| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Physiology, Zlotowski Center for Neuroscience, Ben-Gurion University, Beer-Sheva, Israel
Submitted 16 May 2006; accepted in final form 3 September 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
10 ms) and suppression prevailing at longer intervals (>10 ms). These temporal interactions resulted in a time-dependent regulation of direction tuning of cortical neurons. The synaptic mechanisms underlying L1 inputs’ influences were examined using whole cell recordings in vitro while pairing L1 and white-matter stimulations. We found time-dependent, layer-specific differences in synaptic summation of the two inputs, with supralinearity at shorter intervals and sublinearity at longer intervals that resulted mainly from shunting inhibition. Taken together, our results demonstrate that L1 inputs impose a time- and layer-specific regulation on sensory-evoked responses. This in turn may lead to a dynamic transmission of sensory information in the somatosensory cortex. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
; Zeki and Shipp 1988). Although this model has been relatively successful in explaining certain aspects of cortical sensory information processing, it fails to account for anatomical, physiological, and psychophysical studies indicating that feedback and top-down effects play a crucial role in sensory integration (Engel et al. 2001; Lamme et al. 1998, and references therein). Numerous studies have shown that cortical areas at a higher stage of the hierarchy influence receptive field properties, firing rates, and the degree of neuronal synchronization at lower cortical areas (Bullier et al. 2001; Fries et al. 2001; Hupe et al. 2001; Steinmetz et al. 2000). These influences may be associated with attention, stimulus context, figure-background discrimination, memory organization, and touch discrimination (Cauller 1995; Cauller and Kulics 1991; Fries et al. 2001; Hupe et al. 1998; Lamme et al. 1998; Roelfsema et al. 1998; Squire and Zola-Morgan 1991; Steinmetz et al. 2000).
The anatomical substrate for these feedback and top-down effects resides in both the network of local intrinsic horizontal connections (Amir et al. 1993; Gilbert and Wiesel 1989; Livingstone and Hubel 1984; Rockland and Lund 1983) and the dense network of top-down and feedback interareal connections (Angelucci and Bullier 2003; Angelucci et al. 2002). The mechanisms by which these pathways produce their impact are largely unexplored.
One of the targets of interareal connections is the superficial L1 (Cauller and Kulics 1991; Mitchell and Cauller 2001). In addition to axons originating from local pyramidal and nonpyramidal neurons, this layer contains significant inputs from other regions (Vogt 1991), including diffuse top-down projections from higher cortical areas and some thalamocortical axons from specific and nonspecific nuclei (Diamond 1995; Herkenham 1980; Mitchell and Cauller 2001; Rockland and Pandya 1979; Zeki and Shipp 1988). The axons in L1 are predominantly glutamatergic and GABAergic and their main targets are the distal apical dendritic tufts of pyramidal cells in layers 2/3 and 5 (Budd 1998; Hestrin and Armstrong 1996; Tamas et al. 2003; Vogt 1991; Zhu and Zhu 2004). Furthermore, a small number of nonpyramidal neurons are also located in L1 (Chu et al. 2003; DeFelipe and Jones 1988; Zhou and Hablitz 1996).
Although much is known about the anatomical organization of L1, the influences of L1 distal synaptic inputs on synaptic integration and the contribution of L1 to sensory information processing remain largely unknown. Several mechanisms may combine to amplify distal excitatory inputs and allow their propagation to the soma, but it has been generally accepted that inhibition contacting the apical tuft affects dendritic processing only locally (Cauller and Connors 1994; Miles et al. 1996; Salin and Prince 1996; however, see Williams and Stuart 2003). We explored the impact of L1 activation on cortical processing using extracellular recordings in vivo in the primary somatosensory (barrel) cortex of rodents. We found that L1 inputs regulate the magnitude, variance, and direction selectivity of sensory responses. Pairing L1 and white-matter (WM) stimulation in vitro revealed corresponding layer-specific, time-dependent nonlinear summations of synaptic potentials, which most likely underlie the interactions of the two activated pathways in vivo.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Adult male Sabra rats (250–300 g) were used. All experiments were conducted in accordance with international and institutional standards for the care and use of animals in research. Surgical anesthesia was induced by urethane (1.5 g/kg, administered intraperitoneally) and maintained at a constant level by monitoring hindpaw withdrawal, corneal reflex, respiratory rate, and administering extra doses (10% of original dose) as necessary. Atropine methyl nitrate (0.3 mg/kg) was administered intramuscularly before general anesthesia to prevent respiratory complications. Body temperature was maintained near 37.1°C using a servo-controlled heating blanket (Harvard, Holliston, MA). After placing subjects in a stereotaxic apparatus (TSE, Bad Homburg, Germany), the somatosensory cortex was exposed by a 3 x 4-mm craniotomy centered on a point 2 mm posterior and 6 mm lateral to bregma. The vibrissae region of the somatosensory cortex was then identified according to vascular landmarks and stereotaxic coordinates (Chapin and Lin 1984; Hall and Lindholm 1974).
In vivo recording and stimulation
Cortical recordings were obtained using tungsten microelectrodes and two types of 16-channel silicone probes: a vertical probe and a 4 x tetrodes probe (NeuroNexus Technologies, Ann Arbor, MI). The vertical probe sites were separated vertically by 100 µm, whereas the 4 x tetrodes probe sites were 25 µm apart. Both of these probes had impedances of 1–2 M
. After alignment and mapping of the barrel, the electrodes were placed into the estimated center of the barrel (A–D, based on maximal responses from the principal vibrissa and minimal longer-latency responses from surrounding vibrissae). The electrodes were aligned normal to the cortical surface such that recording sites remained within the same barrel column. The laminar position of the electrodes was determined by both the electrode depth (resolution of 20 µm) and response latency (granular layer with the shortest latency). The location of layer 4 was estimated to be at 500–750 µm from the pia, a depth that usually showed the shortest response latency to sensory stimulation. The recorded signals were amplified (1,000-fold), band-pass filtered (1 Hz to 10 kHz), digitized (30 kHz/channel), and stored for off-line spike sorting and analysis. The data were then separated to local field potentials (LFPs; 1–150 Hz), and isolated single-unit activity (0.5–10 kHz). Spike extraction and sorting were accomplished with MClust (by A. D. Redish, available from http://www.cbc.umn.edu/
redish/mclust), which is a Matlab (The MathWorks, Natick, MA) based spike-sorting software. The extracted and sorted spikes were stored at a 1-ms resolution and peristimulus time histograms (PSTHs) were computed.
Handheld probes were used to identify the whisker evoking the strongest response from an isolated unit, i.e., the principal whisker. A piezoelectric mechanical stimulator was then attached to this whisker about 2 mm from the face. We used simulated-textures stimuli in which constant-seed, Gaussian-distributed white-noise (500 Hz, low-pass filtered to prevent ringing; the piezoelectric wafer was modified to increase its resonant frequency to 650 Hz) voltages were used to drive the piezoelectric wafer stimulator (Arabzadeh et al. 2005). To calibrate and monitor the piezoelectric element, a noncontact optical displacement-measuring system was used (Micro-Epsilon, Ortenburg, Germany). The 500-ms single-axis, two-direction stimulus was adjusted to the direction preference of the recorded cell. When measuring the direction selectivity of cortical neurons, single-axis, single-direction stimulation was used. Quantitative data were obtained in response to 50–100 presentations of stimuli at numerous amplitudes (50–350 µm) every 2 s. Displayed responses show averages over multiple stimulations.
In some experiments, we paired a preceding L1 electrical stimulation with a subsequent sensory stimulation and varied the interval between them. A bipolar stimulating electrode was placed on the surface of the cortex at about 2 mm laterally to the recording column, to activate afferent L1 fibers (pulse duration: 100 µs). Drugs aimed at blocking L1 inputs such as 6,7-dinitroquinoxaline-2,3-dione (DNQX, 20 µM, Sigma–Aldrich) or tetrodotoxin (TTX, 5 µM, Alomone, Jerusalem, Israel) were applied to the surface of the cortex.
Slice preparation, recordings, and stimulation
Mice (CD1, 21–28 days old) were deeply anesthetized with pentobarbital and decapitated; their brains were quickly removed into cold (5°C) physiological solution. Coronal cortical slices (350 µm thick) were cut with a vibratome (Campden Instruments, London, UK) and then transferred to a holding chamber, where they were kept at room temperature for 
1 h before recording, continuously bubbled by 95% O2-5% CO2. Recording was done in a chamber mounted on an upright microscope equipped with IR/DIC optics (Nikon Physiostation EC-600), where they were held at 32–34°C and constantly perfused. The normal bathing solution contained (in mM): 124 NaCl, 3.5 KCl, 2 MgSO4, 1.25 NaHPO4, 2 CaCl2, 26 NaHCO3, and 10 dextrose, and was saturated with 95% O2-5% CO2 (pH 7.4).
Whole cell recordings were made from layer 2/3 or layer 5 pyramidal neurons in the barrel field. Patch micropipettes (4–6 M) were filled with a solution containing (in mM): 125 K gluconate, 5 NaCl, 2 MgCl2, 10 EGTA, 10 HEPES, and Na2ATP (pH 7.2, 280 mOsm). Voltages were recorded with a patch-clamp amplifier (AxoPatch 2B, Axon Instruments), and digitally sampled at 10 kHz. Series resistance was typically <15 M. One bipolar stimulating electrode was placed in the white matter (WM) under the recording area, to activate a column of neurons. The other stimulating electrode was used to stimulate L1 and was placed 1 mm lateral to the recording area to activate afferent L1 fibers while minimizing the activation of neurons in the same column. The two pathways were stimulated alternately and in conjunction, at 0.1 Hz for each pathway, using pulses of 40 µs and about 50 µA. The intensities of the stimuli were adjusted such that the L1 stimulus was set to produce the maximal response (which was always subthreshold), whereas WM stimulation was set to produce synaptic potentials of amplitudes around half spike threshold.
Data analysis
One-dimensional current-source density (CSD) was calculated from the second spatial derivative of the LFP profile (Freeman and Nicholson 1975; Mitzdorf 1985)
 | (1) |
is the field potential, z is the coordinate perpendicular to the laminae, 
z is the sampling interval (100 µm), and nz is the differentiation grid (n = 1). To obtain upper and lower boundary sites, we used an extrapolation method that assumes no additional decay in the field potential above and below the uppermost and lowermost recording sites, respectively (Vaknin et al. 1988). In CSD traces, current sinks are indicated by upward deflections and sources by downward deflections. To facilitate visualization of CSD profiles, we generated color image plots using linear interpolation along the depth axis.
The analysis of response reliability was similar to that of Mainen and Sejnowski (1995), with some differences, and is illustrated in Fig. 3. The first step consisted of constructing a PSTH with a bin width of 1 ms (e.g., Fig. 3A). The next step was to define an "event"—that is, an action potential obtained with a sufficiently high temporal precision to yield large bins in the PSTH. To do that, we arbitrarily chose a threshold value corresponding to a height of 3 SDs of spontaneous activity (horizontal dashed line in Fig. 3A). For simplicity, we restricted the analysis to the first 25 ms after whisker deflection, and only the first event, which corresponds to the initial response, was considered. The dashed vertical lines in Fig. 3, A and B delimit such events.
|
|
|
The direction selectivity of single neurons was determined by assigning each with a direction index (DI) according to the formula
 |
To quantify the synaptic reversal potential and conductance, we used an established technique (Anderson et al. 2000; Hirsch et al. 1998) based on the measurement of input conductance. We measured membrane potential (Vm) responses to each stimulus while injecting, in turn, different levels of steady current. Using these Vm signals we fit the current–voltage relationship at each time point with the equation
 | (2) |
We made the simplifying assumption that the conductance of a cell is the sum of its excitatory and inhibitory synaptic conductances [ge(t) and gi(t), respectively] and a constant resting conductance (grest)
 | (3) |
In the absence of synaptic stimulation, we take the resting conductance grest to be equal to the total conductance of the cell and the synaptic conductance to be equal to zero. Given this relation, Vsyn can be expressed as a function of conductance
 | (4) |
-aminobutyric acid types A and B (GABAA and GABAB, respectively) equilibrium potentials. Vrest was obtained from the Vm recordings during the control conditions and grest was calculated from the voltage deflections in response to hyperpolarizing current pulses. Equations 2 and 3 can be solved at each time point to yield values of ge(t) and gi(t). Thus for each series of Vm measurements, these calculations yield two corresponding time series reflecting the excitatory and inhibitory synaptic conductances of the cell. The significance of the differences between the measured parameters was evaluated using one-way ANOVA. When significant differences were indicated in the F-ratio test (P < 0.05), the Tukey method for multiple comparisons was used to determine those pairs of measured parameters that differed significantly within the pair (P < 0.05 or P < 0.01). Averaged data are expressed as means ± SD. Error bars in all the figures indicate the SD.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
To examine how L1 regulates sensory-evoked responses we recorded LFPs evoked by white noise whisker stimulation with a linear probe of 16 electrodes inserted perpendicular to the cortical surface. The location and direction of transmembrane currents were determined using CSD analysis. The early laminar CSD profile of response to standard brief whisker deflection (Fig. 1A) was similar to that of previous reports (Mitzdorf 1985; Swadlow et al. 2002), revealing an early postsynaptic current sink centered in layer 4 in parallel with current sources in infra- and supragranular layers. Although more complex, the early CSD profile of responses to white noise stimuli revealed a similar current sink in layer 4 (Fig. 1B).
We next applied TTX to the surface of the cortex (n = 7). To ensure that the effects of TTX were limited to L1, we monitored the amplitude of evoked responses recorded at a depth of about 100 µm (L1) and about 200 µm (upper layer 2). Only sessions in which whisker-evoked LFP responses in layer 2 did not exhibit any significant reduction, but rather reached an elevated plateau (Fig. 1D, gray area), were considered for analysis. During this plateau period, TTX application did not change the frequency of spontaneous activity nor its spectral content in layer 2 (Fig. 1H). On the other hand, it increased the whisker-evoked response magnitude (Fig. 1C, right) and variance (Fig. 1E, right), as expressed in both an augmentation and prolongation of all current sinks and sources. To quantify these changes we used two methods: First, we subtracted the CSD profile in control conditions from the CSD after TTX application. The results revealed positive values in most layers (Fig. 1F), confirming a general enhancement of the responses. Second, we quantitatively estimated the borders of whisker-evoked CSD activity by testing, at each point in the figure, the null hypothesis that the neural response is not different from spontaneous activity. Contours where this null hypothesis is rejected at the P = 0.05 level were then superimposed on the map to demarcate the border of the significant activity. Figure 1G displays an example of the borders of the responses taken from the right panel in Fig. 1C. The mean level of activity within each border was compared between the control and the TTX conditions. The analysis revealed a significant increase in response magnitude and variability that was consistent across all animals (n = 7). These changes in response characteristics were more prominent in infragranular than those in supragranular layers (Fig. 1; Table 1; P < 0.01). It is important to note that no significant differences were observed in the initial response in layer 4 (Fig. 1C, arrows). These results indicate that L1 modifies sensory-evoked responses by regulation of intracortical excitability.
|
-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)–receptor blocker DNQX. As before, to ensure that the effect of DNQX is confined to L1, we monitored the amplitude of evoked responses in the superficial electrodes and only sessions that did not exhibit any significant reduction in layer 2 responses were considered (Fig. 2C). Application of DNQX to the cortical surface did not influence whisker-evoked response magnitude (Fig. 2, A and C; Table 1), but it reduced responses’ variance (Fig. 2B; P < 0.01; n = 6). This reduction was more prominent in infragranular layers (P < 0.01). Taken together, these results indicate that L1 can exert a strong inhibitory influence on whisker-evoked responses that serves to reduce their magnitude and variability (see also Shlosberg et al. 2003). We next examined the effect of blocking L1 on single-neuron activity. Whisker-evoked responses were recorded from infra- and supragranular layers neurons while applying TTX (infragranular: n = 11; supragranular: n = 10) or DNQX (infragranular: n = 10; supragranular: n = 9) to the surface of the cortex as before. Drug-induced changes in response properties were evaluated by measuring the magnitude, reliability, and precision of the spiking responses (see METHODS). An example of a supragranular neuron response to white noise whisker stimulation is displayed in Fig. 3. Application of TTX resulted in a significant increase of response magnitude while reducing its reliability and precision (Fig. 3, C and D; infragranular: n = 9; supragranular: n = 7). Application of DNQX had no effect on either the magnitude or precision of the responses, although it did increase their reliability (Fig. 3, C and D; infragranular: n = 8; supragranular: n = 6). These results further support the notion that L1 may influence both the magnitude and the temporal fidelity of sensory responses by dominant inhibitory input.
Effects of layer 1 stimulation on sensory responses
The above results allude to a considerable inhibitory function for L1 in sensory integration. However, it was previously shown that the contribution of L1 to sensory-evoked responses is reduced in anesthetized animals (Cauller and Kulics 1991). Therefore we electrically stimulated L1 in conjunction with sensory stimulation.
Stimulation of L1 evoked two types of laminar activity, depending on stimulus intensity. At lower intensities (
10 µA), a short-latency (1.2 ± 0.3 ms) negative potential appeared in L1 and upper layer 2 (Fig. 4A, asterisk). This negative potential was followed by a local positive potential (3.6 ± 0.2 ms) and by global succeeding negative potentials (5.4 ± 0.4 ms), which lasted 50 ms. A CSD analysis revealed that the superficial current sink was associated with a current source immediately below and no current dipoles were observed in deeper layers (Fig. 4B), indicating that the synaptic activation evoked by this electrical stimulation was restricted to L1 and upper layer 2. In contrast, at higher stimulus intensities (>10 µA), a short-latency negative potential also appeared in layer 5 (Fig. 4C). These negative potentials were followed by local positive potentials and by global succeeding negative potentials. A CSD analysis revealed that current dipoles were observed in both superficial layers and layer 5 (Fig. 4D), indicating that L1 stimulation at higher intensities activated the entire cortical column. Thus we used lower stimulus intensities (<10 µA) for the rest of the study to examine the effect of L1 activation on sensory inputs. To rule out the possibility that L1 electrical activation indirectly alters sensory responses by changing the cortical network state, we compared the integrated power of spontaneous activity in layer 2 before and between stimuli in low- (1–12 Hz), medium- (12–25 Hz), and high-frequency (25–40 Hz) bands. This analysis did not reveal significant changes in these parameters, indicating that the cortical rhythmic activity was not systematically driven in one way or another by the stimulation (Fig. 4E).
|
10 ms) resulted in an enhancement of whisker-evoked LFP magnitude accompanied by a reduction in response variability and duration (Fig. 5A, top traces), suggesting an increase in neuronal synchronization. Again, the increase in response magnitude was more pronounced for infragranular than for supragranular layers (Fig. 5C, P < 0.05). Longer interstimulus intervals (10 ms < ISIs < 50 ms) resulted in a reduction of both magnitude and variability of sensory-evoked responses (Fig. 5). The reduction in response magnitude was more pronounced for infragranular layers at ISIs of 20 and 30 ms, whereas the change in variability did not differ between layers.
|
10 ms) increased all three parameters of sensory responses, whereas at longer ISIs (>10 ms) there was a reduction of sensory-evoked responses and a return of reliability and precision measures to control levels (Fig. 6B). These data further demonstrate that the temporal relationship between L1 activation and whisker deflection may regulate the flow of sensory information.
|
Several studies demonstrated that neurons in the barrel cortex have angular deflection preference (Bruno and Simons 2003; Simons and Carvell 1989). Based on our findings, we conjectured that L1 activation might also influence the direction selectivity of the neurons in a time-dependent manner. Figure 7A depicts the effect of L1 activation on a direction-tuning curve by plotting the neurons’ responses to all directions normalized to its preferred direction. L1 stimulation 5 ms before the sensory stimulation resulted in a reduced direction selectivity, as reflected in a decrease of DI from 0.46 to 0.29. An ISI of 20 ms increased the direction selectivity (DI change from 0.46 to 0.88). A summary of L1 stimulation effects across all neurons confirms that shorter ISIs reduce the direction selectivity, whereas longer ISIs increase it (Fig. 7B, Table 2, n = 11). Altogether, these findings demonstrate that L1 inputs can have inverse effects on sensory responses, depending on the temporal relationship: when activated at close temporal proximity to sensory stimuli they enhance response magnitude and reliability while decreasing its selectivity. In contrast, at longer ISIs they reduce the response magnitude while increasing its direction selectivity.
|
|
In a second series of experiments, we sought to determine the synaptic mechanisms underlying the interactions between L1 inputs and sensory responses by performing whole cell recordings in slices. We first measured the responses of layer 2/3 (n = 17) and 5 (n = 14) neurons to cortical superficial stimulation while injecting steady hyperpolarizing and depolarizing currents to hold the cells at different membrane potentials. Two types of L1 stimuli were used; a distant stimulating electrode was placed more than 1 mm lateral to the recorded neuron and its stimulus intensity was set to produce the maximal response, which was always subthreshold. When this electrode was placed at the same horizontal distances but in lower laminae it did not evoke synaptic responses on the recorded neurons (not shown), confirming that the synaptic responses resulted from selective activation of incoming pathways within L1. Furthermore, L1 stimulation did not evoke synaptic responses in any of the layer 4 neurons recorded (n = 7), indicating that our stimulation affected only neurons that have an apical dendrite that ascends to L1 (i.e., layers 2/3 and 5). For comparison, another stimulating electrode was placed near the recording area right under the pia and its intensity was set to produce postsynaptic potential (PSP) amplitudes of around half spike threshold.
We then calculated the synaptic reversal potentials and conductances associated with the responses based on two simplifying assumptions. First, we assumed that the current–voltage relationship in the recorded neurons is linear. To evaluate this, we calculated the linear correlation coefficient between injected current and the Vm (Fig. 8B). In general, the correlation coefficients were quite high (range 0.79–0.94) and statistically significant. Second, we assumed single equilibrium potentials for synaptic excitation (Ee) and inhibition (Ei). We chose Ei to be –80 mV, a value between those for GABAA and GABAB conductances, and assumed a single type of voltage-independent conductance for Ee (0 mV). To control for possible effects of incorrectly estimating the synaptic equilibrium potentials, we repeated the analyses after varying the values of Ee and Ei by 10 mV in either direction (Anderson et al. 2000). These changes did not significantly affect the basic results.
|
|
|
; Holt and Koch 1997). One of those is the reduction in the synaptic driving force. Preceding L1 stimulation usually evoked depolarization that would decrease the synaptic driving force to a subsequent WM-evoked response. To address this possibility, we plotted the amplitude of the WM-evoked synaptic response against the baseline Vm and fit a linear regression to these points. The regression line was used to extrapolate the expected WM-evoked amplitude at the depolarized Vm caused by the preceding L1 activation. For the population, we plotted the WM-evoked PSP amplitude when preceded by L1 stimulation at 25-ms intervals (Fig. 10A, arrow) against the expected amplitudes derived by extrapolation (Fig. 10B). The data show a significant but weak correlation (r2 = 0.24; P < 0.01), indicating that the driving force plays a role in reducing the PSP amplitude. However, all the data points fall below the diagonal, suggesting that the observed amplitude was reduced more than could be accounted for by the decrease in driving force. This finding suggests that an additional mechanism contributes to the reduction in PSP amplitudes.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
; Chu et al. 2003; Hestrin and Armstrong 1996; Tamas et al. 2003; Zhu and Zhu 2004). Our results demonstrate that activation of axonal pathways in L1 regulates the magnitude and reliability of whisker-evoked responses in a time-dependent and layer-specific manner, and thus we propose that membrane shunting caused by these distal inputs is the main mechanism by which they exert their influence. Time-dependent modulation
To study how the barrel system integrates feedback, top-down, and feedforward inputs, we measured the response of cortical neurons to whisker deflection in vivo or WM stimulation in vitro preceded by L1 activation. Consistent with previous findings in thalamocortical slices (Llinas et al. 2002), we found that this interaction resulted in enhancement of the sensory responses or supralinear synaptic summation of the two inputs at shorter time intervals. In contrast, longer intervals strongly suppressed the spike output of the cells and reduced the WM-evoked PSP amplitude. These observations are similar to findings of previous studies exploring the columnar integration of multiple whiskers. In these studies, the interval between the stimuli also determined the outcome of the interaction, with facilitation dominating the short ISIs (Shimegi et al. 1999; but see Higley and Contreras 2003) and suppression of responses more prevalent at longer ISIs (Higley and Contreras 2003; Shimegi et al. 1999; Simons 1985; Simons and Carvell 1989). However, the mechanisms underlying response modulation in multiwhisker interactions are not fully understood because their anatomical substrate involves not only intracortical pathways, but also feedback loops through the thalamus. Our study, on the other hand, explores the interaction between two separate pathways, where the columnar organization is better defined.
Although the response enhancement at short ISIs is most likely mediated by dendritic voltage-gated conductances (see following text; London and Häuser 2005; Reyes 2001, and references therein), several mechanisms may explain the spike suppression and reduction in PSP amplitude in the present study. Because L1 and whisker input use independent pathways, the suppression of thalamocortical excitation by L1 has to be mediated by mechanisms affecting synaptic integration in the postsynaptic cortical neurons. Two processes, which are not mutually exclusive, may be involved: First, shunting inhibition occurs when an increased postsynaptic conductance divisively reduces the amplitude of coinciding synaptic potentials. Second, L1-induced synaptic depolarization can decrease the driving force of subsequent PSPs. Our data clearly demonstrate that the reduction in driving force may contribute, but is not the main mechanism underlying the reduction in WM-evoked PSP amplitude, because it correlates only weakly with L1-evoked Vm depolarization. Instead, the sublinear summation of the PSPs can be attributed to the increase in L1-evoked conductance caused mainly by shunting inhibition because the amount of reduction correlates with the calculated change in impedance caused by the preceding L1 stimulation (Fig. 10C).
Our interpretation provides only a partial explanation consistent with experimental evidence and additional mechanisms may be involved in the suppression of spike output in vivo. One likely candidate is the spike-generating mechanism. It has been demonstrated that spike threshold varies as a function of the amplitude and time course of membrane potential (Azouz and Gray 2000; Henze and Buzsáki 2001; Wilent and Contreras 2005). Thus subthreshold depolarizations resulting from L1 stimulation may lead to an increase in spike threshold and a reduction of spike-generation probability. An additional mechanism may involve indirect effects of increased membrane conductance. Because only small changes in Vm near threshold are necessary to change spikes output, a reduction in PSP variability arising from an increase in membrane conductance (Destexhe and Par é 1999; Par é et al. 1998) may cause a disproportionately larger suppression in spike output (Chance et al. 2002). Nonetheless, our results lend further support to the notion that synaptic inputs in the dendritic periphery may affect synaptic integration in neurons with dendrites in L1 (Shlosberg et al. 2003) to influence their sensory response.
Laminar specificity
In addition to receiving excitatory pathways, L1 contains axons originating from nonpyramidal Martinotti neurons that have vertically projecting axons that arborize profusely and nonspecifically (Fairén et al. 1984; Kawaguchi and Kubota 1997; Wahle 1993). We cannot rule out the possibility that at least part of the observed effects of L1 stimulation is mediated by synaptic inputs from collaterals of antidromically activated deeper-layer neurons. However, we did not observe an antidromic spike in any of the neurons we recorded from in vitro, and additional data from our laboratory prove that the threshold for evoking an antidromic spike in pyramidal neurons by L1 stimulation is much higher than the threshold required to elicit synaptic responses in the same neuron (not shown).
L1 also contains two types of nonpyramidal inhibitory neurons (Chu et al. 2003; DeFelipe and Jones 1988; Zhou and Hablitz 1996). The first is a "local circuit" interneuron with a large receptive field and an extended axonal branching field in L1. The second is a "deep-layer projecting" interneuron with a small receptive field and axonal projections to layer 2/3 (Zhu and Zhu 2004). The latter preferentially suppress the firing of other interneurons in layer 2/3 (Christophe et al. 2002). Thus selective activation of the different types of interneurons within L1 may result in layer-specific modulatory influences on columnar activation. Our results fit well with this functional organization scheme. We show that distant stimulation of L1 in vitro, which is likely to activate the long-range inhibitory axonal branching of Martinotti cells and "local circuit" interneurons, results in sublinear summation in layer 2/3 neurons. In contrast, local activation—which is likely to activate inhibitory axons and all L1 interneurons—results in supralinear summation in layer 2/3 neurons. This may result from disinhibition stemming from inactivation of layer 2/3 interneurons by "deep-layer projecting" neurons.
An additional mechanism capable of producing layer-specific L1 influence on sensory responses may be provided by dendritic voltage-gated conductances that have profound effects on synaptic integration (London and Hausser 2005). Calcium-dependent regenerative potentials in layer 5 neurons boost distal synaptic inputs to cause somatic spiking (Helmchen et al. 1999; Larkum and Zhu 2002; Schiller et al. 1997; Williams and Stuart 1999), whereas in layer 2/3 neurons these potentials are more labile and hardly spread to the soma (Waters et al. 2003). Thus dendritic amplification may have a more prominent effect on L1 excitatory activity in layer 5 neurons than that in layer 2/3 neurons. Such a mechanism may explain our findings that L1 stimulation resulted in a larger supralinear summation in layer 5 and that block of L1 activity in vivo had more prominent short-latency effects on sensory-evoked responses in infragranular layers.
Functional implications
Theories of top-down cortical processing propose that lower areas that are closer to sensory periphery serve as an interface where sensory information and predictions about features of a stimulus are compared. In this scheme, higher areas provide modulatory signals that control sensory information flow (Grossberg 1980; Mumford 1992; Ullman 1995). Our findings show that such feedback and top-down regulation may be mediated, at least in part, by L1. This regulation of sensory information transmission may be expressed in several aspects of cortical activity: First, neurons in the barrel cortex respond differentially to a variety of stimulus parameters such as angular direction, velocity, frequency, and amplitude of deflections (Bruno et al. 2003; Simons and Carvell 1989). We show that although activation of L1 does not change angular direction tuning of cortical neurons, it does regulate their selectivity in a time-dependent manner. Second, several studies have shown that the majority of information about stimulus location in the somatosensory system (Panzeri et al. 2001) and luminance contrast in the primary visual cortex (Reich et al. 2001) is carried by the time of the first spike after sensory stimulation. Here we show that activation of L1 controls the magnitude, precision, and reliability of the initial response after each whisker deflection.
L1 may also regulate sensory information transmission by promoting burst firing in layer 5 neurons, thereby securing the transmission of sensory information to recipient neurons (Larkum and Zhu 2002; Lisman 1997; Swadlow and Gusev 2001). This mechanism may also modulate the input–output relation (gain) of the neuron (Azouz 2005; Chance et al. 2002; Larkum et al. 2004).
A different model of how feedback and top-down influences might be implemented neurally suggests that neuronal synchronization can serve to integrate distributed neurons into cell assemblies and that this process underlies the selection of perceptually and behaviorally relevant information (Gray 1999; Singer 1999; Singer and Gray 1995; von der Malsburg 1995). This conjecture is consistent with our findings showing that activation of L1 adjusts the magnitude and reliability of cortical LFPs, indicating changes in neuronal synchronization. Thus activation of L1 regulates the temporal structure and coordination of neuronal responses. Such mechanisms may participate in top-down processing of incoming sensory information in which attention to behaviorally relevant stimuli regulates the degree of synchronized oscillations (Fries et al. 2001; Steinmetz et al. 2000).
|
GRANTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: R. Azouz, Dept. of Physiology, Faculty of Health Sciences, Ben-Gurion University, Beer-Sheva, Israel 84105 (E-mail: razouz{at}bgu.ac.il)
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Anderson JS, Carandini M, and Ferster D. Orientation tuning of input conductance, excitation, and inhibition in cat primary visual cortex. J Neurophysiol 84: 909–926, 2000.
Angelucci A and Bullier J. Reaching beyond the classical receptive field of V1 neurons: horizontal or feedback axons? J Physiol (Paris) 97: 141–154, 2003.[CrossRef][ISI][Medline]
Angelucci A, Levitt JB, Walton EJ, Hupe JM, Bullier J, and Lund JS. Circuits for local and global signal integration in primary visual cortex. J Neurosci 22: 8633–8646, 2002.
Arabzadeh E, Zorzin E, and Diamond ME. Neuronal encoding of texture in the whisker sensory pathway. PLoS Biol 3: e17, 2005.[CrossRef][Medline]
Azouz R. Dynamic spatiotemporal synaptic integration in cortical neurons: neuronal gain, revisited. J Neurophysiol 94: 2785–2796, 2005.
Azouz R and Gray CM. Dynamic spike threshold reveals a mechanism for synaptic coincidence detection in cortical neurons in vivo. Proc Natl Acad Sci USA 97: 8110–8115, 2000.
Bruno RM, Khatri V, Land PW, and Simons DJ. Thalamocortical angular tuning domains within individual barrels of rat somatosensory cortex. J Neurosci 23: 9565–9574, 2003.
Budd JM. Extrastriate feedback to primary visual cortex in primates: a quantitative analysis of connectivity. Proc R Soc Lond Biol Sci 265: 1037–1044, 1998.[Medline]
Bullier J, Hupe JM, James AC, and Girard P. The role of feedback connections in shaping the responses of visual cortical neurons. Prog Brain Res 134: 193–204, 2001.[ISI][Medline]
Cauller LJ. Layer I of primary sensory neocortex: where top-down converges upon bottom-up. Behav Brain Res 71: 163–170, 1995.[CrossRef][ISI][Medline]
Cauller LJ and Connors BW. Synaptic physiology of horizontal afferents to layer I in slices of rat SI neocortex. J Neurosci 14: 751–762, 1994.[Abstract]
Cauller LJ and Kulics AT. The neural basis of the behaviorally relevant N1 component of the somatosensory-evoked potential in SI cortex of awake monkeys: evidence that backward cortical projections signal conscious touch sensation. Exp Brain Res 84: 607–619, 1991.[ISI][Medline]
Chance FS, Abbott LF, and Reyes AD. Gain modulation from background synaptic input. Neuron 35: 773–782, 2002.[CrossRef][ISI][Medline]
Chapin JK and Lin CS. Mapping the body representation in the SI cortex of anesthetized and awake rats. J Comp Neurol 229: 99–213, 1984.
Christophe E, Roebuck A, Staiger JF, Lavery DJ, Charpak S, and Audinat E. Two types of nicotinic receptors mediate an excitation of neocortical layer I interneurons. J Neurophysiol 88: 1318–1327, 2002.
Chu Z, Galarreta M, and Hestrin S. Synaptic interactions of late-spiking neocortical neurons in layer 1. J Neurosci 23: 96–102, 2003.
DeFelipe J and Jones EG. Cajal on the cerebral cortex. New York: Oxford UP, 1998, p. 158 and 208.
Destexhe A and Paré D. Impact of network activity on the integrative properties of neocortical pyramidal neurons in vivo. J Neurophysiol 81: 1531–1547, 1999.
Engel AK, Fries P, and Singer W. Dynamic predictions: oscillations and synchrony in top-down processing. Nat Rev Neurosci 2: 704–716, 2001.[CrossRef][ISI][Medline]
Fairén A, DeFelipe J, and Regidor J. Nonpyramidal neurons: general account. In: Cerebral Cortex, Cellular Components of the Cerebral Cortex, edited by Peters A. and Jones EG. New York: Plenum Press, 1984, vol. 1, p. 201–253.
Felleman DJ and Van Essen DC. Distributed hierarchical processing in the primate cerebral cortex. Cereb Cortex 1: 1–47, 1991.
) and about 200 µm (upper layer 2, ). Time 0 denotes TTX application. Gray box denotes the time interval from which responses were analyzed. E: variability across trials for the same data as in B. Scale bar units are in (µV/mm2)2. F: difference between the CSD profiles of control and TTX in B. Scale bar is in µV/mm2. G: borders of the current sources and sinks taken from the right panel in A (see text). Time and depth scales in E, F, and G are as in C. H: Fourier power spectra of layer 2 LFP traces during spontaneous activity in control (red) and during TTX application (blue).
and 
