisker. This activation pattern starts with a current sink centered in layer IV at 5–7 ms after the deflection and lasts for roughly 20 ms. Both the onset and the spatial location of the subsequent events that we report here are almost identical to the results of CSD analyses performed in the rat barrel cortex (Castro-Alamancos and Connors 1996Go; Di et al. 1990Go). Current sinks and sources originate mainly from local changes in neuronal membrane conductance produced by synaptic activation and less by those created by propagated APs (Leung 1990Go; Mitzdorf 1985Go). The early layer IV sink presumably reflects the excitatory drive of the glutamatergic TC projections that arborize predominantly in this layer (Agmon and Connors 1991Go; Agmon and Odowd 1992Go; White 1978Go). This assumption is supported by the dependency of the negative going phases of LFPs evoked in layer IV on {alpha}-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) and N-methyl-D-aspartate (NMDA) glutamate receptors as shown in rat cortical slices (Wirth and Luscher 2004Go). After 1–3 ms the sink in layer IV invades layers II & III and is followed by a sink in the infragranular layers. This sequence is in agreement with observations in rats demonstrating excitatory outputs of layer IV to layers II & III (Feldmeyer et al. 2000Go; Lübke et al. 2000Go; Petersen and Sakmann 2001Go) and of layer IV and layers II & III to infragranular layers (Gottlieb and Keller 1997Go; Harris and Woolsey 1983Go; Zhang and Deschenes 1998Go). It is in agreement with the postulated "VB-to-layer IV-to-layers II & III-to-layers V and VI" pathway in rodents (Armstrong-James and Fox 1987Go; Armstrong-James et al. 1992Go; Welker et al. 1993Go). The current source that follows the early sink in layer IV may be driven by feedforward and feedback inhibition from layer IV inhibitory interneurons (Porter et al. 2001Go). This current source temporally matches a positive LFP phase and goes with a weak spiking period. Wirth and Lusher (2004)Go showed that the positive LFP phase that interrupts the negative LFP phase evoked by the activation of layer IV is dependent on {gamma}-aminobutyric acid type A (GABAA) receptors. Castro-Alamancos and Connors (1996)Go showed that the spatiotemporal sequence of barrel cortex activation as determined by CSD analysis in rats can be altered by repetitive thalamic stimulation. However, our study reveals that a chronic stimulation period does not modify the laminar locations, onset latencies, and duration of the current sinks and sources evoked by the stimulated whisker.

Altered activity of single units after chronic stimulation

DECREASED SPONTANEOUS FIRING IN THE STIMULATED BARREL COLUMN. Spontaneous activity of single units decreases in layer IV and layers II & III of the stimulated but not of the adjacent nonstimulated barrel columns (Fig. 2). This may be explained by modifications in the intrinsic properties of neurons within the stimulated barrel column, but could also be the consequence of modifications in the chemical environment of the neurons. For instance, an elevated concentration of extracellular GABA induced by the 24-h stimulation period could persist in the stimulated column during the recording session (Welker et al. 1989Go; see following text). This could modulate neuronal firing by enhancing tonic inhibitory currents by high-affinity extrasynaptic GABAA receptors (Banks and Pearce 2000Go; Fritschy et al. 1992Go; Rossi and Hamann 1998Go; Semyanov et al. 2003Go) and induce a decrease in the probability of spontaneous spikes without preventing spiking driven by evoked activity. Alternatively, the decreased spontaneous firing could be the consequence of a persistent reduction in the concentration of glutamate in the extracellular medium.

Studies from our laboratory have generated arguments in favor of both alternatives. Involvement of the GABAergic innervation in the effects of whisker stimulation was first observed by an increase in glutamate decarboxylase (GAD) immunoreactivity in chronically stimulated barrels (Welker et al. 1989Go). In an ultrastructural analysis, we subsequently discovered that 24 h of whisker stimulation significantly decreases the ratio between excitatory and inhibitory synapses inside the barrel (Knott et al. 2002Go). This shift in the ratio arose from an increase in the density of inhibitory synapses that exceeded the increase in the density of excitatory synapses. Besides this relative increase in the density of inhibitory synapses, we also observed a significant increase in the occurrence of inhibitory synapses on spines that invariably also received an excitatory synapse. Turning to the glutamatergic hypothesis, we showed an upregulation of glutamate transporters GLAST and GLT-1 in the stimulated barrel (Genoud et al. 2006Go) that could indicate an increased capacity to remove glutamate from the extracellular space and as such influence the level of excitation. These two sets of analyses elucidated modifications that were restricted to the stimulated barrel and could therefore have locally contributed to a decrease in the level of spontaneous activity that in general is considered to be synchronized across barrel columns (Armstrong-James et al. 1985Go; Erchova et al. 2002Go; Petersen et al. 2003aGo).

ATTENUATION OF RESPONSES TO THE PW IN LAYER IV OF THE "STIMULATED" BARREL COLUMN. Layer IV is the cortical entry of peripheral signals. In rodents, each barrel processes primarily sensory signals ascending from the corresponding whisker (Armstrong-James and Fox 1987Go; Simons 1978Go; Welker et al. 1992Go), reflecting the segregation of TC-axonal branches from individual barreloids within the corresponding barrels (Arnold et al. 2001Go; Bernardo and Woolsey 1987; Frost and Caviness 1980Go; Jensen and Killackey 1987Go; Killackey 1973Go). In mice and rats barrels contain glutamatergic neurons, mainly spiny stellate cells, and GABAergic inhibitory aspiny neurons (Harris and Woolsey 1983Go; Porter et al. 2001Go; White 1978Go; White and Rock 1980Go). Both cell types receive glutamatergic TC synapses of high efficacy, which should ensure a robust impact of TC activation in layer IV (Agmon and Odowd 1992Go; Benshalom and White 1986Go; Gabernet et al. 2005Go; Gil et al. 1999Go). After initial monosynaptic TC activation of layer IV through AMPA and NMDA receptors, subsequent evoked neuronal spiking is mediated by excitatory interactions of cortical origin that depend almost entirely on NMDA receptors, as demonstrated in the rat and in the mouse (Armstrong-James et al. 1993Go; Fleidervish et al. 1998Go; Gil and Amitai 1996Go). In both species, spiny stellate neurons form a densely interconnected excitatory network within the barrel (Beierlein et al. 2003Go; Feldmeyer et al. 1999Go; Gil et al. 1999Go; Lübke et al. 2000Go) that mediates a selective amplification of thalamic signals before its propagation throughout the cortical column by axonal collaterals. Concomitantly, inhibitory interneurons generate strong disynaptic feedforward inhibition: on TC activation, an excitatory postsynaptic potential (EPSP) is evoked at the soma of virtually all barrel neurons and is followed within 1–3 ms by an inhibitory postsynaptic potential (IPSP) that curtails the EPSP as soon as 10–11 ms after its onset, as shown both in rats (Carvell and Simons 1988Go; Higley and Contreras 2003Go; Moore and Nelson 1998Go; Sun et al. 2006Go) and in mice (Agmon and Connors 1992Go; Gabernet et al. 2005Go; Porter et al. 2001Go). Evoked inhibition is often followed by a second EPSP at variable latencies (50–200 ms poststimulus). The sequence of those subthreshold events is consistent with the suprathreshold profile generated by the PW that we observed in layer IV population PSTHs (Fig. 5A): a strong feedforward inhibition could explain the decrease in evoked AP firing after it has peaked around 11 ms, whereas a second wave of EPSP after release from inhibition could match the late period of low evoked firing (T4). This biphasic profile remains unaltered after chronic stimulation.

In a previous work we demonstrated that the magnitude of PW-evoked single-unit spiking decreases in the stimulated barrel relative to spiking in the same barrel of nonstimulated mice (Knott et al. 2002Go); here supplementary neurons were added to the data and we confirm this observation: RM significantly decreases by 25%. In addition, we show that the depression of PW-evoked response is restricted to the "stimulated" barrel because responses in adjacent "nonstimulated" barrels C1 and D2 to their corresponding PW are not altered. These observations correlate well with the similarly localized decrease in deoxyglucose uptake described by Welker et al. (1992)Go. The depression of neuronal firing may be attributable to homeostatic downregulations of the cortical network to stabilize its own level of activity in response to a period of chronic increase in sensory drive. Even though far removed from a natural setting, one can also imagine that the depression of cortical responses to the stimulated whisker represents a habituation-like phenomenon to a continuous and behaviorally irrelevant stimulus.

To investigate the functional consequences of the decreased response we modified in a set of experiments the amplitude and the repetition rate of the whisker deflection during the recoding session. As previously reported in rats (Armstrong-James and Fox 1987Go; Peterson et al. 1998Go; Pinto et al. 2000Go; Wilent and Contreras 2004Go), here we show in the mouse that the magnitude of layer IV neuronal response is dependent on the amplitude of the PW deflection (Fig. 6). With incremented deflection amplitude, latency decreases while firing robustly increases, which presumably reflects stronger and more synchronized TC inputs to layer IV units (Pinto et al. 2000Go). After chronic stimulation the neuronal responses were weaker at all deflection amplitudes, although at the smallest amplitude this difference does not reach significance. At this amplitude RM still exceeds spontaneous level in both groups of animals. This amplitude is near-threshold; it induces an upward deflection of only 0.06°, close to the threshold estimated in rats by Peterson et al. (1998)Go, and mean RM of neurons is weak, barely exceeding 0.1 spike per deflection, one third of neurons being unresponsive in both groups of mice. Thus RM varies from virtually 0 at the smallest amplitude to an average of 1.18 spikes in nonstimulated and 0.85 spike in stimulated animals at the largest amplitude. Thus the dynamic range of response for deflection amplitude is reduced after chronic whisker stimulation.

We further analyzed the cortical responses after varying the repetition rate of the whisker deflection. In rats, layer IV neurons exhibit adaptation in their response to increasing rates of repetitive deflections (Ahissar et al. 2000Go; Garabedian et al. 2003Go; Khatri et al. 2004Go; Melzer et al. 2006Go; Simons 1978Go). Although the absolute value of the response magnitude varies between experimental paradigms used (method of stimulation, type of anesthesia, determination of RM), we showed here that layer IV neurons in the mouse adapt their response magnitude in a manner similar to that in the rat. It is thought that adaptation in layer IV results from short-term depression at TC synapses to both excitatory and inhibitory neurons (Chung et al. 2002Go; Hartings et al. 2003Go) and at cortico-cortical excitatory and inhibitory synapses (Finnerty et al. 1999Go; Gabernet et al. 2005Go; Gil et al. 1997Go; Higley and Contreras 2006Go; Sosnik et al. 2001Go; Webber and Stanley 2006Go). We observed here that in layer IV the adaptation of the neuronal response is not affected after chronic stimulation, suggesting that short-term dynamics at TC and cortico-cortical synapses within layer IV is presumably unaltered. However, neurons in layers II & III show a significant modification in the degree of adaptation of their responses to incremented deflection rates. This difference between cortical layers is most likely explained by the fact that the RM of neurons in layers II & III is more depressed than that of layer IV neurons after the chronic whisker stimulation period (see following text).

To conclude, chronic stimulation leads to a modulation of the magnitude of the response that does not alter the sensitivity of neurons for the deflection of the stimulated whisker. The reduced dynamic ranges of amplitude- and repetition-rate–dependent responses, however, may affect the retrieving and encoding of crucial stimulus clues (Moore 2004Go; Petersen 2002Go).

The epoch analysis further elucidated the possible origin of RM depression of layer IV neurons after chronic whisker stimulation. Although onset of the layer IV response is not modified (T1; 3–10 ms after deflection onset), the analysis revealed a significant RM depression during the second epoch (T2; 12–25 ms poststimulus; see Fig. 5). In urethane-anesthetized rats, the period of strongly synchronized spiking of barreloid neurons peaks 5–6 ms after the stimulus and lasts between 4 and 10 ms poststimulus (Armstrong-James and Callahan 1991Go; Diamond et al. 1992Go; Glazewski et al. 1998Go; Pinto et al. 2000Go). The strength of the TC volley is then strongly reduced, presumably because of the strong inhibition generated by the reticular nucleus (Brecht and Sakmann 2002bGo). Taking this to the mouse, the unaltered T1 response indicates a TC activation that is not modified after chronic stimulation. Quantification of evoked firing in barreloid neurons of stimulated and nonstimulated mice will be useful to confirm this assumption.

Other studies on the barrel cortex showed that short-term experience-dependent changes of cortical receptive field are mediated by synaptic plasticity occurring predominantly at intracortical synapses (Armstrong-James et al. 1994Go; Feldman and Brecht 2005Go; Glazewski et al. 1998Go; Rema et al. 1998Go; Wallace and Fox 1999aGo). Only after long periods of modified sensory experience such as 30 days of whisker-pairing (Armstrong-James et al. 1994Go; Diamond et al. 1993Go) does TC plasticity contribute to an altered cortical response. However, this concept—based on quantitative analysis of response properties of layer IV neurons in anesthetized rats—was recently challenged by an in vivo recording study that indicates that TC activation may be altered within 4 h of sensory deprivation (Sellien and Ebner 2006Go).

The depression of PW-evoked firing during T2 in the "stimulated" barrel is likely a functional consequence of the increased density of