Skip to main content
eLife logoLink to eLife
. 2017 Jan 6;6:e21843. doi: 10.7554/eLife.21843

Feedforward motor information enhances somatosensory responses and sharpens angular tuning of rat S1 barrel cortex neurons

Mohamed Khateb 1, Jackie Schiller 1, Yitzhak Schiller 1,2,*
Editor: Sacha B Nelson3
PMCID: PMC5271607  PMID: 28059699

Abstract

The primary vibrissae motor cortex (vM1) is responsible for generating whisking movements. In parallel, vM1 also sends information directly to the sensory barrel cortex (vS1). In this study, we investigated the effects of vM1 activation on processing of vibrissae sensory information in vS1 of the rat. To dissociate the vibrissae sensory-motor loop, we optogenetically activated vM1 and independently passively stimulated principal vibrissae. Optogenetic activation of vM1 supra-linearly amplified the response of vS1 neurons to passive vibrissa stimulation in all cortical layers measured. Maximal amplification occurred when onset of vM1 optogenetic activation preceded vibrissa stimulation by 20 ms. In addition to amplification, vM1 activation also sharpened angular tuning of vS1 neurons in all cortical layers measured. Our findings indicated that in addition to output motor signals, vM1 also sends preparatory signals to vS1 that serve to amplify and sharpen the response of neurons in the barrel cortex to incoming sensory input signals.

DOI: http://dx.doi.org/10.7554/eLife.21843.001

Research Organism: Rat

Introduction

Rodents are equipped with an array of vibrissae on their snout (mystacial vibrissae) that serve as a highly developed tactile somatosensory organ (Woolsey and Van der Loos, 1970; Carvell and Simons, 1990; Petersen, 2007; Diamond et al., 2008; Diamond and Arabzadeh, 2013; Feldmeyer et al., 2013). To probe their environment, rodents typically produce whisking movements of their vibrissae, and palpate objects located within the vibrissae reach. The physical contact between external objects and vibrissae generate forces and vibrations at the base of the vibrissae, which are sensed by specialized mechanoreceptors located at the vibrissae pad. In turn, the sensory information collected by these mechanoreceptors is conveyed to the cortex via the VPM and PoM nuclei of the thalamus (Petersen, 2007; Diamond et al., 2008). The main cortical region receiving sensory information from the vibrissae is the somatosensory S1 barrel cortex (vS1), which is arranged somatotopically, with each vibrissa represented by a separate and discrete barrel-like region (Izraeli and Porter, 1995; Petersen, 2007; Diamond et al., 2008; Bosman et al., 2011; Diamond and Arabzadeh, 2013; Feldmeyer et al., 2013).

The vibrissae-barrel somatosensory system typically uses active sensing (Kleinfeld et al., 2006; Schroeder et al., 2010). Whisking movements are evoked by descending commands from the primary vibrissae motor cortex (vM1) (Grinevich et al., 2005; Gerdjikov et al., 2013; Hill et al., 2011; Petersen, 2014). It is still unclear whether neurons in vM1 directly drive the facial nucleus, or alternatively activate a central pattern generator (CPG) in the brainstem that in turn rhythmically drives the facial nucleus (Grinevich et al., 2005Kleinfeld et al., 2014; Moore et al., 2014Petersen, 2014). In addition to the descending output motor information to the brain stem, vM1 also directly sends information to other structures in the sensory-motor vibrissae-barrel loop, and especially the vS1 barrel cortex (Aronoff et al., 2010; Veinante and Deschênes, 2003; Mao et al., 2011; Feldmeyer et al., 2013). This direct vM1 to vS1 pathway originates mostly from layer 2–3 and layer 5A pyramidal neurons in vM1 and terminates mostly on layer 5 neurons and to a lesser degree layer 2–3 neurons of vS1 (Mao et al., 2011). Previous functional studies have shown that vM1 inputs to vS1 carry information regarding both motor parameters of whisking, as well as more complex sensory information (Petreanu et al., 2012). Moreover, inputs from vM1 have been shown to modify the network state of the vS1 barrel cortex, and increase the coding reliability of complex sensory stimuli in vS1 (Zagha et al., 2013).

In this study, we aimed to investigate the effect of vM1 inputs on sensory processing in the barrel vS1 cortex. A major obstacle we encountered in tackling this question was the fact the vibrissae-barrel system functions as a motor-sensory loop (Kleinfeld et al., 1999; Diamond et al., 2008), and thus, vM1 can influence vS1 neurons both directly and indirectly. For example, vM1 activation can affect vS1 neuron both via direct inputs connecting the two areas and indirectly via whisking-evoked stimulation of mechanoreceptors in the vibrissae pad. To tackle this problem, we disconnected the motor-sensory loop of the vibrissa-barrel system, and dissociated activation of the motor and sensory components. We passively activated the principal vibrissa by piezo-mediated vibrissa movements or artificial whisking against sandpaper (Szwed et al., 2003; Garion et al., 2014), and independently activated vM1 with optogenetic stimulation. In all our experiments, we cut the buccolabialis branch of the facial nerve to eliminate cortical-driven whisking movements. Using this experimental paradigm, we aimed to investigate the effect vM1 activation on the response of vS1 neurons to vibrissa stimulation; characterize temporal interactions between vM1 and vS1; and examine the effect of vM1 on angular tuning of vS1 neurons.

Results

Pairing optogenetic vM1 activation with passive vibrissa activation

The goal of this study was to investigate the effect of vM1 on sensory processing in the vS1 barrel cortex. Under physiological conditions, the barrel-vibrissae system functions as a loop, and thus vM1 can affect the vS1 barrel cortex both directly and indirectly. To address this obstacle, we passively stimulated the principal vibrissa, and independently activated vM1 neurons using optogenetic stimulation. In the experiments we passively deflected, the principal vibrissa (Typically B2) with piezo-mediated 200 ms ramp and hold vibrissa deflection (Bruno et al., 2003; Lavzin et al., 2012), while concomitantly recording single-unit activity from neurons in the vS1 barrel cortex. We recorded the single-unit activity from neurons in different layers of the vS1 using single-shaft 16 channel silicon probe electrodes (inter contact distance of 50 µm; contacts located at depth of 300–1100 µm from the pia) (Figure 1A). The principal barrel was identified prior to electrode insertion using intrinsic imaging (for details see Garion et al., 2014), and verified by manual deflection of vibrissae during extracellular multi-unit recordings.

Figure 1. Optogenetic vM1 activation and passive ramp and hold vibrissa deflection.

(A) Scheme of the experimental design with the recording electrode in vS1, optogenetic stimulation of vM1 and ramp and hold passive vibrissa deflection. (B) Peri-stimulus histograms (PSTH, mean ± SEM) recorded from a vS1 neuron during isolated vM1 optogenetic stimulation (blue), isolated passive vibrissa deflection (green), and paired vM1-vibrissa stimulation (red). In addition, the expected linear sum of the solitary vM1 stimulation and vibrissa deflection is shown (gray). (C) Fluorescence images at different magnifications of neurons co-expressing ChR2 and mCherry. (D) PSTH recorded from a vM1 neuron during optogenetic stimulation (473 nm laser pulse). (E) Average (mean ± SEM) responses recorded in vS1 neurons during optogenetic stimulation of vM1 with laser pulse duration (147 neurons in four rats). In each neurons, the spike count responses during optohgenetic stimulation were presented as percent of the spike count during the pre-stimulus control value. Later, the responses were averaged over the different neurons.

DOI: http://dx.doi.org/10.7554/eLife.21843.002

Figure 1.

Figure 1—figure supplement 1. Raw data of unit recordings.

Figure 1—figure supplement 1.

Four separate traces of raw recording filtered at 1–5 kHz. Each trace shows simultaneous recordings from eight different channels.
Figure 1—figure supplement 2. Raw clustering data.

Figure 1—figure supplement 2.

The raw clustering data is presented for ten individual electrodes, which yielded 26 individual sorted units. For each electrode, we present the putative layer, the 2D principal components graph, the mean ± SD of each unit and the clustering statistical parameters.
Figure 1—figure supplement 3. Piezo bimorph movement.

Figure 1—figure supplement 3.

The movement of the piezo bimorph was monitored during a ramp and hold stimulation pulse using a high-speed camera (1000 fps). Panel A shows individual traces of piezo bimorph movement after applying different values to the smoothening factor (0.03–1). Panel B presents the average amplitude of the maximal ringing (measured from peak to turf) with the different smoothening factor values.

Typically, the 200 ms passive ramp and hold vibrissa deflection-evoked short on and off responses (Figure 1B). For our analysis, we measured the additional number of spikes (spike count) evoked by the stimulus compared to the pre-stimulus control value during the on (initial 100 ms of the ramp and hold stimulation) and off responses (initial 100 ms after the end of the ramp and hold stimulation). In addition, we independently activated neurons in vM1 cortex with optogenetic stimulation. We expressed ChR2 in vM1 pyramidal neurons by local injection of AAV viral vectors containing the ChR2 gene expressed under the control of the CaMKII promotor (Figure 1C), and ChR2 expressing neurons in vM1 were activated with 20 ms laser pulses (Figure 1D; for details see Materials and methods).

Extracellular recordings from vM1 during optogenetic stimulation showed that short 473 nm laser pulses effectively activated vM1 neurons (Figure 1D). Overall 84% of all recorded neurons in vM1 significantly increased their firing rate during 20 ms 473 nm laser pulses (109 vM1 neurons from five rats). On average vM1 neurons increased their average firing by 563% ± 84% during the pulse time. Moreover, the average latency between laser onset and increased firing in vM1 was 4.81 ± 0.43 ms (mean ± SEM, 109 neurons from five rats), with 41% of neurons showing a unimodal response.

When we recorded from vS1 neurons during optogenetic activation of vM1 neurons, we found small responses in vS1 barrel neurons (Figure 1B,E, and Figure 2). The magnitude of the responses recorded in vS1 neurons depended on the duration of the laser pulse, and gradually increased as the duration of the laser pulse increased from 5 to 20 ms (Figure 1E). Based on this data, we used 20 ms laser pulses for all our experiments.

Figure 2. Supra-linear summation of paired optogenetic vM1 activation and passive ramp and hold vibrissa deflection-averaged results.

Figure 2.

Average (mean ± SEM) single-unit response (presented as percent above the pre-stimulus control activity) of vS1 neurons to the On response during the different stimulation conditions, isolated vM1 optogenetic stimulation, isolated passive vibrissa deflection, expected linear sum of the solitary vM1 stimulation and vibrissa deflection and paired vM1-vibrissa stimulation presented for single-unit analysis (A) and multi-unit analysis (B), (237 units from 91 electrode contacts in seven rats). Note that the recorded activity during paired vM1-vibrissa stimulation was significantly larger than the linear sum of the response to vM1 activation and passive ramp and hold vibrissa deflection applied separately for both single- and multi-unit analyses. (C) Average (mean ± SEM) supra-linearity of the paired vM1-vibrissa stimulation response (measured/expected linear sum) presented for all neurons and for neurons in the different putative neocortical layers (2–3, 4 and 5). Thirty-five neurons from putative layers 2–3, 79 neurons from putative layer 4 and 123 neurons in putative layer 5. ***p<0.001 and **p<0.01 using the student's t-test.

DOI: http://dx.doi.org/10.7554/eLife.21843.006

To investigate the effect of vM1 activation on sensory processing in vS1 neurons, we applied three different stimulation conditions in random order. First, isolated vM1 optogenetic stimulation; second, isolated ramp and hold vibrissa deflection; third, paired optogenetic vM1 activation with passive vibrissa deflection. When we paired optogenetic activation of vM1 with passive ramp and hold deflection (paired vM1-vibrissa stimulation), vM1 activation supra-linearly amplified the responses in vS1 neurons. During the on response, paired vM1-vibrissa stimulation were significantly larger (by an average of 39%) than the expected linear sum of the two stimuli applied separately (Figure 2A, 237 neurons from seven rats). We also confirmed these findings with multi-unit analysis (Figure 2B). In the case of multi-unit analysis the on response to paired ramp and hold vibrissa deflection and optogenetic vM1 stimulation was 34% larger than the expected sum of the two stimuli applied separately. Moreover, we found that in 57% of individual neurons (135 out of 237 neurons) paired vM1-vibrissa stimulation was significantly larger than the expected sum of solitary vibrissa deflection and vM1 optogenetic activation.

Interestingly, a small yet significant supra-linearity was also observed during the off response despite the fact we applied the 20 ms optogenetic stimulation at the onset of the 200 ms ramp and hold vibrissa deflection. Paired vM1-vibrissa deflection stimulation yielded an average off response that was 15% larger than the expected sum of the two stimuli applied separately (expected average linear sum of 207.6% ± 11.7% compared to 238% ± 12.4% of the averaged measured paired vM1-vibrissa stimulation, p<0.05, 237 units from seven rats).

We also investigated the effect of pairing vM1 activation with passive ramp and hold vibrissa deflection in the different cortical layers of the vS1 barrel cortex. Our recordings extended between 300–1100 µm from the cortical surface. We divided the recorded neurons according to their putative cortical layers, based on their distance from the pia. We defined putative layers 2–3 as extending up to 600 µm from the pia, layer 4 as extending 650–800 µm from the pia, and layer 5 as extending beyond 800 µm from the pia. It is important to stress that cortical layers were determined solely by the distance of the recording contact from the cortical surface, as we could not verify cortical layers by histological criteria. Paired opto-stimulation of vM1 supra-linearly amplified the on response vS1 neurons to ramp and hold vibrissa deflection in all cortical layers recorded. Supra-linearity was largest in putative layers 2–3 and 5, and smallest in putative layer 4 neurons. No significant differences were observed between layers 2–3 and 5 (Figure 2D, analysis for the different layers is shown for the on response).

We further investigated the effect of vM1 activation on the response to vibrissa activation using a second passive stimulation paradigm, artificial whisking against sandpaper. With this passive stimulation paradigm, we cut the buccolabial branch of the facial nerve and stimulated the severed branch to generate artificial whisking movements at 5.5 Hz while contacting sandpaper (Szwed et al., 2003; Bagdasarian et al., 2013; Garion et al., 2014) (Figure 3A). vM1 neurons were optogenetically activated in a similar manner to the ramp and hold vibrissa deflection experiments. For our analysis, we measured the additional number of spikes (spike count) evoked by the protraction phase of artificial whisking (90 ms) as compared to the pre-stimulus baseline. We chose to concentrate on the earlier protraction phase as we opto-stimulated vM1 at the onset of artificial whisking.

Figure 3. Supra-linear summation of paired optogenetic vM1 activation and artificial whisking against sandpaper.

Figure 3.

(A) Scheme of the experimental design with the recording electrode in vS1, optogenetic stimulation of vM1 and artificial whisking against sandpaper by electrically stimulating the buccolabialis nerve. The timing of individual nerve stimulation and protraction phase is marked in black. (B) Peri-stimulus histograms (PSTH, mean ± SEM) recorded from a vS1 neuron during solitary vM1 optogenetic stimulation (blue), solitary artificial whisking against sandpaper (green), and paired vM1-vibrissa stimulation (red). In addition, the expected linear sum of the solitary vM1 stimulation and artificial whisking is shown (gray). (C) Average (mean ± SEM) responses (presented as percent above the pre-stimulus control activity) of vS1 neurons to the different stimulation conditions (294 neurons from six rats). Note that the recorded activity during paired vM1-vibrissa stimulation was significantly larger than the linear sum of the response to solitary vM1 activation and artificial whisking. ***p<0.001 with the paired student's t-test. (D) Average (mean ± SEM) supra-linearity of the paired vM1-vibrissa stimulation response (measured/expected linear sum) presented for all neurons and for neurons in the different putative neocortical layers, 76 neurons from putative layers 2–3, 80 neurons from putative layer 4 and 138 neurons in putative layer 5.

DOI: http://dx.doi.org/10.7554/eLife.21843.007

Artificial whisking against sandpaper-evoked spikes during the protraction and retraction phases (Figure 3B). Similar to the ramp and hold stimulation paradigm optogenetic co-activation of vM1 supra-linearly amplified the response of vS1 neurons to artificial whisking against sandpaper. On average, the response evoked by paired vM1 optogenetic activation with artificial whisking against sandpaper was 42% larger than the expected linear sum of the two individual responses applied separately (Figure 3B,C). When we calculated the ratio of the measured versus expected responses for individual neurons we found that 69% of neurons (164 out of 241 neurons) had measured paired vM1-vibrissa stimulation that were significantly larger than the expected linear sum of isolated vibrissa deflection and vM1 optogenetic activation. Similar to ramp and hold vibrissa stimulation, pairing vM1 activation with artificial whisking supra-linearly amplified the response in all recorded neocortical layers (Figure 3D).

Temporal roles governing the interactions between vM1 and vibrissa sensory inputs

In contrast to physiological conditions, our experimental paradigm allowed us to activate the vibrissae and vM1 in a temporally independent manner. We utilized this capability to investigate the temporal rules governing the effect vM1 activation on the response of vS1 barrel neurons to passive vibrissa stimulation. In these experiments, the vibrissa and vM1 were stimulated independently with different time delays ranging between −100 ms and +50 ms (time difference calculated by subtracting the onset time of vibrissa stimulation from the onset time of the laser pulse, namely the onset of vibrissa stimulation was defined as time 0 ms). For these experiments, we passively stimulated the principal vibrissa by either ramp and hold vibrissa deflection (Figure 3A) or artificial whisking against sandpaper (Figure 3B). When vM1 was optogenetically activated 0–50 ms prior to passive vibrissa stimulation the paired response was significantly larger than that evoked by isolated passive vibrissa stimulation (Figure 3). In contrast, when vM1 was activated 75–100 ms before or 20–50 ms after passive vibrissa stimulation the combined response did not significantly differ from the isolated vibrissa stimulation. The largest facilitating effect of vM1 activation was observed when the onset of the optogenetic vM1 activation preceded the onset of vibrissa stimulation by 20 ms (Figure 3). These findings were observed for both ramp and hold vibrissa deflection and artificial whisking against sandpaper (Figure 3A,B).

It is important to stress that in our experiments the time difference between vM1 optogenetic activation and vibrissa stimulation was in fact the time difference between the onset of vibrissa stimulation and onset of optogenetic stimulation, without taking into account transmission and synaptic delays, nor the exact timing of slip and stick events for artificial whisking.

The effect of vM1 activation on angular tuning of neurons in the vS1 barrel cortex

Previous studies have reported that neurons in the vS1 barrel cortex show a preference for the direction in which the vibrissa is deflected (angular tuning) (Bruno et al., 2003; Andermann and Moore, 2006; Kremer et al., 2011; Lavzin et al., 2012). In the previous section, we have shown that paired vM1 activation supra-linearly enhances the response of vS1 barrel cortex neurons to vibrissa stimulation. We next set out to examine whether in addition, paired vM1 activation also effects angular tuning of neurons in the vS1 barrel cortex. In these experiments, we recorded the response of vS1 barrel neurons to passive ramp and hold deflection of the principal vibrissa to eight different directions with and without paired optogenetic activation of vM1. vM1 optogenetic activation was applied at the optimal temporal time window (laser onset preceding vibrissa deflection by 20 ms). In these experiments, we randomly alternated the direction of vibrissa deflection and whether the laser was activated or not. Consistent with previous results, we found that angular tuning in neurons is located in layers 2–3 and layer 4 (Bruno et al., 2003; Andermann and Moore, 2006; Kremer et al., 2011; Lavzin et al., 2012). We further found that layer 5 neurons in the vS1 barrel cortex also show similar angular tuning. We quantified the percent of neurons with significant angular tuning, and found that 62% ± 8% of all recorded neurons showed significant angular tuning, with no significant differences between putative layers (significant angular tuning was defined by significant difference at the 0.05 level for comparison of the responses to the preferred angle and to the three least preferred angles).

When we paired optogenetic stimulation of vM1 with vibrissa deflection to different angles, we found that the angular tuning of neurons was sharpened. Sharpening of angular tuning by optogenetic activation of vM1 is demonstrated in six different individual vS1 neurons (Figure 5, Figure 5—figure supplement 1). To quantify the effect of vM1 activation on angular tuning of barrel vS1 neurons, we calculated the selectivity index (SI) of angular tuning of the preferred direction (maximal response in the preferred angle/average response to all angles) for each neurons with and without vM1 optogenetic activation. In these experiments, we found that paired vM1 optogenetic activation significantly increased the average SI for angular tuning in vS1 barrel cortex neurons (Figure 6A), which resulted from a right shift of the SI value histogram by vM1 activation (Figure 6B). Interestingly, we observed a significant increase in the SI of the preferred direction in both 50–60 day old and 90–100 day old rats (Figure 6—figure supplement 1A). Furthermore, as expected, while the SI of the preferred direction increased, the SI values of the worst three angular directions decreased by vM1 activation, although this reduction did not reach statistical significance (Figure 6—figure supplement 1B). It is important to stress that although the SI to the worst directions decreased, in many cases the activation of vM1 increased the absolute response to the worst angles (see examples in Figure 5, and Figure 5—figure supplement 1).

Figure 5. Sharpening of the angular tuning of vS1 by paired vM1 activation: examples of individual neurons.

Polar plots of six individual vS1 neurons during isolated ramp and hold vibrissa deflection (blue) and paired vM1-vibrissa stimulation (red). In these experiments, the principal vibrissa was randomly deflected to eight different directions (0°,43°,90°, 135°, 180°, 225°,270°, 315°) with and without paired vM1 optogenetic activation. Vibrissae were deflected in the different directions by two pairs of galvanometers. Upper four panels are from 50–60 day old rats, and the lower two panels are from 90–100 day old rats. Upper left panel-putative layer 2–3, upper right panel putative layer 5, middle left panel-putative layer 4, middle right panel putative layer 5, lower left panel-putative layer 5, lower right panel putative layer 2–3. Note, sharpening of the angular tuning curve following paired vM1 optogenetic activation.

DOI: http://dx.doi.org/10.7554/eLife.21843.009

Figure 5.

Figure 5—figure supplement 1. The effect of vM1 activation on the angular tuning of individual neurons.

Figure 5—figure supplement 1.

The response (mean ± SEM) of six individual neurons to ramp and hold vibrissa deflection in eight different angles under control conditions and during optogenetic co-activation of vM1. The neurons are the same neurons presented in Figure 5. In this figure, we present the data in conventional graphs rather than polar plots, and add the variability of the responses (SEM).

Figure 6. Sharpening of the angular tuning of vS1 by paired vM1 activation: averaged results.

(A) The average (mean ± SEM) ratio between the SI calculated for the preferred angular direction for isolated vibrissa deflection (red) and paired vibrissa deflection with vM1 optogenetic activation (blue). The results are presented for single- (SUA) and multi-unit (MUA) analysis. The results are shown for all neurons examine (304 neurons from 11 rats). Note the significant increase in the SI following pairing with vM1 activation for moth SUA and MUA. (B) The SI magnitude histogram of unit for isolated for isolated vibrissa deflection (red) and paired vibrissa deflection with vM1 optogenetic activation (blue). Note that paired vM1 optogenetic stimulation resulted a right shift of the histogram. (C) The effect of vM1 optogenetic activation on the average amplitude (mean ± SEM) of the vector sum. Vector sum analysis was performed on the same data set presented in panels A and B (204 neurons from 11 rats). **p<0.01. (D) The average (mean ± SEM) ratio between the SI calculated for isolated vibrissa deflection and paired vibrissa deflection with vM1 optogenetic activation. The results are shown for all neurons examine, and for the different putative cortical layers (25.7% putative layer 2–3 neurons, 20% putative layer 4 neurons and 54.3% putative layer five neurons). *p<0.05, **p<0.01. In cases the SI for isolated vibrissa deflections were compared with paired vibrissa deflections and vM1 optogenetic activation using the paired student's t-test. (E) Percent of neurons that retained the same SI with and without vM1 optogenetic activation as a function of the control SI value (SI < 1.5, SI = 1.5–2 and SI > 2). (F) The average (mean ± SEM) ratio between the SI value recorded without (SIcont) and with (SIlaser) vM1 optogenetic activation in all recorded neurons and in neurons which retained the preferred angle with and without optogenetic activation (stable), and in neurons that changed their preferred angle after vM1 activation (unstable). **p<0.01.

DOI: http://dx.doi.org/10.7554/eLife.21843.011

Figure 6.

Figure 6—figure supplement 1. The effect of vM1 activation on angular tuning.

Figure 6—figure supplement 1.

(A) The effect of optogenetic activation of vM1 on the SI of the preferred direction in 50–60 day old (left) and 90–100 day old mice (right). (B) The effect of optogenetic activation of vM1 on the SI of the preferred direction and of the worst three directions. **p<0.01, *p<0.05.

To overcome potential technical problems associated with single-unit analysis, we repeated our analysis for angular tuning using multi-unit activity. Similar to single- unit activity multi-unit activity, again demonstrated that vM1 optogenetic activation significantly sharpened angular tuning of vS1 neurons (Figure 6A). The amplificatory effect of vM1 activation. The fact we observed sharpening of angular tuning with multi-unit analysis is consistent with the previously reported spatial mapping of angular tuning in the barrel cortex (Kremer et al., 2011).

To further quantify the effect of vM1 optogenetic activation on angular tuning, we calculated the vector sum of the responses to vibrissae stimulation in the eight different angles (see Materials and methods for details) with and without vM1 optogenetic activation. We found that similar to the effect on SI, vM1 optogenetic activation significantly increased the amplitude of the vector sum (Figure 6C). On average, the amplitude of the vector sum increased by 29% ± 7%.

A significant sharpening of angular tuning by paired vM1 activation was observed in all recorded neocortical layers (layers 2–5), with no significant differences between the recorded putative layers (Figure 6D). The majority of neurons retained their preferred angular direction during opto-stimulation of vM1. The probability for retaining the preferred angular direction during paired vM1 activation was larger in neurons with higher SI during solitary unpaired vibrissa stimulation, with 70.9% of neurons with SI greater than two retaining their preferred angular direction after paired vM1 activation (Figure 6E). Moreover, neurons that retained their preferred angular direction showed a significantly greater degree of sharpening of their angular tuning curve by vM1 optogenetic activation, as compared to neurons that changed their preferred angular tuning during paired vM1 activation (Figure 6F).

Discussion

In this study, we set out to investigate the direct effects of vM1 activation on sensory processing in neurons of the vS1 barrel somatosensory cortex. The main findings of this study include: (1) vM1 optogenetic activation supra-linearly amplified the response of vS1 barrel neurons to passive vibrissa stimulation. This supra-linear amplification effect of vM1 activation was observed in all neocortical layers recorded (layers 2–5 300–1100 µm from the pia), and recurred in two different passive stimulation paradigms, ramp and hold vibrissa deflection and artificial whisking against sandpaper. (2) The maximal effect of vN1 on the response of vS1 neurons to vibrissa sensory inputs occurred when the onset of vM1 activation preceded vibrissa activation by 20 ms. Smaller yet significant effects were also observed when the onset of vM1 activation preceded vibrissa stimulation by 50 ms or when vibrissa stimulation and vM1 activation occurred simultaneously. This temporal relationship recurred for the two passive stimulation paradigms we examined. The physiological time delay between vM1 and vS1 activation is not fully known. Yet previous studies have shown that the activity of both vM1 and vS1 is phased locked with exploratory rhythmic whisking movements in rats (Ahrens and Kleinfeld, 2004), and vM1 neurons are probably critical in rhythmically driving whisking (Carvell et al., 1996; Kleinfeld et al., 2002; Ahrens and Kleinfeld, 2004; Brecht, 2004; Brecht et al., 2004) (3). We found that in addition to supra-linearly amplification of the response of vS1 barrel neurons to vibrissa stimulation, activation of vM1 also significantly sharpened the angular tuning in these neurons.

vM1 connections to the vS1 barrel cortex

Previous studies have shown that vM1 neurons directly innervate neurons in the vS1 barrel cortex (Izraeli and Porter, 1995; Veinante and Deschênes, 2003; Diamond et al., 2008; Aronoff et al., 2010; Mao et al., 2011; Feldmeyer et al., 2013). The strongest monosynaptic connections between vM1 and vS1 exist between neurons in layers 2–3 and layer 5A of vM1 and layer 5A and 5B neurons in the vS1 barrel cortex (Mao et al., 2011). Interestingly, reciprocal connections also exist between layers 2–3 and 5A neurons in vS1 and vM1 neurons (Hooks et al., 2013). A subset of vM1 axons reach layer 1 of the vS1 barrel cortex, and carry both motor information regarding different vibrissa movement parameters and more complex sensory information regarding contact of vibrissae with objects, and spatial localization of objects (Petreanu et al., 2012). Moreover, vM1 inputs innervating tuft dendrites of vS1 layer 5 pyramidal neurons critically participate in dendritic spike initiation when vibrissae actively touched objects in a location and angle-dependent manner (Xu et al., 2012).

In this study, we show that short (20 ms) optogenetic stimulation of vM1 neurons significantly increased firing of vS1 barrel neurons. These findings are in agreement with the results of a recent study performed in awake behaving rodents that reported activation of vS1 barrel neurons by optogenetic stimulation of vM1 neurons (Zagha et al., 2013). Moreover, Zagha et al. (2013) also reported that optogenetic activation of vM1 resulted in context-dependent changes in the network state of the barrel cortex and increased reliability of responses to complex sensory stimuli.

Although previous studies have established the existence of direct connections between vM1 and vS1, and have demonstrated functional implications of this pathway (Petreanu et al., 2012; Xu et al., 2012; Zagha et al., 2013), our study adds novel yet unknown information regarding the effects of vM1 inputs on vS1 barrel neurons. First, we show that motor information from vM1 neurons and sensory information from the vibrissae summate supra-linearly in vS1 barrel neurons. Second, we defined the temporal rules governing the interactions between incoming vM1 and vibrissa sensory input information in vS1 neurons. We found maximal vM1-vibrissa interactions occurred when vM1 activation preceded vibrissa stimulation by 20 ms. This optimal temporal window for vM1-vibrissa input interactions suggested that vM1 sent primarily feedforward information regarding the planned vibrissa movements, rather than feedback information originating in the sensory motor loop. Finally, we show that vM1 activation, not only amplified the response of vS1 barrel neurons to incoming vibrissa sensory inputs, but also sharpened their angular tuning curve. Taken together, it seems vM1 inputs participated in increasing both the sensitivity and specificity of barrel cortex neurons to incoming sensory information from the vibrissae. The direct effect of vM1 on vS1 neurons may underlie the difference between passive and active sensing in the barrel vibrissa somatosensory system.

Possible mechanisms underlying vM1-mediated response amplification and sharpening of angular tuning in vS1 barrel neurons

In our study, we did not directly address the cellular mechanisms underlying the effects of vM1 inputs on the response of vS1 neurons to incoming vibrissa stimulation. Several different potential cellular and network mechanisms can explain our findings. First, is a non-linear effect on the axonal initiation zone. Specifically with this mechanism, excitatory vM1 inputs depolarize vS1 neurons, and as a result, concomitant vibrissae-evoked EPSPs will generate a larger number of action potentials in the axonal initiation zone. Second, are non-linear dendritic amplification mechanisms. Previous studies have shown that dendrites of neocortical excitatory neurons can support generation of local dendritic sodium, calcium and NMDA spikes (Stuart et al., 1997; Larkum et al., 1999, 2009; Schiller et al., 2000; Polsky et al., 2004London and Häusser, 2005; Branco et al., 2010; Lavzin et al., 2012; Xu et al., 2012; Harnett et al., 2013; Major et al., 2013; Smith et al., 2013; Palmer et al., 2014; Cichon and Gan, 2015). It is possible that pairing vM1 inputs with inputs carrying vibrissa sensory information results in initiation of local dendritic spikes vS1 neurons. Consistent with this possibility are the results of several previous in vivo studies that reported initiation of dendritic spikes in tuft dendrites of vS1 layer-5 pyramidal neurons in response to vM1 and M2 activation (Xu et al., 2012; Manita et al., 2015). Third, are local network mechanisms. For example, vM1 inputs may selectively innervate VIP inhibitory inter neurons, as described for the medial prefrontal inputs to the auditory cortex (Lee et al., 2013; Pi et al., 2013). In this case, vM1 inputs will activate VIP inter neurons, which will inhibit other inhibitory inter neurons, and in turn secondary excite pyramidal neurons in vS1.

Possible functional significance of vM1 mediated response amplification and sharpening of angular tuning in vS1 barrel neurons

The barrel vibrissa system uses active sensing to palpate objects in the near vicinity of the rodents head. Typically, vM1, the motor cortex of the barrel vibrissa system, is responsible for generating the whisking movements of the mystacial vibrissae. vM1 can drive whisking either by directly activating lower motor neurons in the facial nucleus, especially during protraction movements, or alternatively, by activating a brain stem CPG located in close proximity to or within the Botzinger complex. In the latter case the CPG drives the facial nucleus to generate rhythmic movements (Grinevich et al., 2005; Haiss and Schwarz, 2005; Gerdjikov et al., 2013; Moore et al., 2013; Petersen, 2014; Sreenivasan et al., 2015). In addition to the brainstem output motor commands, vM1 also directly innervates the vS1 barrel cortex. These inputs convey both motor parameters of whisking, as well as more complex sensory information regarding contacting the object and object localization (Petreanu et al., 2012; Harnett et al., 2013).

Functionally, there are several scenarios for how the vM1 to vS1 pathway participates in sensory computations. First, vM1 inputs can convey a simple ‘attentional cue’ to prepare vS1 barrel cortex neurons for the incoming sensory information from the vibrissae. Regarding this possibility, it is interesting to note that previous studies have shown that increased attention and especially activation of the frontal eye field area increases the sensitivity and specificity of cortical neurons in sensory visual regions to different visual stimuli (McAdams and Maunsell, 1999; Reynolds and Chelazzi, 2004; Noudoost et al., 2010). The second possibility is that vM1 inputs convey a motor efference copy that can be used for computing the location as well as shape and texture of objects by comparing the expected (vM1) with the measured (thalamo-cortical) vibrissa movements (Diamond et al., 2008; Curtis and Kleinfeld, 2009; Kleinfeld and Deschênes, 2011). Third, multiple inputs converge onto the barrel motor cortex, thus vM1 inputs can serve to convey top-down dynamical modifications of sensory processing in the primary vS1 cortex (Xu et al., 2012; Zagha et al., 2013; Manita et al., 2015), see also Squire et al. (2013) regarding the visual system). Regardless of the exact computational role vM1 inputs play in sensory processing in vS1 neurons, the significant effects vM1 inputs have on vS1 neurons may underlie the importance of active sensing in the somatosensory system. Further studies in awake behaving rodents are needed to distinguish between these three possibilities and decipher the functional role of M1 inputs in sensory processing in somatosensory processing.

Materials and methods

Surgical preparation for recording in anesthetized rats

We conducted experiments in accordance with NIH and institutional standards for the care and use of animals in research, and received the approval of our institutional animal ethics committee (protocol 007-01-2014). P50-60 Wistar rats were anesthetized by intra-peritoneal injection of Urethane (20% dissolved in normal saline). Prior to surgery lidocaine (2%) was applied locally over the scalp, and the buccolabialis branch of the facial nerve, which innervates muscles in the vibrissae pad, was exposed and severed at its proximal segment. The skull was exposed and well cleaned in order to identify important anatomical landmarks such as the midline, bregma and lambda. A craniotomy (2–3 mm2) was drilled over the vS1 barrel cortex (2.5 mm posterior to bregma, 4.5 mm lateral to the midline), and a well surrounding the craniotomy was constructed using dental cement. In addition, a short metal pole was glued to the skull rostral to the dental cement well and later used to hold the rats head in place. After the dental cement was constructed, the dura matter was carefully removed over a small area (less than 1 mm2) to allow access to the cortex. We minimized the area of exposed cortical surface in order to reduce brain pulsation and damage during the hours of the recordings. The well surrounding the craniotomy was filled with aCSF to cover the exposed neocortical surface to prevent the brain from drying. Body temperature was carefully maintained at 36–37°C using a heating pad (FHC, Montana, USA).

Optical intrinsic imaging

To identify the principal barrel, we initially performed optical intrinsic imaging as previously described (Lavzin et al., 2012; Garion et al., 2014). The cortical surface was illuminated and imaged through the thinned skull. We used 650 nm LED to illuminate the cortex, and light absorbance images were acquired with a Q-cam CCD camera (QImaging, British Columbia, Canada). Individual vibrissae were deflected using a pair of galvanometers at 10 Hz for a 2 s duration, to identify the location of the barrel representing the stimulated vibrissa. Sensory stimulation and data acquisition were controlled via an isolated pulse stimulator (model 2100, A-M systems, Washington, USA) using custom home made software written in Matlab (MathWorks, Massachusetts, USA). The location of the principal barrel was later used to guide electrode insertion.

Electrophysiologcal unit recordings

Electrophysiological recordings were performed simultaneously from multiple neurons in layers 2–5 of the vS1 barrel cortex using silicone multi-contact probes (NeuroNexus, Michigan, USA). Our silicone probes (A 1 × 16 MEA) were composed of a single-shaft electrode with 16 recording contacts arranged along a vertical line and separated by 50 micrometers. Electrodes were inserted to pre-mapped principal barrel using a stereotactic micro-manipulator (TSE, Bad Homburg, Germany). Electrodes were slowly lowered until the electrode tip reached approximately a depth of 1000–1100 µm from the pial surface. After the electrode was inserted into the barrel cortex, we verified the identity of the principal vibrissa by recording multi-unit activity during manual deflection of the vibrissa. During the experiments, we trimmed all vibrissae aside from the principal vibrissa.

Electrophysiological data was acquired with the 16-channel ME-16 system and MC Rack software (Multi channel systems, Reutlingen, Germany). The recorded data were initially amplified (X1000), filtered at 0–25 kHz and stored in the computer, and the analysis was performed primarily offline. In addition, to allow for online monitoring of the unit activity during the experiments the data was filtered online at 1–5 kHz and displayed. For the offline analysis, the raw recorded data were replayed and filtered at 1–5 kHz to obtain the unit activity (Figure 1—figure supplement 1). Later single-units were sorted using the OFS offline spike sorter (Plexon, Texas, USA), and the sorted spike trains were further analyzed using the Neuroexplorer software (Nex Technologies, Alabama, USA) and home written software in MatLab (MathWorks, Massachusetts, USA). The results were presented as peri-stimulus histograms (PSTH), and the increase in spike count during stimulation, defined as the additional number of spikes evoked during stimulation stimuli above the number of spikes during the control pre-stimulus baseline (Spike count during stimulation-Spike count during a similar baseline control).

For offline sorting, we initially detected events with an amplitude >3.5 SD of the baseline value. These events served for multi-unit analysis (MUA). For single-unit analysis, we further sorted the thresholded events using semi-automatic clustering algorithms, followed by manual verification and correction of these clusters, if needed. Clusters were accepted as single-units if all the following criteria were met: (1) the waveform shape remained consistent and stable throughout recording (verified by the ‘Sort-Quality Vs Time’ analysis in the OFS software). Moreover, units were excluded in case the average amplitude or half width of unit changed significantly (ANOVA test) between the initial and last 20% of recorded spikes. (2) Firing rate was >0.5 Hz to allow for adequate sampling. (3) Inter-spike interval (ISI) was >2 ms to reflect the absolute refractory period of neurons. (4) ISI distribution showed a smooth exponential-like curve. (5) Finally and most importantly, statistical criterion of p<0.05 (multivariate ANOVA) of cluster separation. Figure 1—figure supplement 2 shows sorting parameters of 10 individual units. The average signal to noise (SNR) value of our units was 11.2 ± 0.4 (range of 7–15, Rousche et al., 1999)

In addition, we performed cross-correlation analysis between units recorded for all adjacent contact pairs, and excluded the unit from one of the electrodes (with the smaller amplitude) in case the peak of the cross-correlation was >0.9 (2 ms time bin) to avoid recording of the same unit with two contacts. In addition, we excluded all units that showed >0.9 value of the cross-correlogram in more than one electrode pair to exclude noise.

All the averaged results are presented as the mean±SEM values, and statistical testing was performed using the paired and unpaired student's t-test. The source code for the Matlab homemade software are enclosed in supplementary files (Source code 1).

To verify the recording location, at the end of the experiment a fluorescent dextran (florescent dextran-A solution of 2 mM fluorescent dextran Alexa-488 or Texas Red; Invitrogen, USA) was injected into the electrode tract using a pressure injector. Later the rat was sacrificed, and trans-cardially perfused with 4% Para-Formaldehyde for histological processing.

Vibrissa stimuli

We used three different passive stimulation paradigms applied to the principal vibrissa:

  1. Artificial whisking paradigm as previously described (Derdikman et al., 2006; Garion et al., 2014). In this stimulation paradigm, we exposed and stimulated the buccolabialis branch of the facial nerve to induce artificial whisking movements. The buccolabialis nerve was cut, and its distal end mounted on a pair of bipolar tungsten electrodes. We applied a train of 10 protraction-retraction cycle applied at 5.5 Hz. For each protraction. Ten bipolar rectangular electrical pulses (0.5–4.0 V, 40 µs duration) were applied through an isolated pulse stimulator (A360, WPI) at 100 Hz to produce vibrissa protraction, followed by a passive vibrissa retraction. The stimulation magnitude was adjusted to the minimal value that reliably generated the maximal possible movement amplitude. All other vibrissae except the principal vibrissa were cut. The tip of the principal vibrissa contacted a piece of P320 sandpaper (2 Cm2), which the vibrissa brushed against during the artificial whisking. Artificial whisking was visually monitored under a stereo microscope to ensure proper movements of the vibrissa. We repeated the stimulation at each condition 132 times, and divided the repetitions to three equal blocks (each containing 44 repetitions). Blocks of the different stimulation conditions are applied in random order during the experiments.We repeated the stimulation in each condition 132 times, divided to three equal blocks (each containing 44 repetitions). Blocks of the different stimulation conditions were applied in random order during the experiments.

  2. Passive ramp-and-hold stimulation of the principal vibrissa. With this stimulation paradigm, the principal vibrissa was rapidly deflected (1300°/second) with a single ceramic piezoelectric bimorph for a period of 200 ms (Simons, 1983; Wilent and Contreras, 2004; Lavzin et al., 2012; Garion et al., 2014). To avoid ringing of the vibrissa during the rapid deflection phases, we generated a sigmoidal onset and offset of the ramp and hold pulses. To generate the sigmoidal onset and offset of the pulse we first, generated a sloped onset and offset phase lasting five millisecond. Second we applied a forward low-pass filter on the stimulus waveform using the following equation: Xn=X(n-1)+(Xn-X(n-1))*SF, where SF designates the smoothening factor.

  3. 3. Third, we used a consecutive backward low-pass filter on the stimulus wave form using the following equation: Xn=X(n+1)+(Xn-X(n+1))*SF.We tested SF ranging from 0.03–1, and for our experiments we used a SF = 0.03, resulting in an effective rise/fall duration of approximately 10–20 ms (Figure 1—figure supplement 3). Piezo bimorph deflections were controlled via a National Instruments board (PCI 6713), using custom routines written in MatLab. To monitor and calibrate vibrissa movements, and confirm lack of distortion or ringing of the stimulated vibrissa the deflection was monitored using a laser displacement sensor (LD1605-2; Micro-Epsilon OptoNCDT 1700) (Lottem and Azouz, 2009; Lavzin et al., 2012; Garion et al., 2014), and a high-speed camera (1000 fps) (Flare, 4M180MCL, 4 Megapixel, Dalsa Xcelera-x4-CL, IO industries) the Streams six acquisition software (IO industries) and homemade analysis software written in Matlab (MathWorks, NA) (Garion et al., 2014). Figure 1—figure supplement 3A presents single traces of piezo bimorph movements during ramp and hold pulses generated with different smoothening factors (0.03–1). Figure 1—figure supplement 3B shows the average (mean±SEM) peak-to-peak ringing amplitude during ramp and hold stimulation pulses. Note that under our stimulation conditions (SF = 0.03) overshoot and ringing was minimal.

  4. Passive ramp-and-hold deflection of the principal vibrissa to eight different directions to identify the angular tuning of neurons. The principal vibrissa was deflected in the different directions with 200 ms ramp-and-hold stimuli using two perpendicular pairs of ceramic piezoelectric bimorphs (For calibration of the piezoelectric bimorphs see previous section, Lavzin et al., 2012; Garion et al., 2014). Altogether, the principal vibrissa was deflected to eight different directions separated by 45° (0°, 45°, 90°, 135°, 180°, 225°, 270°, 315°), and delivered at 0.5 Hz to prevent steady-state adaptation of vibrissa-evoked responses. We repeated the ramp and hold vibrissa deflection to each angle direction 50 times with the vM1 laser off and 50 times with the vM1 laser on. The repetitions for each direction with and without vM1 laser activation were divided into two equal blocks and the different deflection blocks were applied in a random order.

Similar to the case of ramp and hold vibrissa deflection to a single direction we calibrated the perpendicular bimorphs, and made sure that no ringing and distortions occurred during stimulation we used a sigmoid function for the onset and offset piezo movements, resulting in an effective rise/fall duration of approximately 10–20 ms, and monitored movements of the vibrissa using both a laser displacement sensor (LD1605-2; Micro-Epsilon OptoNCDT 1700) and visual vibrissa tracking with a high-speed camera (Flare, 4M180MCL, 4 Megapixel, Dalsa Xcelera-x4-CL, IO industries at 1000 fps) (See above).

To quantify angular tuning we used two different methods:

  1. Selectivity index (SI). The SI reflected the ratio between the response to the preferred angle and the average response to all angles. For each neuron, we calculated the Selectivity index (SI) using the following formula:
    SI=Preferred /(Ri0.125)

    With Rpreffered, designating the response at the preferred angle, and Ri designating the sum of responses obtained in each one of the eight directions. In some case, we also calculated the SI of non preferred angles, by replacing Rmax with response to the relevant angle.

  2. Vector sum. For each neuron, we plotted the response to each angle as a vector on a 2-demminsional plane. In turn, the eight vectors obtained for the different stimulation angles were summed to a single vector. Thus, the overall response was represented as a single vector with an amplitude and direction (Mazurek et al., 2014).

Viral vector injection and optogenetic stimulation

To transfect pyramidal neurons of vM1 with channelrhodopsin 2 (ChR2), a craniotomy was drilled over the vM1 region (2 mm anterior to bregma and 1.25 mm lateral to the midline) under Isoflurane anesthesia and local anesthetic injection in P25-30 rats. A solution (500 nl) containing the ChR2 expressing viral vector (AAV2.1.CAMKIIa-hChR2(H134)-mCherry.WPRE.hGH, UNC viral core facility, North Carolina, USA) was injected into the vM1 region via a small craniotomy. ChR2 expressed in excitatory neurons by using the promotor CaMKII. Viral vby a micromanipulator (Sutter Instruments, California, USA) and connected to glass pipettes with tip diameters of 30–50 um. To allow for ChR2 expression in the target pyramidal neurons electrophysiological recordings were performed 4–6 weeks after viral vector injections. At the end of each experiment, the location and extent of ChR2 expression were verified histologically. Typically, viral vectors were injected at two depths of 300 and 500 µm. These injections typically resulted in ChR2 expression in a cortical region with vertical and horizontal extents of 500–600 (Figure 1).

  1. To activate vM1 during the experiments (performed 4–6 weeks after viral vector injections to allow for ChR2 expression), a craniotomy was re-opened over the vM1 and vS1 region, and a dental cement well was constructed around the craniotomy. The dental cement well was filled with artificial cerebrospinal fluid (aCSF) to prevent drying of the cortical surface during the recording. To activate ChR2 channel, short 473 nm light pulses were generated with a laser (OEM laser pulses, Utah, USA) controlled by an isolated pulse generator (STG4, multichannel system, Germany). The 473 nm laser was connected to a light guide that was held 10–20 mm above the neocortical surface. Control experiments revealed that in the absence of ChR2 multiple repeated laser pulses do not affect the activity of neurons in vM1. We repeated the stimulation in each condition (including paired optogenetic stimulation and vibrissa stimulation) 132 times, divided to three equal blocks (44 repetitions), and the blocks were applied in random order.

In control rats (intact un-severed buccolibial nerve), optogenetic stimulation of vM1 resulted in whisking movements (n = 5), functionally confirming the location of vM1. Upon cutting the buccolabial nerve, optogenetic stimulation of vM1 showed whisking movements in all rats tested, including the five rats in which vM1 optogenetic activation yielded whisking prior to cutting the buccolabial nerve.

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Funding Information

This paper was supported by the following grants:

  • Israel Science Foundation to Jackie Schiller.

  • Adelis Foundation to Yitzhak Schiller.

  • Prince Neurodegeneration Center to Yitzhak Schiller.

Additional information

Competing interests

The authors declare that no competing interests exist.

Author contributions

MK, Data curation, Formal analysis, Investigation.

JS, Resources, Supervision, Funding acquisition.

YS, Conceptualization, Formal analysis, Supervision, Funding acquisition, Validation, Methodology, Writing—original draft, Writing—review and editing.

Ethics

Animal experimentation: This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All of the animals were handled according to approved institutional animal care and use committee (IACUC) protocols of the technion.(protocol 007-01-2014) All surgery and experiments were performed under sodium urethane anesthesia, and every effort was made to minimize suffering.

Additional files

Source code 1. Matlab homemade software.

DOI: http://dx.doi.org/10.7554/eLife.21843.013

elife-21843-code1.zip (19.3KB, zip)
DOI: 10.7554/eLife.21843.013

References

  1. Ahrens KF, Kleinfeld D. Current flow in Vibrissa motor cortex can phase-lock with exploratory rhythmic whisking in rat. Journal of Neurophysiology. 2004;92:1700–1707. doi: 10.1152/jn.00020.2004. [DOI] [PubMed] [Google Scholar]
  2. Andermann ML, Moore CI. A somatotopic map of vibrissa motion direction within a barrel column. Nature Neuroscience. 2006;9:543–551. doi: 10.1038/nn1671. [DOI] [PubMed] [Google Scholar]
  3. Aronoff R, Matyas F, Mateo C, Ciron C, Schneider B, Petersen CC. Long-range connectivity of mouse primary somatosensory barrel cortex. European Journal of Neuroscience. 2010;31:2221–2233. doi: 10.1111/j.1460-9568.2010.07264.x. [DOI] [PubMed] [Google Scholar]
  4. Bagdasarian K, Szwed M, Knutsen PM, Deutsch D, Derdikman D, Pietr M, Simony E, Ahissar E. Pre-neuronal morphological processing of object location by individual whiskers. Nature Neuroscience. 2013;16:622–631. doi: 10.1038/nn.3378. [DOI] [PubMed] [Google Scholar]
  5. Bosman LWJ, Houweling AR, Owens CB, Tanke N, Shevchouk OT, Rahmati N, Teunissen WHT, Ju C, Gong W, Koekkoek SKE, De Zeeuw CI. Anatomical pathways involved in generating and sensing rhythmic whisker movements. Frontiers in Integrative Neuroscience. 2011;5:53. doi: 10.3389/fnint.2011.00053. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Branco T, Clark BA, Häusser M. Dendritic discrimination of temporal input sequences in cortical neurons. Science. 2010;329:1671–1675. doi: 10.1126/science.1189664. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Brecht M, Schneider M, Sakmann B, Margrie TW. Whisker movements evoked by stimulation of single pyramidal cells in rat motor cortex. Nature. 2004;427:704–710. doi: 10.1038/nature02266. [DOI] [PubMed] [Google Scholar]
  8. Brecht M. What makes whiskers shake? focus on "Current flow in vibrissa motor cortex can phase-lock with exploratory rhythmic whisking in rat". Journal of Neurophysiology. 2004;92:1265–1266. doi: 10.1152/jn.00404.2004. [DOI] [PubMed] [Google Scholar]
  9. Bruno RM, Khatri V, Land PW, Simons DJ. Thalamocortical angular tuning domains within individual barrels of rat somatosensory cortex. Journal of Neuroscience. 2003;23:9565–9574. doi: 10.1523/JNEUROSCI.23-29-09565.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Carvell GE, Miller SA, Simons DJ. The relationship of vibrissal motor cortex unit activity to whisking in the awake rat. Somatosensory & Motor Research. 1996;13:115–127. doi: 10.3109/08990229609051399. [DOI] [PubMed] [Google Scholar]
  11. Carvell GE, Simons DJ. Biometric analyses of vibrissal tactile discrimination in the rat. Journal of Neuroscience. 1990;10:2638–2648. doi: 10.1523/JNEUROSCI.10-08-02638.1990. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Cichon J, Gan WB. Branch-specific dendritic Ca(2+) spikes cause persistent synaptic plasticity. Nature. 2015;520:180–185. doi: 10.1038/nature14251. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Curtis JC, Kleinfeld D. Phase-to-rate transformations encode touch in cortical neurons of a scanning sensorimotor system. Nature Neuroscience. 2009;12:492–501. doi: 10.1038/nn.2283. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Derdikman D, Yu C, Haidarliu S, Bagdasarian K, Arieli A, Ahissar E. Layer-specific touch-dependent facilitation and depression in the somatosensory cortex during active whisking. Journal of Neuroscience. 2006;26:9538–9547. doi: 10.1523/JNEUROSCI.0918-06.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Diamond ME, Arabzadeh E. Whisker sensory system - from receptor to decision. Progress in Neurobiology. 2013;103:28–40. doi: 10.1016/j.pneurobio.2012.05.013. [DOI] [PubMed] [Google Scholar]
  16. Diamond ME, von Heimendahl M, Knutsen PM, Kleinfeld D, Ahissar E. 'Where' and 'what' in the whisker sensorimotor system. Nature Reviews Neuroscience. 2008;9:601–612. doi: 10.1038/nrn2411. [DOI] [PubMed] [Google Scholar]
  17. Feldmeyer D, Brecht M, Helmchen F, Petersen CC, Poulet JF, Staiger JF, Luhmann HJ, Schwarz C. Barrel cortex function. Progress in Neurobiology. 2013;103:3–27. doi: 10.1016/j.pneurobio.2012.11.002. [DOI] [PubMed] [Google Scholar]
  18. Garion L, Dubin U, Rubin Y, Khateb M, Schiller Y, Azouz R, Schiller J. Texture coarseness responsive neurons and their mapping in layer 2-3 of the rat barrel cortex in vivo. eLife. 2014;3:e03405. doi: 10.7554/eLife.03405. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Gerdjikov TV, Haiss F, Rodriguez-Sierra OE, Schwarz C. Rhythmic whisking area (RW) in rat primary motor cortex: an internal monitor of movement-related signals? Journal of Neuroscience. 2013;33:14193–14204. doi: 10.1523/JNEUROSCI.0337-13.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Grinevich V, Brecht M, Osten P. Monosynaptic pathway from rat vibrissa motor cortex to facial motor neurons revealed by lentivirus-based axonal tracing. Journal of Neuroscience. 2005;25:8250–8258. doi: 10.1523/JNEUROSCI.2235-05.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Haiss F, Schwarz C. Spatial segregation of different modes of movement control in the whisker representation of rat primary motor cortex. Journal of Neuroscience. 2005;25:1579–1587. doi: 10.1523/JNEUROSCI.3760-04.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Harnett MT, Xu NL, Magee JC, Williams SR, Nl X, Williams SR. Potassium channels control the interaction between active dendritic integration compartments in layer 5 cortical pyramidal neurons. Neuron. 2013;79:516–529. doi: 10.1016/j.neuron.2013.06.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Hill DN, Curtis JC, Moore JD, Kleinfeld D. Primary motor cortex reports efferent control of vibrissa motion on multiple timescales. Neuron. 2011;72:344–356. doi: 10.1016/j.neuron.2011.09.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Hooks BM, Mao T, Gutnisky DA, Yamawaki N, Svoboda K, Shepherd GM. Organization of cortical and thalamic input to pyramidal neurons in mouse motor cortex. Journal of Neuroscience. 2013;33:748–760. doi: 10.1523/JNEUROSCI.4338-12.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Izraeli R, Porter L. Vibrissal motor cortex in the rat: connections with the barrel field. Experimental Brain Research. 1995;104:41–54. doi: 10.1007/BF00229854. [DOI] [PubMed] [Google Scholar]
  26. Kleinfeld D, Ahissar E, Diamond ME. Active sensation: insights from the rodent vibrissa sensorimotor system. Current Opinion in Neurobiology. 2006;16:435–444. doi: 10.1016/j.conb.2006.06.009. [DOI] [PubMed] [Google Scholar]
  27. Kleinfeld D, Berg RW, O'Connor SM. Anatomical loops and their electrical dynamics in relation to whisking by rat. Somatosensory & Motor Research. 1999;16:69–88. doi: 10.1080/08990229970528. [DOI] [PubMed] [Google Scholar]
  28. Kleinfeld D, Deschênes M. Neuronal basis for object location in the vibrissa scanning sensorimotor system. Neuron. 2011;72:455–468. doi: 10.1016/j.neuron.2011.10.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Kleinfeld D, Moore JD, Wang F, Deschênes M. The brainstem oscillator for whisking and the case for breathing as the master clock for orofacial motor actions. Cold Spring Harbor Symposia on Quantitative Biology. 2014;79:29–39. doi: 10.1101/sqb.2014.79.024794. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Kleinfeld D, Sachdev RN, Merchant LM, Jarvis MR, Ebner FF. Adaptive filtering of vibrissa input in motor cortex of rat. Neuron. 2002;34:1021–1034. doi: 10.1016/S0896-6273(02)00732-8. [DOI] [PubMed] [Google Scholar]
  31. Kremer Y, Léger JF, Goodman D, Brette R, Bourdieu L. Late emergence of the vibrissa direction selectivity map in the rat barrel cortex. Journal of Neuroscience. 2011;31:10689–10700. doi: 10.1523/JNEUROSCI.6541-10.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Larkum ME, Nevian T, Sandler M, Polsky A, Schiller J. Synaptic integration in tuft dendrites of layer 5 pyramidal neurons: a new unifying principle. Science. 2009;325:756–760. doi: 10.1126/science.1171958. [DOI] [PubMed] [Google Scholar]
  33. Larkum ME, Zhu JJ, Sakmann B. A new cellular mechanism for coupling inputs arriving at different cortical layers. Nature. 1999;398:338–341. doi: 10.1038/18686. [DOI] [PubMed] [Google Scholar]
  34. Lavzin M, Rapoport S, Polsky A, Garion L, Schiller J. Nonlinear dendritic processing determines angular tuning of barrel cortex neurons in vivo. Nature. 2012;490:397–401. doi: 10.1038/nature11451. [DOI] [PubMed] [Google Scholar]
  35. Lee S, Kruglikov I, Huang ZJ, Fishell G, Rudy B. A disinhibitory circuit mediates motor integration in the somatosensory cortex. Nature Neuroscience. 2013;16:1662–1670. doi: 10.1038/nn.3544. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. London M, Häusser M. Dendritic computation. Annual Review of Neuroscience. 2005;28:503–532. doi: 10.1146/annurev.neuro.28.061604.135703. [DOI] [PubMed] [Google Scholar]
  37. Lottem E, Azouz R. Mechanisms of tactile information transmission through whisker vibrations. Journal of Neuroscience. 2009;29:11686–11697. doi: 10.1523/JNEUROSCI.0705-09.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Major G, Larkum ME, Schiller J. Active properties of neocortical pyramidal neuron dendrites. Annual Review of Neuroscience. 2013;36:1–24. doi: 10.1146/annurev-neuro-062111-150343. [DOI] [PubMed] [Google Scholar]
  39. Manita S, Suzuki T, Homma C, Matsumoto T, Odagawa M, Yamada K, Ota K, Matsubara C, Inutsuka A, Sato M, Ohkura M, Yamanaka A, Yanagawa Y, Nakai J, Hayashi Y, Larkum ME, Murayama M. A Top-Down cortical circuit for accurate sensory perception. Neuron. 2015;86:1304–1316. doi: 10.1016/j.neuron.2015.05.006. [DOI] [PubMed] [Google Scholar]
  40. Mao T, Kusefoglu D, Hooks BM, Huber D, Petreanu L, Svoboda K. Long-range neuronal circuits underlying the interaction between sensory and motor cortex. Neuron. 2011;72:111–123. doi: 10.1016/j.neuron.2011.07.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Mazurek M, Kager M, Van Hooser SD. Robust quantification of orientation selectivity and direction selectivity. Frontiers in Neural Circuits. 2014;8:92. doi: 10.3389/fncir.2014.00092. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. McAdams CJ, Maunsell JH. Effects of attention on the reliability of individual neurons in monkey visual cortex. Neuron. 1999;23:765–773. doi: 10.1016/S0896-6273(01)80034-9. [DOI] [PubMed] [Google Scholar]
  43. Moore JD, Deschênes M, Furuta T, Huber D, Smear MC, Demers M, Kleinfeld D. Hierarchy of orofacial rhythms revealed through whisking and breathing. Nature. 2013;497:205–210. doi: 10.1038/nature12076. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Moore JD, Kleinfeld D, Wang F. How the brainstem controls orofacial behaviors comprised of rhythmic actions. Trends in Neurosciences. 2014;37:370–380. doi: 10.1016/j.tins.2014.05.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Noudoost B, Chang MH, Steinmetz NA, Moore T. Top-down control of visual attention. Current Opinion in Neurobiology. 2010;20:183–190. doi: 10.1016/j.conb.2010.02.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Palmer LM, Shai AS, Reeve JE, Anderson HL, Paulsen O, Larkum ME. NMDA spikes enhance action potential generation during sensory input. Nature Neuroscience. 2014;17:383–390. doi: 10.1038/nn.3646. [DOI] [PubMed] [Google Scholar]
  47. Petersen CC. The functional organization of the barrel cortex. Neuron. 2007;56:339–355. doi: 10.1016/j.neuron.2007.09.017. [DOI] [PubMed] [Google Scholar]
  48. Petersen CC. Cortical control of whisker movement. Annual Review of Neuroscience. 2014;37:183–203. doi: 10.1146/annurev-neuro-062012-170344. [DOI] [PubMed] [Google Scholar]
  49. Petreanu L, Gutnisky DA, Huber D, Xu NL, O'Connor DH, Tian L, Looger L, Svoboda K, Nl X, Svoboda K. Activity in motor-sensory projections reveals distributed coding in somatosensation. Nature. 2012;489:299–303. doi: 10.1038/nature11321. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Pi HJ, Hangya B, Kvitsiani D, Sanders JI, Huang ZJ, Kepecs A, Hj P, Kepecs A. Cortical interneurons that specialize in disinhibitory control. Nature. 2013;503:521–534. doi: 10.1038/nature12676. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Polsky A, Mel BW, Schiller J. Computational subunits in thin dendrites of pyramidal cells. Nature Neuroscience. 2004;7:621–627. doi: 10.1038/nn1253. [DOI] [PubMed] [Google Scholar]
  52. Reynolds JH, Chelazzi L. Attentional modulation of visual processing. Annual Review of Neuroscience. 2004;27:611–647. doi: 10.1146/annurev.neuro.26.041002.131039. [DOI] [PubMed] [Google Scholar]
  53. Rousche PJ, Petersen RS, Battiston S, Giannotta S, Diamond ME. Examination of the spatial and temporal distribution of sensory cortical activity using a 100-electrode array. Journal of Neuroscience Methods. 1999;90:57–66. doi: 10.1016/S0165-0270(99)00061-8. [DOI] [PubMed] [Google Scholar]
  54. Schiller J, Major G, Koester HJ, Schiller Y. NMDA spikes in basal dendrites of cortical pyramidal neurons. Nature. 2000;404:285–289. doi: 10.1038/35005094. [DOI] [PubMed] [Google Scholar]
  55. Schroeder CE, Wilson DA, Radman T, Scharfman H, Lakatos P. Dynamics of active sensing and perceptual selection. Current Opinion in Neurobiology. 2010;20:172–176. doi: 10.1016/j.conb.2010.02.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Simons DJ. Multi-whisker stimulation and its effects on vibrissa units in rat SmI barrel cortex. Brain Research. 1983;276:178–182. doi: 10.1016/0006-8993(83)90561-9. [DOI] [PubMed] [Google Scholar]
  57. Smith SL, Smith IT, Branco T, Häusser M. Dendritic spikes enhance stimulus selectivity in cortical neurons in vivo. Nature. 2013;503:115–120. doi: 10.1038/nature12600. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Squire RF, Noudoost B, Schafer RJ, Moore T. Prefrontal contributions to visual selective attention. Annual Review of Neuroscience. 2013;36:451–466. doi: 10.1146/annurev-neuro-062111-150439. [DOI] [PubMed] [Google Scholar]
  59. Sreenivasan V, Karmakar K, Rijli FM, Petersen CC. Parallel pathways from motor and somatosensory cortex for controlling whisker movements in mice. European Journal of Neuroscience. 2015;41:354–367. doi: 10.1111/ejn.12800. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Stuart G, Schiller J, Sakmann B. Action potential initiation and propagation in rat neocortical pyramidal neurons. The Journal of Physiology. 1997;505:617–632. doi: 10.1111/j.1469-7793.1997.617ba.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Szwed M, Bagdasarian K, Ahissar E. Encoding of vibrissal active touch. Neuron. 2003;40:621–630. doi: 10.1016/S0896-6273(03)00671-8. [DOI] [PubMed] [Google Scholar]
  62. Veinante P, Deschênes M. Single-cell study of motor cortex projections to the barrel field in rats. Journal of Comparative Neurology. 2003;464:98–103. doi: 10.1002/cne.10769. [DOI] [PubMed] [Google Scholar]
  63. Wilent WB, Contreras D. Synaptic responses to whisker deflections in rat barrel cortex as a function of cortical layer and stimulus intensity. Journal of Neuroscience. 2004;24:3985–3998. doi: 10.1523/JNEUROSCI.5782-03.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Woolsey TA, Van der Loos H. The structural organization of layer IV in the somatosensory region (SI) of mouse cerebral cortex. The description of a cortical field composed of discrete cytoarchitectonic units. Brain Research. 1970;17:205–242. doi: 10.1016/0006-8993(70)90079-X. [DOI] [PubMed] [Google Scholar]
  65. Xu NL, Harnett MT, Williams SR, Huber D, O'Connor DH, Svoboda K, Magee JC, Nl X, Magee JC. Nonlinear dendritic integration of sensory and motor input during an active sensing task. Nature. 2012;492:247–251. doi: 10.1038/nature11601. [DOI] [PubMed] [Google Scholar]
  66. Zagha E, Casale AE, Sachdev RN, McGinley MJ, McCormick DA. Motor cortex feedback influences sensory processing by modulating network state. Neuron. 2013;79:567–578. doi: 10.1016/j.neuron.2013.06.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
eLife. 2017 Jan 6;6:e21843. doi: 10.7554/eLife.21843.014

Decision letter

Editor: Sacha B Nelson1

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "Feedforward motor information enhance somatosensory responses and sharpen angular tuning of S1 barrel cortex neurons" for consideration by eLife. Your article has been favorably evaluated by Gary Westbrook (Senior Editor) and three reviewers, one of whom, Sacha Nelson, is a member of our Board of Reviewing Editors. Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

Each of the reviewers felt the underlying issues were important and the major findings were potentially interesting, but each also had significant (often overlapping) concerns with the rigor of the methods, the main points being

1) the rigor of spike sorting and quality of the recordings

2) the details of stimulus timing and stimulus control, and

3) clarity surrounding the issue of response variability and stability – both generally and as it relates to the significance of the angular tuning and enhancement thereof.

It is the policy at eLife to reject manuscripts if the reviewers feel that additional work is needed to support the conclusions of the paper, and if in the editors' opinions this work is extensive enough to require more than about 2 months. We would be open to considering a new revised manuscript if the authors want to try to address the issues outlined above in a new manuscript. We wish to stress, however, that it might be difficult for the authors to raise the manuscript over the bar, because of the many concerns raised.

Reviewer #1:

The authors use multi-site extracellular recording in anesthetized rats to study the influence of M1 projections on sensory responses to whisker stimulation in S1. This is a well studied area and has been the subject of a handful of recent high profile papers. The present study makes an important contribution to this field by breaking the normally recurrent sensory-motor loop so as to be able to study the effect of motor cortex activity in the absence of the indirect effects of the movement that activity would normally produce. To do this, they sever the motor nerve to prevent centrally generated whisking and then stimulate the whiskers passively or by causing artificial whisking by directly stimulating the nerve, while at the same time stimulating the motor cortex ontogenetically. They find that there is a feed-forward facilitation of sensory responses and that this facilitation is strongest for the preferred stimulus, thereby enhancing stimulus specificity. Although there have been other demonstrations of circuits that might produce this facilitation, the sensory and temporal features of the facilitation have not been studied under conditions in which it can be cleanly isolated as here.

1) The manuscript is in need of substantial editing for usage and for clarity. I have tried to catch as many of the errors as I can, but it extends to being unsure of precisely what procedures were followed and what some of the figures depict.

2) The inclusion criteria with respect to response stability are not clear and may be inappropriate. It is stated that two parameters were used: 1) the baseline pre-stimulus frequency – but this is not justified and no criteria are given, and 2) "…conditions was repeated twice…and neurons in which the responded differentially to the two stimulation blocks were excluded." This is not clear and no definition of difference is given, but treating the responses as an n=2 repetitions and excluding based on differences is not valid without some statistical definition of repeatability. Along these lines, the variability of the responses to different directions of stimulation does not seem to have been adequately assessed. Tuning curves for individual neurons should have error bars so that the claim of enhanced tuning can be assessed statistically on a neuron-by neuron basis.

3) The results are quantified solely with respect to the on response. Off responses should also be shown (examples of the full-time course of individual responses with and without M1 stimulation should be shown) so as to reveal the kinetics of facilitation and to perhaps justify the quantification of on responses only.

4) It is not clear what is different about Figure 1F (39% larger) and 1G all (73.5% larger). The same lack of clarity applies to Figure 2. Is Figure 3B a single example neuron? If so, should a) make it clear that the response on the left and the response on the right are two individual neurons and b) the variability of the response across trials should be indicated. Figure 4 should also include measures of response variability. In Figure 5, the asterisk should be defined and exactly which groups were being compared should be stated clearly (e.g. in the Methods or in the legend or text). If the pooled controls are being compared to individual layers (i.e. the layers don't have their own separate control groups – which would not make sense but which is somewhat implied by the structure of the figure) a correction for multiple comparisons must be used.

Figure 4. Temporal rules governing the interactions between of vM1 activation and vibrissa stimulation on the response of vS1 barrel cortex neurons.

Figure 4.

(A) The peak of the peri-stimulus histograms (PSTH) recorded from individual vS1 neurons during paired vM1-vibrissa stimulation with different time lags between the vM1 optogenetic activation (20 ms laser pulse) and vibrissa stimulation (−100 ms up to +50 ms). The time lags were calculated by subtracting the onset time of vibrissa stimulation from the onset time of vM1 optogenetic activation. The data are presented for two different individual neurons during passive ramp and hold vibrissa deflection (left panel) and artificial whisking against sandpaper (right panel). (B) The response (mean ± SEM) of two different individual vS1 neurons to paired vM1-vibrissa stimulation applied with different inter-stimuli time lags between the optogenetic and vibrissa deflection. In the left panel, the vibrissa was stimulated with a ramp and hold vibrissa deflection, and in the right panel, the vibrissa was stimulated with artificial whisking against a P320 sand paper. (C) The average (mean ± SEM) response of all recorded vS1 neurons to isolated passive vibrissa stimulation (control) and paired vM1-vibrissa stimulation applied with different inter-stimuli time lags. For both B and C, the data are presented for passive ramp and hold vibrissa deflection (left panels) and artificial whisking against sandpaper (right panels). 238 neurons from six rats for vibrissa deflection experiments; 334 neurons from seven rats in the artificial whisking experiments; *p<0.05, ***p<0.001.

DOI: http://dx.doi.org/10.7554/eLife.21843.008

5) The authors should show some example histology that convinces the reader that the assignment to layers is accurate.

Reviewer #2:

The supralinear main effect observed in this paper is undeniably interesting.

Major:

I) Stimulus control is essential for this experiment:

1) Perpendicularly-arrayed bimorphs are notorious for ringing and distortion. I would like quantification of the amplitude and velocity of motion for all 8 directions for the piezo's used, and most importantly the ringing they demonstrated both for the first deflection and the other deflections in a train. Such quantification and description is essential.

2) I'm confused by the description of stimulation in the Results (200 ms ramp and hold stimuli) and in the Methods (100 ms).

II) I am skeptical about the 50 micron MEAs from neuronexus finding any single units of demonstrable selectivity in layers II/III. I certainly am not impugning the author's basic credibility (I don't think they are lying!) but getting well-isolated units in those layers, or layer IV especially, with such probes is quite hard. Without the virtue of a tetrode configuration or the use of smaller diameter contacts the claim of single unit identification is particularly suspect.

I also found the following criteria to feel a bit arbitrary and wasn't sure what it meant:

"For our analysis, we excluded all unstable neurons, with stability determined using two parameters. First, the baseline pre-stimulus frequency. Second, for every experimental paradigm at least one of the stimulation conditions was repeated twice during the experiment, and neurons in which the responded differentially to the two stimulation blocks were excluded."

First, what does the phrase "the baseline pre-stimulus frequency" have to do with stability? Second, multi-unit recordings can easily show run-by-run stability and, in fact, are potentially more stable than units for such recordings.

So, I'd like to see the following to give me confidence:

1) Application of autocorrelogram exclusion criteria.

2) Quantification of spike feature stability across the entire run (whether or not the response properties changed), and a discussion of which spike features were used.

3) Demonstration of the cluster quality for some number of units –.I don't want to make this infinite, but as I said I'm skeptical, so at least a fair number of examples of stability in layers II/III units would be appreciated (say, 10 such examples from 10 experiments or so).

Moderate:

I) In the 56 neurons recorded in vMI, what was the latency to firing following optogenetic stimulation, and was this a bimodal distribution (implying directly opto activated and synaptically opto activated populations).

II) The authors should quantify the relative effect of stimulation on the optimal direction angle versus the worst direction angle, whatever these might have been. Obviously, there must be enhancement on average across randomly associated directions or else the main effect of Figure 1 couldn't stand: The sharpening obviously suggests, though, a significant different between best and worst, etc.

Reviewer #3:

Khateb et al. report on modulation of whisker-stimulation evoked spike rates in rat barrel cortex by optogenetic activation of projections from motor cortex. To avoid confounding effects by sensory reafference activation they prevented initiation of whisker movements by cutting the motor branch of the facial nerve. For two stimulation paradigms under anesthesia they find that pairing of motor cortex activation with whisker stimulation leads to supralinear spiking responses in barrel cortex neurons across all layers, most prominently in L2/3 and L5. This effect depended on the relative timing of the two inputs, peaking when motor cortex activation shortly preceded the whisker stimulation. In addition, they report a sharpening of tuning to the angular direction of stimulation upon pairing.

How a cortical sensory area such as the barrel cortex integrates the various input streams it receives, is a fundamental, most relevant question. Modern tools, especially optogenetic precise control of specific pathways as applied here, now enable addressing this question in vivo. This study therefore is timely and presents highly original data. The finding that neuronal spiking activity in barrel cortex is amplified by activation of the M1-S1 pathway is highly significant, even though the underlying potential mechanisms are not investigated here, as the authors admit. The results regarding changes in angular tuning I find less convincing, given that this feature is a complex, debated issue and little data are provided here. Overall, this is a nice study, making an important contribution to signal processing in the barrel cortex. The manuscript would benefit, however, from a more detailed and extensive presentation of (raw) data and a more in-depth treatment of the crucial issue of relative input timing. Find my specific comments below.

1) Introduction, last paragraph: Why only 'partially disconnected'? Isn't the motor arm entirely uncoupled? Was this actually verified by confirming the absence of whisker movements following vM1 activation after the nerve cut?

2) Subsection “Electrophysiological unit recordings” and figures. The authors should provide some raw electrophysiology data, so that one can better judge the quality of data. What were the noise levels? How well could spike sorting be performed and on how many channels? How well could spike waveforms be separated and were there any putative fast-spiking units present? Do the reported spike rates represent changes in spike rate? It would also be interesting to see the LFP responses for the paired stimulation paradigms.

3) Furthermore: what were the respective baseline spike frequencies and why (and how) were these used to exclude data from the analysis (subsection “Electrophysiological unit recordings”, third paragraph). I also do not understand the motivation for excluding units that did not show a consistent response pattern upon repeating the stimulation paradigm. What type of responses did these cells actually show, in how far did they deviate from a nonlinear summation results? Why should one set such bias?

4) Artificial whisking paradigm (subsection “Whisker stimuli”, Figure 2, Figure 3). How exactly was the optogenetic stimulation paired with motor activation pulses? Was it a 20-ms light pulse at the beginning of the 10 pulses to the nerve? How large were the whisker protractions induced by this protocols? The spiking response appears only 20-ms long, how come? What happened in the remaining 80 ms of protraction/retraction cycle? In the eighth paragraph of the subsection “Pairing optogenetic vM1 activation with passive whisker activation” it is stated that spikes were evoked during protraction and retraction phase, but I can't see that. More details are needed here.

5) Subsection “Viral vector injection and optogenetic stimulation”, first paragraph. How precisely was the vM1 hit with the injections and the light stimulation? Was this verified by measuring whisker movements (or other movements?) by optogenetic vM1 stimulation before the facial nerve was cut?

6) I find the timing experiments very interesting but I am not sure how to interpret the peak revealed at -20 ms. This is a time scale where the conduction delays etc. play an important role. So how was the timing exactly defined (onset of piezo drive and onset of LED illumination, I presume)? Was there any dead time for mechanical stimulation of the whisker considered? In particular, how was the timing defined for the artificial whisking stimulus? Were the whisker movements (presumably stick-slip events) monitored and their timing analyzed? Were any axonal conduction times and synaptic delays taken into account? Obviously, these questions are important to understand what the real timing difference at the integrating neurons in barrel cortex might be. Supplementary whole-cell recordings could be very helpful here, to quantify when exactly inputs from both pathways actually arrive with these stimulation paradigms.

7) Angular tuning in barrel cortex apparently is a complicated matter and among other things seems to especially depend on age (Kremer study). In addition, the housing conditions (use of whiskers) may affect the outcome. The age of rats used here is just in between the ages when no angular tuning was observed and when it later was established. I find the examples in Figure 4 not convincing, as the pure whisker-evoked responses appear relatively untuned and the responses enhanced by optogenetic vM1 activation mostly display multiple peaks (often in orthogonal directions). Cells from what layer are actually shown in Figure 4? It might be helpful to show distributions of the absolute SI values for the different stimulation protocols.

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for submitting your article "Feedforward motor information enhance somatosensory responses and sharpen angular tuning of S1 barrel cortex neurons" for consideration by eLife. Your article has been favorably evaluated by Gary Westbrook (Senior Editor) and three reviewers, one of whom, Sacha B Nelson (Reviewer #1), is a member of our Board of Reviewing Editors, and another one is Fritjof Helmchen (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

All three reviewers were pleased with the addition of new data and new analyses addressing the issue of spike sorting. The reviewers remain concerned about several issues outlined below. In addition, the manuscript could still benefit from additional editing.

Essential revisions:

1) One reviewer states: "I appreciate the addition of details on how the stimulator was calibrated-these are all the right approaches and tools. I can guarantee that using these parameters, that unless they use a specific compensatory algorithm, they almost certainly cannot get ringing under 5% of stroke magnitude, and I doubt it can be under 10%. I want the following included in the final manuscript → An actual analysis of the mean amplitude of the ringing. Saying you did not see it does not reflect quantification of the effect. Say how large in angle of vibrissal base motion and at what frequency the ring is on average across trials and across the different stimulators used." The reviewer notes that this is likely to have an effect on the direction selective responses.

2) A second reviewer felt that the documentation of the degree of direction selectivity was inadequate. They did not feel the statement "Consistent with previous results we found angular tuning in neurons.…" (subsection “The effect of vM1 activation on angular tuning of neurons in the vS1 barrel cortex”, first paragraph) was well supported. They felt that the criteria for when a cell's response is considered well tuned were not clear and that the SI as calculated made comparisons across neurons with very different response amplitudes difficult. It is suggested that the authors "confirm their interpretation with an alternative, more robust analysis method of direction-tuning, based on the mean response vector not the Rmax (Kremer et al. 2011; Mazurek et al., Front Neural Circ 2014). And provide a statistical argument for their statement 'we found angular tuning'."

3) One of the reviewers also notes that "laser stimulation of vM1 alone in essentially all cases did not evoke spiking activity at all (except perhaps for Figure 1B?). Thus, the 'supralinear' effect essentially consists in an upregulation or facilitation of the sensory-evoked response in vS1, which could be simply explained by additional pre-depolarization mediated by the M1-to-S1 projections, given vM1 is stimulated at the right time briefly before the sensory stimulus. While the authors mention this simple explanation (vM1 projection fibers helping vS1 neurons to reach the – nonlinear – spike threshold), they only refer to it as 'additional cellular mechanism' in the third paragraph of the subsection “Possible mechanisms underlying vM1 mediated response amplification and sharpening of angular tuning in vS1 barrel neurons”. Their primary 'attractive potential mechanism' of dendritic amplification (in the first paragraph of the aforementioned subsection) in my view is, however, largely speculative." The authors should consider toning down their use of 'supralinear responses' and may consider using terms like 'facilitation' or 'modulation.' At the very least they should give more equal weight to the simplest interpretation of these effects.

eLife. 2017 Jan 6;6:e21843. doi: 10.7554/eLife.21843.015

Author response


[Editors’ note: the author responses to the first round of peer review follow.]

Each of the reviewers felt the underlying issues were important and the major findings were potentially interesting, but each also had significant (often overlapping) concerns with the rigor of the methods, the main points being

1) the rigor of spike sorting and quality of the recordings

2) the details of stimulus timing and stimulus control, and

3) clarity surrounding the issue of response variability and stability – both generally and as it relates to the significance of the angular tuning and enhancement thereof.

1) Following the concerns raised by all reviewers regarding the rigor of our spike sorting we redefined stringent spike sorting criteria, and reanalyzed all our data according to these criteria. 2) To overcome inherent difficulties associated with single unit spike sorting we performed multi-unit analysis of our data. This analysis, which is much less prone to technical inaccuracies, confirmed all the main findings of the manuscript. 3) We performed additional experiments, most important of which are a new series of angular tuning experiments in more mature rats (90-100 day old). 4) We added data regarding the variability of our recordings. 5) We added raw experimental data. We present raw filtered (1-5 KHz) recording traces and spike sorting of 10 individual electrodes with 26 individual single units. 6) We greatly expanded the description of various technical issues regarding the stimulation paradigms and timing of stimulation.

Reviewer #1:

[…] 1) The manuscript is in need of substantial editing for usage and for clarity. I have tried to catch as many of the errors as I can, but it extends to being unsure of precisely what procedures were followed and what some of the figures depict.

Following the reviewer’s comment we have extensively edited the manuscript.

2) The inclusion criteria with respect to response stability are not clear and may be inappropriate. It is stated that two parameters were used: 1) the baseline pre-stimulus frequency – but this is not justified and no criteria are given, and 2) "…conditions was repeated twice…and neurons in which the responded differentially to the two stimulation blocks were excluded." This is not clear and no definition of difference is given, but treating the responses as an n=2 repetitions and excluding based on differences is not valid without some statistical definition of repeatability. Along these lines, the variability of the responses to different directions of stimulation does not seem to have been adequately assessed. Tuning curves for individual neurons should have error bars so that the claim of enhanced tuning can be assessed statistically on a neuron-by neuron basis.

We thank the reviewer for his comments on this subject. Upon re-reading the manuscript, we also saw that our inclusion and exclusion criteria of units were not sufficient. For the revised manuscript, we re-defined criteria for including units in our analysis, and re-analyzed all our data according to these new and better-defined criteria.

Specifically, we initially recorded the multi-unit activity (MUA) by recording events with an amplitude >3.5 SD above baseline from the filtered raw data (1-5 KHz). From this data we next sorted single unit activity using the OFS offline spike sorting software from Plexon. Sorting was initially performed by semi-automatic algorithms, and later verified and corrected manually. We accepted clusters as single unit if they met all the following criteria: 1) the waveform shape remained consistent and stable throughout recording. This was verified by the "Sort-Quality Vs. Time" analysis in the OFS software. Moreover, we excluded unstable units in case the average amplitude or half width of units changed significantly (ANOVA test) between the first and last 20% of recorded spikes. 2) The firing rate was >0.5 Hz to allow for adequate sampling. 3) The inter-spike interval (ISI) was >2 ms to reflect the absolute refractory period of neurons. 4) The ISI distribution showed a smooth exponential-like curve. 5) Finally and most importantly statistical criterion of p<0.05 (multivariate ANOVA) of cluster separation.

These points were added to the revised manuscript (subsection “Electrophysiological unit recordings”). In addition to the revised manuscript we added examples of raw data traces (Figure 1—figure supplement 1) as well as raw clustering data (Figure 1—figure supplement 2).

In addition, in the revised manuscript we took a second approach to overcome the inherent difficulties associated with spike sorting. We repeated all our analysis for multi-unit data (threshold >3.5 SD above baseline). We found the similar to single unit analysis, multi-unit analysis also showed that vM1 optogenetic activation both supra-linearly amplified the responses of whisker stimulation, as well sharpened angular tuning. Thus, our multi-unit data verified the single unit results. The multi-unit data was added to the revised manuscript (subsection “Pairing optogenetic vM1 activation with passive whisker activation”, fifth paragraph and subsection “The effect of vM1 activation on angular tuning of neurons in the vS1 barrel cortex”, third paragraph; Figures 2B and 6A). Note that in the case of angular tuning we added date from 6 additional 90-100 day old mice.

Additionally, following the reviewer's comments, we added error bars to the curves of all individual neurons presented in the manuscript (revised Figures 1, 35).

3) The results are quantified solely with respect to the on response. Off responses should also be shown (examples of the full-time course of individual responses with and without M1 stimulation should be shown) so as to reveal the kinetics of facilitation and to perhaps justify the quantification of on responses only.

We chose to concentrate on the "on response" as the duration of the whisker stimulus is 200 ms, and thus the "OFF response" occurs approximately 180 ms after the optogenetic pulses ended.

Following the reviewer’s comments we also analyzed the off responses to ramp and hold stimulation. To our surprise, we found that pairing of ramp and hold stimuli with optogenetic activation of vM1 also resulted in supra-linear amplification of the "off response", although the magnitude of the amplification was smaller. The additional data regarding the "off response" is added the revised manuscript (subsection “Pairing optogenetic vM1 activation with passive whisker activation”, sixth paragraph).

4) It is not clear what is different about Figure 1F (39% larger) and 1G all (73.5% larger). The same lack of clarity applies to Figure 2.

The difference in the values result from different calculation methods of the same data. While in Figure 1F and 2F we initially measured the response for individual units and averaged the response for each of the conditions, in 1G and 2G we initially calculated the ratio between the recorded and linearly expected responses for individual neurons, and averaged the ration over all neurons. Following the reviewer’s comment we agree that this difference is confusing. In the revised manuscript we used a single unified method for calculating the averaged results. We preferred the more conventional method of averaging the response for each of the conditions (as used in 1F and 2F of the original manuscript). The new modified calculations are presented in Figures 2 and 3 and in the subsection “Pairing optogenetic vM1 activation with passive whisker activation”.

Is Figure 3B a single example neuron? If so, should a) make it clear that the response on the left and the response on the right are two individual neurons and b) the variability of the response across trials should be indicated.

These are indeed responses from two different individual neurons, one responding to a ramp and hold passive whisker stimulation, and the other to artificial whisking against sandpaper. Following the reviewer’s comments we clarified this point in the revised manuscript (Figure legend 4). In addition, we added error bars to the averaged values.

Figure 4 should also include measures of response variability. In Figure 5, the asterisk should be defined and exactly which groups were being compared should be stated clearly (e.g. in the Methods or in the legend or text). If the pooled controls are being compared to individual layers (i.e. the layers don't have their own separate control groups – which would not make sense but which is somewhat implied by the structure of the figure) a correction for multiple comparisons must be used.

Following the reviewers’ comments we added error bars to the angular tuning experiments presented in Figure 5. As error bars are difficult to view in polar plots we present this data in a separate figure (Figure 5—figure supplement 1). Upon re-reading the manuscript, we agree that we were not clear about the statistical comparisons presented for the different cortical layers. For each layer we compared SI of the preferred angle with and without laser. This is now clarified in the figure legend of the revised manuscript (Figure legend 6).

5) The authors should show some example histology that convinces the reader that the assignment to layers is accurate.

We performed our experiments with a single shaft 16 contact silicone probe (NeuoNexus). The inter-contact distance in our electrodes was 50 microns. We categorized our recordings to putative layers based on the recording depth from pia (depth 0 was determined visually during electrode insertion, and the depth was calculated by the location of the electrode along the electrode shaft, assuming the silicone probe remained straight and unfolded). The recording location of each contact could not be confirmed histologically, and thus, we used the term putative cortical layer. Following the reviewer’s comment we further emphasized this points in the revised manuscript (in the subsection “Pairing optogenetic vM1 activation with passive whisker activation”).

Reviewer #2:

The supralinear main effect observed in this paper is undeniably interesting.

Major:

I) Stimulus control is essential for this experiment:

1) Perpendicularly-arrayed bimorphs are notorious for ringing and distortion. I would like quantification of the amplitude and velocity of motion for all 8 directions for the piezo's used, and most importantly the ringing they demonstrated both for the first deflection and the other deflections in a train. Such quantification and description is essential.

As pointed by the reviewer, distortions and ringing of the bimorphs during ramp and hold paradigm, especially in the case of two perpendicular bimorphs, are significant issues. Our lab has worked and published in the past on the issue (Garion et al., 2014 for ramp and hold stimuli and Lavzin et al., 2012 for angular tuning). In the previous version of the manuscript, we relayed on quoting these articles. However, following the reviewer's comment in the revised manuscript we detailed our procedure for eliminating ringing of the bimorphs (subsection “Whisker stimuli”). Specifically, passive ramp-and-hold stimulation the principle whisker was rapidly deflected (1300°/second) with a single ceramic piezoelectric bimorph for a period of 200 ms (Simons, 1983; Wilent and Contreras, 2004; Lavzin et al., 2012; Garion et al., 2014). To avoid ringing of the whisker during the rapid deflection phases we used a sigmoid function for the onset and offset piezo movements, resulting in an effective rise/fall duration of 20 ms. The deflections were controlled via a National Instruments board (PCI 6713), using custom routines written in MatLab. To monitor and calibrate whisker movements, and confirm lack of distortion or ringing of the stimulated whisker the deflection was monitored using a laser displacement sensor (LD1605-2; Micro-Epsilon OptoNCDT 1700) (Lotem and Azouz, 2009; Lavzin et al., 2012; Garion et al., 2014). Moreover, we confirmed our calibration, lack of distortions and lack of ringing by whisker tracking with a high-speed camera (1000 fps). For this procedure, whisking movement was photographed with a high-speed camera (Flare, 4M180MCL, 4 Megapixel, Dalsa Xcelera-x4-CL, IO industries at 1000 fps) and software (Streams 6; IO industries) with resolutions 600 × 350 pixels. Movement of full-length whisker was tracked semi-manually, and the angle and curvature of the whisker were calculated as described in Garion et al., 2014, using a homemade software written in MatLab (MathWorks, NA).

2) I'm confused by the description of stimulation in the Results (200 ms ramp and hold stimuli) and in the Methods (100 ms).

In all our experiments we used a 200 ms ramp and hold stimulation. The "on response" was quantified by recording the spike count during the initial 100 ms of the ramp and hold stimulation, and the "off response" was quantified by the spike count during the initial 100 ms after termination of the ramp and hold stimulus. In one instance in the Methods section, we mistakenly wrote the ramp and hold stimulus lasted only 100 ms (angular tuning). We corrected this point in the revised manuscript. The stimulation parameters are described in the subsection “Whisker stimuli”.

II) I am skeptical about the 50 micron MEAs from neuronexus finding any single units of demonstrable selectivity in layers II/III. I certainly am not impugning the author's basic credibility (I don't think they are lying!) but getting well-isolated units in those layers, or layer IV especially, with such probes is quite hard. Without the virtue of a tetrode configuration or the use of smaller diameter contacts the claim of single unit identification is particularly suspect.

No doubt, technical issues are critical for single unit analysis. I would like to address these concerns along two parallel routes:

1. In our recordings we used NeuroNexus 177 µ2 electrodes, which are the smallest contact size electrodes by NeuroNexus. We previously used 2-5 mega Ohm commercial tungsten electrodes, and our experience is that the NeuroNexus electrodes are superior in quality. For our spike sorting we used stringent spike sorting criteria for spike sorting based on the commercial spike sorting by the OFS offline spike sorting software from Plexon. Specifically, after thresholding of the raw filtered data (>3.5 SD of the baseline) sorting was performed by semi-automatic algorithms later verified and corrected manually, if necessary. Clusters were accepted as single unit if all the following criteria were met: 1) The waveform shape remained consistent and stable throughout recording (verified by the "Sort- Quality Vs. Time" analysis in the OFS software). Units were also excluded in case the average amplitude or half width of unit changed significantly (ANOVA test) between the first and last 20% of recorded spikes. 2) The firing rate was >0.5 Hz to allow for adequate sampling. 3) The inter-spike interval (ISI) was >2 ms to reflect the absolute refractory period of neurons.

4) The ISI distribution showed a smooth exponential-like curve. 5) Finally and most importantly statistical criterion of p<0.05 (multivariate ANOVA) of cluster separation. These points were added to the revised manuscript (subsection “Electrophysiological unit recordings”).

2. To further tackle the technical challenges associated with single unit sorting we repeated all our analysis for multi-unit data (threshold >3.5 SD above baseline). We found the similar to single unit analysis, multi-unit analysis also showed that vM1 optogenetic activation both supra-linearly amplified the responses of whisker stimulation, as well sharpened angular tuning. Thus, our multi-unit data verified the single unit results. The multi-unit data was added to the revised manuscript (subsections “Electrophysiological unit recordings”, third paragraph, “Pairing optogenetic vM1 activation with passive whisker activation”, fifth paragraph and “The effect of vM1 activation on angular tuning of neurons in the vS1 barrel cortex”, third paragraph, Figures 2B and 6A). Note that in the case of angular tuning we also added date from 6 additional 90-100 day old mice.

In addition, following the reviewer’s comment (as well as other reviewers) we added examples raw data traces (Figure 1—figure supplement 1), as well as raw clustering data (Figure 1—figure supplement 2).

I also found the following criteria to feel a bit arbitrary and wasn't sure what it meant:

"For our analysis, we excluded all unstable neurons, with stability determined using two parameters. First, the baseline pre-stimulus frequency. Second, for every experimental paradigm at least one of the stimulation conditions was repeated twice during the experiment, and neurons in which the responded differentially to the two stimulation blocks were excluded."

First, what does the phrase "the baseline pre-stimulus frequency" have to do with stability? Second, multi-unit recordings can easily show run-by-run stability and, in fact, are potentially more stable than units for such recordings.

In hindsight, we agree with the reviewer that we did not sufficiently define our inclusion and exclusion criteria. For the revised manuscript, we better defined criteria for including units in our analysis, and re-analyzed all our data according to these new and better-defined criteria. The criteria for spike sorting and stability of units is fully described in our response to a previous comment by the reviewer.

So, I'd like to see the following to give me confidence:

1) Application of autocorrelogram exclusion criteria.

We performed an auto-correlogram for each unit and excluded all units, which showed a response at ± 2 ms.

2) Quantification of spike feature stability across the entire run (whether or not the response properties changed), and a discussion of which spike features were used.

We monitored the amplitude and shape of the waveform of each sorted unit using the "Sort-Quality Vs. Time" analysis in the OFS software. In addition, units were excluded in case the average amplitude or half width of unit changed significantly (ANOVA test) between the first and last 20% of recorded spikes.

3) Demonstration of the cluster quality for some number of units – I don't want to make this infinite, but as I said I'm skeptical, so at least a fair number of examples of stability in layers II/III units would be appreciated (say, 10 such examples from 10 experiments or so).

Following the reviewer’s comment we added 10 individual examples (10 electrodes with 26 units), in which both cluster quality is demonstrated. This data is presented in Figure 1—figure supplement 2 of the revised manuscript. In addition, we present raw traces of our recordings (Figure 1—figure supplement 1 of the revised manuscript).

Moderate:

I) In the 56 neurons recorded in vMI, what was the latency to firing following optogenetic stimulation, and was this a bimodal distribution (implying directly opto activated and synaptically opto activated populations).

Following the reviewer’s comment we measured the latency between laser onset (10 ms pulse) and a significant (P<0.05) increase in firing. On average, we found the latency between laser onset and increased firing in M1 to be 4.81 ± 0.43 ms (mean ± SEM, 109 neurons from 5 rats, note we performed an additional experiment and increased the number of recorded neurons). Of the recorded neurons, 41% showed a mono-modal response, 43% showed a bi or multi-modal response, and 16% did not respond to the optogenetic stimulus. This data was added to the second paragraph of the subsection “Pairing optogenetic vM1 activation with passive whisker activation”.

II) The authors should quantify the relative effect of stimulation on the optimal direction angle versus the worst direction angle, whatever these might have been. Obviously, there must be enhancement on average across randomly associated directions or else the main effect of Figure 1 couldn't stand: The sharpening obviously suggests, though, a significant different between best and worst, etc.

As suggested by the reviewer we compared the effect of vM1 activation on the selectivity to the best and worst directions. To do so we calculated the selectivity index for the preferred direction and the worst three directions. vM1 activation increased the SI to the preferred direction (as also shown in the original manuscript). In contrast, optogenetic activation of vM1 decreased the SI to the worst directions (average of the worst 3 directions). The reduction in the SI occurred despite an absolute increase in the response amplitude to both the preferred and worst angular directions. This data is presented in the second paragraph of the subsection “The effect of vM1 activation on angular tuning of neurons in the vS1 barrel cortex” and Figure 6—figure supplement 1 of the revised manuscript.

Reviewer #3:

[…] The manuscript would benefit, however, from a more detailed and extensive presentation of (raw) data and a more in-depth treatment of the crucial issue of relative input timing. Find my specific comments below.

1) Introduction, last paragraph: Why only 'partially disconnected'? Isn't the motor arm entirely uncoupled? Was this actually verified by confirming the absence of whisker movements following vM1 activation after the nerve cut?

We indeed confirmed the location of vM1 by whisker movements induced by vM1 optogenetic stimulation in rats with intact buccolabial nerve. We further confirmed lack of whisker movements by vM1 optogenetic stimulation after cutting the buccolabial nerve. We added this information to the revised manuscript (subsection “Viral vector injection and optogenetic stimulation”, last paragraph).

In our original manuscript we used the wording "To partially dissociate the vibrissae sensory-motor loop" as there are connections between the sensory and motor limbs at the cortical, subcortical and brainstem levels, while we only severed the peripheral motor nerve. Yet following the reviewer's comment, we see that these wording may be confusing, and we omitted the word "partially" from the revised manuscript.

2) Subsection “Electrophysiological unit recordings” and figures. The authors should provide some raw electrophysiology data, so that one can better judge the quality of data. What were the noise levels? How well could spike sorting be performed and on how many channels? How well could spike waveforms be separated and were there any putative fast-spiking units present? Do the reported spike rates represent changes in spike rate? It would also be interesting to see the LFP responses for the paired stimulation paradigms.

No doubt, these points are very important, and I would like to address each point separately:

1) Following the reviewer’s comments we added filtered (1-5 KHz) raw traces (Figure 1—figure supplement 1), as well as examples of sorting in 10 individual electrodes with 26 individual units (Figure 1—figure supplement 2).

2) Based on the SNR definition of Rousche and diamond, 1999, typically, our unit's SNR were in the range 7-15, and the average was 11.2 ± 0.4 (mean

± SEM). This data was added to the revised manuscript (subsection “Electrophysiological unit recordings”, third paragraph).

3) Our recording electrode contained 16 channels of which we typically analyzed 13-15 channels.

4) We greatly elaborated on our spike sorting procedures, and clarified our inclusion and exclusion criteria for units (subsection “Electrophysiological unit recordings”, for details see response to the next comment by the reviewer).

5) In our recordings high resolution sorting can be performed, and the spike waveforms could be very well separated (for example see Figure 1—figure supplement 1 and 2, and response to the next comment).

6) We did not analyze data for fast spiking neurons. We believe that with the introduction of optogenetics, solely classifying neurons according to their waveform and firing characteristics is not sufficiently accurate.

7) To tackle with the inherent technical issues associated with single unit spike sorting we repeated our analysis on multi-unit activity (events with an amplitude >3.5 SD of the baseline). The multi-unit analysis confirmed the results with single unit activity, and also showed that vM1 activation amplified the response to ramp and hold whisker activation (Figure 2 and subsection “Pairing optogenetic vM1 activation with passive whisker activation”, fifth paragraph), and sharpened angular tuning (Figure 6 and subsection “The effect of vM1 activation on angular tuning of neurons in the vS1 barrel cortex”, third paragraph).

8) Our spike count are an increase in the spikes above baseline (Spike count response-Spike count baseline). This is further explained in the Methods section of the revised manuscript.

3) Furthermore: what were the respective baseline spike frequencies and why (and how) were these used to exclude data from the analysis (subsection “Electrophysiological unit recordings”, third paragraph). I also do not understand the motivation for excluding units that did not show a consistent response pattern upon repeating the stimulation paradigm. What type of responses did these cells actually show, in how far did they deviate from a nonlinear summation results? Why should one set such bias?

Upon re-reading the manuscript, we also saw that our inclusion and exclusion criteria of units were not sufficiently clear. For the revised manuscript, we better defined criteria for including units in our analysis, and re-analyzed all our data according to these new and better-defined criteria.

Specifically, we initially recorded the multi-unit activity by recording events with an amplitude >3.5 SD above baseline from the filtered raw data (1-5 KHz). From this data we next sorted single unit activity using the OFS offline spike sorting software from Plexon. Sorting was initially performed by semi- automatic algorithms, and later verified and corrected manually. We accepted clusters as single unit if all the following criteria were met: 1) The waveform shape remained consistent and stable throughout recording. This was verified by the "Sort-Quality Vs. Time" analysis in the OFS software. Moreover, we excluded unstable units in case the average amplitude or half width of unit changed significantly (ANOVA test) between the first and last 20% of recorded spikes. 2) The firing rate was >0.5 Hz to allow for adequate sampling. 3) The inter-spike interval (ISI) was >2 ms to reflect the absolute refractory period of neurons. 4) The ISI distribution showed a smooth exponential-like curve. 5) Finally and most importantly statistical criterion of p<0.05 (multivariate ANOVA) of cluster separation.

These points were added to the revised manuscript (subsection “Electrophysiological unit recordings”). In addition to the revised manuscript, we added examples raw data traces (Figure 1—figure supplement 1) as well as raw clustering data (Figure 1—figure supplement 2). Following the reviewer comments (as well as of the other reviewers) in the revised manuscript, we no longer used the pre-stimulus firing as a criteria for including units in our analysis, but rather included all units with stable sorting and amplitude parameters.

Finally to further tackle the technical challenges associated with single unit sorting we repeated all our analysis for multi-unit data (threshold >3.5 SD above baseline). Our multi-unit data verified the single unit results. The multi-unit data was added to the revised manuscript (subsection “Pairing optogenetic vM1 activation with passive whisker activation”, fifth paragraph and subsection “The effect of vM1 activation on angular tuning of neurons in the vS1 barrel cortex”, third paragraph, Figures 2 and 6).

4) Artificial whisking paradigm (subsection “Whisker stimuli”, Figure 2, Figure 3). How exactly was the optogenetic stimulation paired with motor activation pulses? Was it a 20-ms light pulse at the beginning of the 10 pulses to the nerve?

First, we want to point out that the artificial whisking experiments served mainly as a control for the ramp and hold stimulation. As a result, we wanted to keep stimulation conditions of vM1 as similar as possible to the ramp and hold protocol.

With respect to the timing and duration of vM1 activation, for the experiment described in Figure 2 of the original manuscript (Figure 3 of the revised manuscript) we used 20 ms optogenetic pulses, which were applied at the beginning of artificial whisking (10 pulses at 100 Hz for the protraction phase). It is true that the optogenetic pulse only covered 20% of the protraction phase. However, we found that even these conditions were sufficient to supra- linearly amplify the response in the barrel cortex. In Figure 3 of the revised manuscript we clearly marked the timing of vM1 whisker activation, buccolabial nerve stimulation pulse and protraction and retraction phase (see below).

In the experiments described in Figure 4 of the revised manuscript (Figure 3 of the original manuscript) we also applied 20 ms optogenetic pulses, but at different time intervals from whisker stimulation. 0 ms time interval meant that whisker stimulation and the optogenetic stimulation were initiated simultaneously, -20 ms meant that the optogenetic stimulation was initiated 20 ms before whiskers were stimulated, and so on.

How large were the whisker protractions induced by this protocols?

Our artificial whisking typically resulted in a 45-60° protraction. We have described our protocol, including a movie showing whisker movement in Garion et al., 2014.

The spiking response appears only 20-ms long, how come? What happened in the remaining 80 ms of protraction/retraction cycle? In the eighth paragraph of the subsection “Pairing optogenetic vM1 activation with passive whisker activation” it is stated that spikes were evoked during protraction and retraction phase, but I can't see that. More details are needed here.

In the original manuscript, we only showed a fraction of the protraction phase. In Figure 3 of the revised manuscript, we show a full protraction and retraction cycle, and the timing of buccolabial nerve stimulation and vM1 optogenetic stimulation. Typically, the protraction phase shows a short response; possibly due to the fact whiskers reach their final position before the end of the 100 ms protraction stimulation (for details see Garion et al. 2014).

5) Subsection “Viral vector injection and optogenetic stimulation”, first paragraph. How precisely was the vM1 hit with the injections and the light stimulation? Was this verified by measuring whisker movements (or other movements?) by optogenetic vM1 stimulation before the facial nerve was cut?

We localized vM1 by published coordinates, and confirmed the location of vM1 by monitoring whisking movements during optogenetic stimulation. We found the optogenetic stimulation of vM1 evoked whisking movements in rats with intact uncut buccolabial nerves. This data was added to the last paragraph of the subsection “Viral vector injection and optogenetic stimulation”.

6) I find the timing experiments very interesting but I am not sure how to interpret the peak revealed at -20 ms. This is a time scale where the conduction delays etc. play an important role. So how was the timing exactly defined (onset of piezo drive and onset of LED illumination, I presume)?

In these experiments vM1 was stimulated with 20 ms laser pulses. We chose the 20 ms pulse after examining the effect to 5-20 ms laser pulse duration (Figure 1E). The relative timing between vM1 stimulation and whisker stimulation was defined as T=onset time of laser pulse-onset time of whisker stimulation (see response to previous comment).

Was there any dead time for mechanical stimulation of the whisker considered? In particular, how was the timing defined for the artificial whisking stimulus? Were the whisker movements (presumably stick-slip events) monitored and their timing analyzed? Were any axonal conduction times and synaptic delays taken into account? Obviously, these questions are important to understand what the real timing difference at the integrating neurons in barrel cortex might be. Supplementary whole-cell recordings could be very helpful here, to quantify when exactly inputs from both pathways actually arrive with these stimulation paradigms.

The points raised by the reviewer are of course very valid and important. In our experiments we only controlled for the timing stimulation was triggered, but not for synaptic or transmission delays, nor for the timing of stick and slip events during artificial whisking. For that reason we used relatively long optogenetic stimulation (20 ms), and concentrated on the on response and initial protraction phase (90-100 ms). We further stressed this point in the last paragraph of the subsection “Temporal roles governing the interactions between vM1 and whisker sensory inputs”.

7) Angular tuning in barrel cortex apparently is a complicated matter and among other things seems to especially depend on age (Kremer study). In addition, the housing conditions (use of whiskers) may affect the outcome. The age of rats used here is just in between the ages when no angular tuning was observed and when it later was established. I find the examples in Figure 4 not convincing, as the pure whisker-evoked responses appear relatively untuned and the responses enhanced by optogenetic vM1 activation mostly display multiple peaks (often in orthogonal directions). Cells from what layer are actually shown in Figure 4? It might be helpful to show distributions of the absolute SI values for the different stimulation protocols.

Following the reviewer’s comments we:

1) Performed additional experiments in which we repeated our angular tuning experiments in 90-100 day old rats (n=6). These experiments again showed sharpening of angular tuning by vM1 optogenetic activation. Interestingly we saw no significant difference in the control angular tuning between 50-60 day old and 90-100 day old rats. This new data is presented in Figure 5, Figure 5—figure supplement 1, Figure 6, Figure 6—figure supplement 1 and in the second paragraph of the subsection “The effect of vM1 activation on angular tuning of neurons in the vS1 barrel cortex”.

2) In Figure 4 of the revised manuscript, we now show four examples from the original set of experiments (50-60 day old) and two additional examples from the new experiments (90-100 day old). In addition, we added the putative layers of the recorded neurons.

3) We believe some of our examples reveal angular tuning, which are consistent with previously published data on the phenomenon, including the existence of more than one peak in the angular tuning curve (See Bruno et al., 2003; Andermann and Moore, 2006). We also show an example of a unit that showed minimal angular tuning prior to opto-stimulation, and transformed into an angular tuned neuron during opto-stimulation.

4) As suggested by the reviewer we added a histogram of the SI values with and without laser activation (Figure 6B of the revised manuscript), as well as the effect of vM1 activation on the SI of both preferred direction and worst three directions (Figure 6—figure supplement 1).

[Editors' note: the author responses to the re-review follow.]

Essential revisions:

1) One reviewer states: "I appreciate the addition of details on how the stimulator was calibrated-these are all the right approaches and tools. I can guarantee that using these parameters, that unless they use a specific compensatory algorithm, they almost certainly cannot get ringing under 5% of stroke magnitude, and I doubt it can be under 10%. I want the following included in the final manuscript → An actual analysis of the mean amplitude of the ringing. Saying you did not see it does not reflect quantification of the effect. Say how large in angle of vibrissal base motion and at what frequency the ring is on average across trials and across the different stimulators used." The reviewer notes that this is likely to have an effect on the direction selective responses.

Ringing of piezo bimorphs is indeed significant problem. Regarding this technical issue, we wish to note that we have already published two papers with the same stimulation system in two very distinguished journals, eLife and Nature. Thus, while writing the previous version of the manuscript that our description of our stimulation protocol combined with quotations of our prior papers would be sufficient.

Following the reviewer's comments and requests we have further elaborated on the technical details of our ramp and hold stimulation protocol, and added experimental data regarding the movement and ringing of the piezo bimorphs during ramp and hold stimulation, as monitored with a high-speed camera (1000 fps). In the revised manuscript we (1) present the formula by which we smoothened the onset and offset of the ramp and hold stimulation pulse. (2) We show single traces of the piezo bimorph movements during ramp and hold pulses smoothened to different degrees. (3) Finally, we present the averaged ringing amplitude of the piezo bimorphs at different degrees of smoothening of the ramp and hold pulses (different smoothening factors).

This data is presented in Figure 1—figure supplement 3 and in the subsection “Vibrissa stimuli”. As shown by the data we succeeded in minimizing ringing in our experimental conditions.

2) A second reviewer felt that the documentation of the degree of direction selectivity was inadequate. They did not feel the statement "Consistent with previous results we found angular tuning in neurons.…" (subsection “The effect of vM1 activation on angular tuning of neurons in the vS1 barrel cortex”, first paragraph) was well supported. They felt that the criteria for when a cell's response is considered well tuned were not clear and that the SI as calculated made comparisons across neurons with very different response amplitudes difficult. It is suggested that the authors "confirm their interpretation with an alternative, more robust analysis method of direction-tuning, based on the mean response vector not the Rmax (Kremer et al. 2011; Mazurek et al., Front Neural Circ 2014). And provide a statistical argument for their statement 'we found angular tuning'."

Following the reviewer's comments, we performed further analysis and quantification of angular tuning. As eloquently presented by Mazurek et al., 2014, quantification of selectivity is a complex issue. In the original manuscript we chose the SI as our main quantification parameter. We introduced two additional parameters:

1) Angular tuned neurons, as defined by a significant difference at the 0.05 level for comparison of the responses to the preferred angle and to the three least preferred angles. We found that about 60% of S1 neurons showed angular tuning, with no significant differences between putative layers 2-5. This data was added to the subsection “The effect of vM1 activation on angular tuning of neurons in the vS1 barrel cortex”.

2) Vector sum analysis for angular tuning. We used a second parameter for quantifying angular tuning, the vector sum. In turn, we used the vector sum to examine the effect of vM1 activation on angular tuning. We found that similar to SI analysis optogenetic activation vM1 significantly increased the amplitude of the summed vector. This data was added to Figure 6 (panel 6C) and at the end of the subsection “Vibrissa stimuli” and “The effect of vM1 activation on angular tuning of neurons in the vS1 barrel cortex”.

3) One of the reviewers also notes that "laser stimulation of vM1 alone in essentially all cases did not evoke spiking activity at all (except perhaps for Figure 1B?). Thus, the 'supralinear' effect essentially consists in an upregulation or facilitation of the sensory-evoked response in vS1, which could be simply explained by additional pre-depolarization mediated by the M1-to-S1 projections, given vM1 is stimulated at the right time briefly before the sensory stimulus. While the authors mention this simple explanation (vM1 projection fibers helping vS1 neurons to reach the – nonlinear – spike threshold), they only refer to it as 'additional cellular mechanism' in the third paragraph of the subsection “Possible mechanisms underlying vM1 mediated response amplification and sharpening of angular tuning in vS1 barrel neurons”. Their primary 'attractive potential mechanism' of dendritic amplification (in the first paragraph of the aforementioned subsection) in my view is, however, largely speculative." The authors should consider toning down their use of 'supralinear responses' and may consider using terms like 'facilitation' or 'modulation.' At the very least they should give more equal weight to the simplest interpretation of these effects.

We agree with the reviewer that "simple" effects on the axonal initiation zone can explain our findings, and have discussed this possibility in the previous version manuscript. Yet we feel that "simple effect" of vM1- mediated EPSPs on the axonal non-linearity of vS1 neurons is less likely to explain the large effect we found. However, as we did not examine this issue experimentally, and following the reviewer's comments, we down toned the dendritic amplification scenario in the revised manuscript.

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    Source code 1. Matlab homemade software.

    DOI: http://dx.doi.org/10.7554/eLife.21843.013

    elife-21843-code1.zip (19.3KB, zip)
    DOI: 10.7554/eLife.21843.013

    Articles from eLife are provided here courtesy of eLife Sciences Publications, Ltd

    RESOURCES