Sensorimotor integration and amplification of reflexive whisking by well-timed spiking in the cerebellar corticonuclear circuit

Purkinje cell activity associated with sensory input to the whisker pad.

Previous studies have reported that crus I/II Purkinje cells respond to sensory input from whisker deflections (Bosman et al., 2010; Chu et al., 2011; Proville et al., 2014) and are also active during spontaneous whisking (Chen et al., 2016). Here, we examined Purkinje cell firing patterns in sensory-evoked whisker movements in awake, head-fixed mice acclimated to sitting in a non-constraining tube. We stimulated the ipsilateral whisker pad with 10-ms 20-psi air puffs while making loose cell-attached recordings of action potentials by crus I/II Purkinje cells (Figure 1A). Recordings were obtained from electrode penetrations into lateral crus I (coordinates relative to bregma: caudal 6.0–6.5 mm, lateral 3.0–3.4 mm) at depths of 150–2000 μm. Purkinje neuron identity was confirmed by the detection of both simple and complex spikes, which were analyzed separately (Figure 1A). In these experiments (N=5 mice), all whiskers except the ipsilateral C2 (or in one case, C1) whisker were trimmed, and whisker position along the rostral-caudal axis was monitored for correlation with Purkinje cell activity.

The puff initially deflected the whisker backward, and then reliably evoked a whisker protraction that outlasted the stimulus for a few hundred ms (Figure 1A). A subset of experiments with whiskers present on both sides confirmed that movements were bilateral. We began by examining spike rates; spike timing is considered below. Instantaneous simple spike firing rates in Purkinje cells were modulated in conjunction with whisker movements, with some cells initially increasing and others decreasing their activity (Figure 1B, n=26). Because baseline firing rates for Purkinje cells during non-whisking periods ranged widely, from 40 to 140 spikes/sec, data were analyzed as the change in firing rate and compared to the change in whisker position. Superimposing the grand mean instantaneous firing rate of all Purkinje cells, and the grand mean of the associated whisker protractions, illustrated a match between the population Purkinje cell response and movement kinematics (Figure 1C, top). A cross-correlation of the mean change in firing rate of all cells with the mean change in whisker position revealed that the increase in firing rate preceded the onset of whisker protraction by 15 ms (Figure 1C, bottom). Linear regression over the points incorporating that lead time gave a slope of 0.98 Hz/degree of protraction with a correlation coefficient (r) of 0.93, plotted as the time-shifted Purkinje firing rate vs. whisker position for the first 750 ms of the movement (Figure 1C, bottom). The population mean also showed a robust increase in complex spike probability time-locked to the onset of the evoked movement (Figure 1D), illustrating that many individual cells fired complex spikes nearly simultaneously. These results indicate that the population simple spike response of crus I/II Purkinje cells carries a linear representation of reflexive whisker movements, that simple spike rate changes precede the protraction, and that complex spike rates increase during these movements. The observations are consistent with a recent study of spontaneous whisking (Chen et al., 2016), and also resemble studies of other motor behaviors, which show Purkinje cell spike modulation in conjunction with, often just preceding, reaching, eye movements, and locomotion (Medina 2011).

The linear relation between the changes in Purkinje cell firing rate and whisker position raises the question of to what extent sensory signals and/or motor commands are encoded in Purkinje activity evoked by air puffs. We reasoned that applying the puff contralaterally would alter the nature of the tactile signal to Purkinje cells, so the simple spike pattern would change substantially if the Purkinje cells activity encoded the location of the sensory stimulus. We therefore compared responses of Purkinje cells to an ipsilateral and contralateral puff. On average, the contralateral puff elicited a larger maximal protraction of the single remaining ipsilateral whisker (Figure 1E, top, ipsilateral puff, 29.6 ± 1.2°; contralateral puff, 42.2 ± 3.8°, p =0.01, paired t-test). With a contralateral puff, the magnitude of the ipsilateral movement was indistinguishable during recordings with and without a contralateral whisker (44.3 ± 1.9°, n=29 cells; 42.2 ± 3.8°, n=10 cells; p=0.6). The larger protraction with a contralateral puff is consistent with reports of asymmetrical whisker movements associated with head turns in freely moving rats, which suggest that contralateral stimuli evoke larger ipsilateral whisking movements just before rats turn toward the stimulus (Towal and Hartmann, 2006).

Despite this difference in whisking amplitude, Purkinje cells continued to generate responses regardless of the puff location (Figure 1E, bottom); for both ipsilateral and contralateral stimuli, mean changes in firing rate in 50-ms bins for 500 ms following the puff were indistinguishable (n=10, all bins, p>0.9, paired t-test, Bonferroni-Holm adjusted p-values). Likewise, the well-timed increase in complex spikes persisted with a contralateral puff (Figure 1F). The similarity of the responses to ipsilateral and contralateral puffs make it seem unlikely that Purkinje cell activity reports sensory input in a manner compatible with fine sensory discrimination. Instead, it seems more likely to relate to factors independent of stimulus location, e.g., either the sensory feedback or motor command associated with bilateral whisker movements.

Purkinje cell activity associated with changes in sensory feedback.

To test the extent to which Purkinje cell firing rates depended upon sensory feedback during reflexive whisking, we removed ipsilateral reafference by suppressing movement with 20 μL lidocaine (2% solution) applied to the ipsilateral facial nerve (Fee et al., 1997). Purkinje cell activity in response to contralateral air puffs was then compared in each cell, before and after lidocaine infusion. In these experiments, all whiskers were kept intact to maximize whisking-related sensory feedback in the control condition. Recordings of whisker position confirmed that lidocaine infusion prevented all movement of the ipsilateral whiskers (Figure 2A, 2B, deflection in control = 33.8 ± 2.6°, in lidocaine at matched time = 0.4 ± 0.6°, N=4 mice, p<3e-9, paired t-test), although the contralateral whiskers continued to whisk normally. Purkinje cells, however, continued to respond to the air puff in the absence of ipsilateral sensory feedback (Figure 2C).

Despite the complete loss of movement of the ipsilateral whiskers, the population instantaneous firing rates superimposed (Figure 2D), and the mean changes in firing rate in each 50-ms bin for 500 ms following the puff were indistinguishable (n=16, all bins, p>0.75, paired t-test, Bonferroni-Holm adjusted p-values). The data indicate that extreme changes in reafference do not greatly alter simple spike firing rates in response to puff stimulation, suggesting that the initial puff triggers motor-related signals carried through the Purkinje cells.

Contributions of brainstem and cortical circuits to Purkinje cell firing and sensory-evoked whisking.

To identify the origin of the putative motor signal, we tested whether Purkinje cell responses and/or puff-evoked movements could be altered by transiently suppressing output from brainstem and/or cortical neurons, both of which project to the cerebellum via mossy fibers (Figure 3A, Proville et al., 2014; Tang and Zhang 1987; Yatim et al., 1996). In VGAT-ChR2 mice, an optical fiber was placed in the ipsilateral reticular formation (Figure 3B), in the vicinity of the whisking-related premotor nuclei, including the whisking central pattern generator located in the intermediate reticular nucleus (IRt) (Moore et al., 2013; Takatoh et al., 2013). VGAT-expressing inhibitory neurons in this region were optically stimulated for 500 ms at 50–100 Hz, and a puff was delivered contralaterally at light onset. In response to this stimulation, whisker movements were greatly suppressed (Figure 3C). On average, the peak evoked whisker protraction was reduced from 42.3 ± 2.1° to 9.8 ± 2.1° (whisks associated with recordings from n=16 cells, p<1e-11, paired t-test, Figure 3D, top), consistent with a successful inhibition of whisking-related premotor centers in the brainstem. In the associated Purkinje cell population responses, the firing rates were decreased significantly in the 50-ms bins between 50 and 200 ms after the puff (n=16, p<0.006, paired t-test, Bonferroni-Holm adjusted p-values), suggesting that the late part of the signal indeed depended on activity in the reticular nuclei (Figure 3D, bottom). The earliest part of the Purkinje cell response was unchanged, however, consistent with crus I/II receiving mossy fiber input directly from the trigeminal nuclei (Yatim et al., 1996). Light alone without the puff generated no movement and did not change Purkinje cell firing rates, verifying that the stimulus did not activate VGAT-positive neurons in the molecular layer of the cerebellum that could directly inhibit Purkinje cells (Figure 3D, bottom). Additionally, light applied directly to the cerebellar cortex led to an increase in Purkinje cell firing rates (Figure 3E), providing further evidence that the reduction in Purkinje cell spiking resulted from inhibition of brainstem afferents to the cerebellum, rather than any direct effects in the cerebellar cortex.

Because anatomical and physiological evidence indicates that vibrissa motor cortex (vM1) projects to the contralateral cerebellar cortex via the pontine nuclei (Proville et al., 2014; Figure 3A), we next repeated the experiment with the optical fiber placed on the surface of contralateral vM1. The relatively small sample size likely gave only a coarse approximation of the control population response; however, optogenetically suppressing vM1 neither reduced puff-evoked whisker movements (deflection in control: 38.7 ± 1.1°; deflection in vM1 inhibition: 38.7 ± 1.9°; p>0.99 paired t-test) nor significantly changed simple spike rates in Purkinje cells measured (Figure 3F, all bins, n=8, p>0.9, paired t-test, Bonferroni-Holm adjusted p-values). Plotting the mean firing rates for the first 250 ms of stimulation with and without optogenetic silencing against each other gave a high linear correlation (r = 0.95), slope near 1 (m=0.96), and intercept near 0 (b=0.9 spike/sec) for vM1 inhibition but a lower correlation and an altered slope and intercept for brainstem inhibition (0.92, 0.79, −3.37; Figure 3G), illustrating the dependence of Purkinje cell activity on the reticular nuclei but not vM1.

Contributions of cerebellar output to sensory-evoked whisking.

The observation that Purkinje cell firing rates correlate with movement raises the question of whether this activity exerts an influence over reflexive whisking. We therefore directly tested whether the puff-evoked protraction was sensitive to elevations in Purkinje cell firing. In mice expressing channelrhodopsin in Purkinje cells (Pcp2-ChR2, Methods; N = 4 mice), crus I/II Purkinje cells were optically stimulated for 200 ms, and the puff was delivered contralaterally at light onset. Optical stimulation greatly accelerated firing by Purkinje cells (Figure 4A, n=5 cells). The light stimulus also reduced the amplitude of the puff-evoked whisker protraction (Figure 4B, 4C, left). The mean data illustrated that silencing cerebellar output suppressed the puff-evoked whisker protraction by about 15°, which accounted for about a third of the total movement (Figure 4C, right, deflection with no-light control = 46.7° ± 1.8°, with light = 32.6°± 3.2°; comparing deflection ± light, p=0.005, N=4 mice). This value is likely to be an underestimate, since it is unlikely that all relevant Purkinje cells were activated by optical stimulation. Differentiating the whisker position record indicated that the peak velocity of the deflection was reduced as well (control vs. light: 1.7 ± 0.1°/ms vs. 1.5 ± 0.1°/ms, p=0.04, N=4 mice). At light offset, an additional whisker protraction that was absent from control trials was also generated (Figure 4D, deflection at light offset = 17.8° ± 3.4°; at matched time point in no-light control = −2.6° ± 1.0°, ± light, p = 0.01 N=4 mice); similar protractions upon the offset of Purkinje cell stimulation have been observed previously, and are attributed to increased activity of CbN cells (Proville et al., 2014). Thus, cerebellar output accounts for at least ~30% of the magnitude of sensory-evoked, apparently reflexive whisking.

Activity of Purkinje target neurons associated with sensory input to the whisker pad.

The suppression of movement by Purkinje cell stimulation is somewhat paradoxical, however, given the match between population Purkinje cell firing rate and movement kinematics. In particular, the greatest elevation of Purkinje cell activity is associated with the greatest deflection of the whiskers, raising the question of how Purkinje-mediated inhibition of CbN neurons normally sculpts the trajectory of whisker protractions. To investigate this question, we made recordings from CbN cells in the lateral cerebellar nucleus of Pcp2-ChR2 mice and measured their responses to contralateral puffs. All CbN cells responded to the sensory stimulus with a short-latency, brief increase in firing rate that abruptly returned to or fell below baseline within 35 ms (n=8 cells, Figure 5A, 5B). Superimposing the mean Purkinje cell response to a contralateral puff from all experiments (n=65 cells) illustrated that the population firing rates of CbN cells and Purkinje cells were both initially elevated. Purkinje cells, however, showed a late phase of increased spiking (after ~25 ms), during which time the CbN cell firing rates dropped below pre-stimulus values (Figure 5B). The initial surge in the CbN response coincided with a drop in the Purkinje response, and the maximal firing rate increase in Purkinje cells occurred between the two peaks in CbN cell firing rates. These observations motivated a closer examination of the spike timing in both sets of cells (below).

First, however, we tested whether optogenetic stimulation of Purkinje cells was sufficient to suppress the increase of CbN cell firing early in the record, since this aspect of the CbN cells response appeared most likely to contribute to the magnitude of the whisker protraction. During recording from CbN cells, we therefore applied light as in previous experiments. Indeed, in all cells (n=5), light stimulation completely eliminated firing (Figure 5C, 5D). This silencing was followed at light offset by a firing rate increase. This increase corresponded to the protraction at light offset, consistent with the idea that elevating CbN cell activity can generate whisker movement. Silencing of CbN cells by the light stimulus occurred whether or not the puff was applied concurrently with the light; however, trials with the puff included ‘ripples’ early in the loose-cell-attached record (Figure 5C, bottom). To examine the ripples more closely, the responses across cells to light applied either alone or with the puff were averaged and subtracted (Figure 5E). We reasoned that these ripples probably resulted from excitatory input and that they could therefore be used to estimate the time course of the mossy-fiber-mediated puff-dependent input to CbN cells.

Timing of Purkinje and CbN cell spikes.

With this information in hand, we next re-examined the timing of action potentials elicited in Purkinje and CbN cells by contralateral puffs. Mean PSTHs with 2-ms bins were constructed from the responses of all 65 Purkinje cells tested (Figure 6A). The plots showed evidence of brief, well-timed windows of alternating decreases and increases in spike probability (on the time scale of milliseconds) at the onset of the response to the puff, across all trials of all neurons. Although it cannot be ascertained which subset of these Purkinje cells converge on CbN cells, the fact that reductions in Purkinje cells firing probability were consistently time-locked to the puff stimulus suggested that CbN cells were likely to be subjected to relatively coherent decreases or gaps in inhibitory input.

Applying the same analysis to CbN cells in response to a contralateral puff revealed two well-timed peaks (Figure 6B), which were also present in PSTHs of individual cells (Figure 6C). Superimposing the Purkinje and CbN cell histograms, along with the putative mossy-fiber-mediated excitatory response, illustrates several points (Figure 6D). First, Purkinje cells showed a consistent decrement in spike probability in a 4-ms window, followed by a 4–6 ms elevation of simple and complex spikes. Second, periods of increased CbN cell firing probability corresponded to the well-timed window in which Purkinje-mediated inhibition to CbN cells would have collectively dropped. Third, the 2–4 ms drop in CbN cell firing probability between the double peak, which was present in all the individual records, matched the window of peak Purkinje cell spike probability. Fourth, the CbN cell firing probabilities fell below baseline concurrently with the second surge in Purkinje cell firing, particularly during the broad peak in complex spike probability. Finally, ~40 ms after the puff, coherent spiking by the population was no longer evident, although Purkinje firing rates remained elevated. In contrast, plotting the same data in broader bins, which do not reveal millisecond-scale spike times, illustrated that Purkinje and CbN cells both increased their firing rates within the first 50 ms following the puff (Figure 6E). Additionally, a spike timing analysis of Purkinje cell responses to ipsilateral puffs to showed that this mean population response (n=26 cells) also had cycles of well-timed (2–4 ms) suppressions and elevations of firing probability (Figure 6F), suggesting that cells on both sides of the cerebellum fire coherently to the puff stimulus.

Together, these observations suggest that sensory input elevates mossy fiber activity, which provides excitatory drive to CbN cells and also generates precisely timed spike suppression and spike generation early in the Purkinje cell response. As a result, Purkinje and CbN cell action potentials rapidly alternate, such that cerebellar output is actually increased. A second burst of less-well-timed complex spikes in Purkinje cells, along with elevated but apparently non-coherent simple spike firing, truncates CbN cell responses, leading to a longer period in which Purkinje and CbN cell firing rates are anti-correlated.

Purkinje and CbN cell responses during spontaneous whisking.

The presence of peaks in the CbN PSTHs and instantaneous firing rate traces raise the possibility that these brief elevations of cerebellar output generate the increased amplitude of protraction that is lost with optogenetic stimulation of Purkinje cells. To test whether such activity was specifically related to sensory-evoked whisking or is a general phenomenon, we measured responses of the same Purkinje and CbN cells during spontaneous non-contact whisking. Spontaneous whisking occurred intermittently in all mice throughout the recording sessions. Comparisons were made of responses associated with puff-evoked and spontaneous whisking in those recordings in which mice had been subjected to contralateral puffs and which had >10 spontaneous whisking events (n=51 Purkinje cells and n=8 CbN cells). Averaging the mean whisker deflections showed that spontaneous whisks were smaller than evoked whisks, with a peak magnitude that was 49.7% of those evoked by the puff (Figure 7A, top). The activity of Purkinje cells, triggered to whisk onset, was reliably modulated during spontaneous movements, but the change in Purkinje cell firing rates was correspondingly lower (Figure 7A, bottom). The relationship between firing rate changes and whisker position changes, however, remained highly correlated for spontaneous (r = 0.73) and puff-evoked whisks (r = 0.97), with gains of 0.7 Hz/° protraction and 1.0 Hz/° protraction, respectively (Figure 7B). The Purkinje cell response led the spontaneous movement by 27 ms, vs. 20 ms in response to the puff. Together the responses of Purkinje cells were comparable for spontaneous and sensory-evoked whisks, with the former simply scaled down according to the whisk amplitude.

In contrast, the CbN cell response was qualitatively as well as quantitatively different in association with spontaneous whisks. Unlike the 80–100 spikes/sec elevation of instantaneous firing rate evoked by the puff, CbN cells on average reduced their rates by about 5 spikes/sec during spontaneous whisks (n=8, Figure 7C), with some cells showing no detectable response (n=3/8). Superimposing the Purkinje and CbN responses illustrated that Purkinje cell and CbN cell activity did not co-modulate at any point in the spontaneous whisks (Figure 7C, Inset).

Plotting the data as PSTHs with 2 ms bins demonstrated that neither the action potentials of Purkinje cells nor of CbN cells showed evidence of precise timing during spontaneous whisks, since no bin or cluster of bins showed a clear increase or decrease in the number of spikes (Figure 7D). These data suggest that spontaneous non-contact whisking does not greatly depend upon changes in cerebellar output. These results are consistent with previous findings that spontaneous whisking persists even with optogenetic silencing of CbN cells, although whisker set point is altered (Proville et al., 2014). They further suggest that sensory input leads to well-timed Purkinje cell action potentials. The resulting cycles of inhibition and disinhibition of CbN cells are compatible with excitation-driven, well-timed, brief periods of elevated CbN activity, which can account for the ability of cerebellar output to increase the magnitude of reflexive whisker protractions.

Relating whisker movement and Purkinje cell firing.

The population response of simple spike rates of crus I/II Purkinje cells has been shown to carry a linear representation of spontaneous whisker movements, preceding the protraction by a few tens of ms (Chen et al., 2016). Here, we further find that the change in firing rate correlates well with the deflection magnitude, with a gain of 1 Hz/degree, over about 30° of protraction, for puff-evoked as well as spontaneous whisks. Such Purkinje cell population responses that match movement kinematics have been observed for several types of movements, including locomotion, reaching, the vestibulo-ocular reflex, smooth pursuit, and saccades (Medina 2011). For saccades in particular, the activity of Purkinje cells with distinct categories of responses during movement (bursts and pauses) can combine to match movement kinematics precisely, especially when responses are sorted to have common complex spike receptive fields (Herzfeld et al., 2015).

Cerebellar contribution to sensory-evoked whisking.

Experimental activation of CbN cells can generate movements in stationary animals (Noda and Fujikado 1987; Heiney et al., 2014; Proville et al., 2014; Lee et al., 2015). Conversely, silencing cerebellar output pharmacologically, electrically, or optogenetically during movements impedes learned eyelid closures, reaching, grasping, and locomotion (Hesslow 1994b; Goodkin and Thach 2003; Sarnaik and Raman 2018). Regarding whisker movements, optogenetic stimulation of Purkinje cells, which silences CbN cells, shifts the whisker set point during whisking (Proville et al., 2014). Here, we find that reflexive whisking in response to air puffs is substantially reduced by suppression of cerebellar output. The full effect may be even larger than seen here, since it is unlikely that all the relevant target cells in the cerebellar nuclei were silenced.

Consistent with the idea that CbN cells amplify reflexive whisker protractions, CbN cell output is greatly elevated during whisking elicited by sensory stimulation. The firing rate changes of CbN cells, however, do not correlate with whisker position throughout the movement, as the increase in their output is only transient. The results suggest that puff-related excitation is appropriately timed to drive the early rise in CbN cell spike probability as well as the early suppression of Purkinje cell spikes, e.g., via granule cells to molecular layer interneurons (Chu et al., 2011) or Purkinje-Purkinje connections, which likely augments the CbN cell response. The brief elevation of Purkinje cell simple spike probability correlates with a 2-ms suppression of CbN cell spikes, and the subsequent brief drop of Purkinje cell spike probability, while mossy fiber excitation remains high, is likely permissive for the second surge in CbN cell spiking. After the second cycle of elevated Purkinje cell firing, which includes many less-well-timed complex spikes, CbN cell spike probability falls below baseline.

Purkinje cell spike patterns and cerebellar output.

These data are relevant to the question of whether synchrony of Purkinje cell simple spikes occurs physiologically and/or affects motor behaviors (de Zeeuw et al., 2011; Person and Raman 2012b). Such millisecond-scale synchrony has been reported repeatedly in anesthetized and awake animals (e.g., Bell and Grimm 1969; Heck et al., 2007; Shin and de Schutter 2006; de Solages et al., 2008; Wise et al., 2010), and in vitro studies have shown that large CbN premotor neurons have brief IPSCs (~2.5 ms), such that even high-frequency synchronous inhibitory postsynaptic potentials (IPSPs, ~100/sec) are followed by synchronous gaps in inhibition during which CbN firing probability can be raised (Person and Raman 2012a; Wu and Raman 2017).

The present results are consistent with a similar scenario occurring in vivo at the onset of the response to the puff. Averaging across all 65 Purkinje cells tested with contralateral puffs, 2–4 ms decreases and increases in spike probability were evident. Such correlated spike times across so many neurons seem most likely to arise from synchronous activity and are consistent with proposals that synaptic inhibition of Purkinje cells initiates simple spike synchrony (de Solages et al., 2008). Although it is not known which Purkinje cells in the present study might converge, coherent spike suppression across the population is necessarily reflected in reduced inhibition downstream. The brief windows of CbN cell spiking indeed correlate with the short intervals of Purkinje cell spike suppression and overlap with elevated excitation. Thus, the present data provide in vivo evidence that even millisecond-scale concerted reductions in Purkinje cell firing can make inhibition effectively ‘get out of the way,’ permitting excitation to drive CbN cell firing more strongly (Wu and Raman 2017).

Complex spikes were also temporally organized in response to puffs, with an early smaller increase at the time of high simple spike probability, and a later, larger increase that is spread over 14–16 ms; similar complex spike synchrony on the time scale of tens of milliseconds is well established (Welsh et al., 1995; Wylie et al., 1995; Ozden et al., 2012). Since each complex spike produces 2 or more propagating action potentials (Khaliq and Raman 2005; Monsivais et al., 2005), both the rate and the temporal dispersion of IPSPs is expected to increase, both of which would increase the efficacy of inhibition of CbN cells. Indeed, the 14–16 ms window correlated with the strongest suppression of CbN cell firing, suggesting that complex spikes may play a disproportionate role in inhibiting CbN cells in real time and thus truncating their effect on movement.

Unlike during puff-evoked whisks, CbN cell output was minimal during spontaneous whisks, consistent with previous reports that optogenetic activation of Purkinje cells did not change spontaneous whisking frequency, amplitude, or speed (Proville et al., 2014). The lack of much change in cerebellar output, despite Purkinje cell activity matching whisker movements (Chen et al., 2016) is reminiscent of cancellation circuits described in cerebellum-like structures of fish, which silence outputs to self-generated inputs (Requarth and Sawtell 2014); here, a similar computation that suppresses cerebellar outputs during well-predicted movements might take place at the level of the CbN. Specifically, the motor command component of the Purkinje cell response may cancel cerebellar output in the absence of unpredicted sensory input.

Origin of signals.

The match between Purkinje cell firing and whisking kinematics during both spontaneous and reflexive whisking makes it seem likely that input from motor regions contributes to the response. Optical illumination of vM1 in VGAT-ChR2 mice, which inhibits cortical output (Guo et al., 2014), reduces spontaneous whisking (Sreenivasan et al., 2016). When vM1 is pharmacologically inactivated with muscimol, however, Purkinje cell modulation remains during spontaneous whisking (Chen et al., 2016). Likewise, no change was seen here in Purkinje cell activity during reflexive whisking with vM1 inhibited. In contrast, optogenetic inhibition of the ipsilateral reticular nuclei reduced Purkinje cell firing rates in the later period of evoked whisking, consistent with the idea that these spikes are driven by internal motor commands from whisking-related premotor circuitry in the brainstem (Moore et al., 2013; Takatoh et al., 2013). Also, the movement-related Purkinje cell responses for sensory-driven and spontaneous whisking resemble one another, as reported for larval zebrafish Purkinje cells during spontaneous and sensory-evoked fictive swimming (Harmon et al., 2017). The data are consistent with models in which the simple spikes of Purkinje cells produce a motor command-related signal that is predictive of movement (Medina 2011).

Mossy fibers also carry sensory input (Bower and Woolston 1983; Ishikawa et al., 2015; Chabrol et al. 2015), and any sensory signals from the puff (Gao et al., 1996, Parsons et al., 1997) and/or reafferent signals from the whisker movements that reach the cerebellum would likely be carried through mossy fibers from the trigeminal nuclei; these cells receive sensory input from the whisker pad and project primarily ipsilaterally but also contralaterally to the cerebellum (Armstrong and Drew 1980; Yatim et al., 1996; Severson et al 2017). The present work indicates that neither the site of the puff nor removal of sensory feedback significantly affects Purkinje cell firing rates associated with reflexive whisking. Nevertheless, the presence or absence of a puff greatly changed Purkinje cell spike timing at the onset of whisking-associated responses, which also varied with puff location. Thus, the earliest part of the Purkinje cell response, which was unaffected by inhibiting the reticular nuclei, seems likely to represent direct trigeminal mossy fiber input associated with the puff.

Indeed, even in anesthetized rodents, brief sensory stimulation of the whiskers and nearby facial regions alters simple spiking by Purkinje cells for tens of milliseconds. For example, in pentobarbitone-anesthetized rats, direct electrical stimulation of vibrissal follicles evoked short latency increases in simple spikes in crus I, crus II, the paramedian lobule, and lobule IX (Armstrong and Drew 1980). In urethane-anesthetized mice air puffs applied to the ipsilateral whisker pad inhibited crus II Purkinje cells for ~100 ms (Chu et al., 2011), and in ketamine/xylazine-anesthetized mice piezoelectrically driven sustained displacement of ipsilateral whiskers variably increased or decreased crus I and crus II Purkinje cell simple spiking in on the time scale of 50 ms (Bosman et al., 2010). In awake mice, we find that sensory stimuli evoke active whisking and Purkinje cell responses are prolonged, suggesting that general anesthesia suppresses the late response, which likely originates from brainstem motor centers. These brainstem nuclei are evidently sufficient to drive whisking independently of the cerebellum, since CbN cell activity is slightly decreased during spontaneous whisking, likely resulting from elevated, non-coherent Purkinje-mediated inhibition. In puff-evoked whisking, however, sensory input drives both direct excitation and well-timed reductions of Purkinje-mediated inhibition to CbN cells, transiently elevating CbN firing rates and amplifying reflexive whisker movements.

Contact for reagent and resource sharing

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Indira M. Raman (i-raman@northwestern.edu)

Experimental model and subject details

All experiments were done in accordance with federal and institutional guidelines and were approved by the Northwestern University IACUC, Animal Welfare Assurance Number, A3283–01, protocol number IS00000242 (IMR). Experiments were done on 7–25 week-old mice of either sex. Female mice acclimated more readily to resting in the recording setup (see below), and so more females (N=21) than males (N=4) were used. Mice were of the following strains: C57BL6 (Charles River), F1 offspring of crosses of Pcp2-creTg/Tg and Ai27DTg/Tg (Jackson, RRID: IMSR_JAX:004146, and RRID: IMSR_JAX:012567), which generated mice heterozygous for channelrhodopsin-2 (hChR2 (H134R)) expression in Purkinje cells (Pcp2-ChR2), and VGAT-ChR2-EYFP mice (Jackson, RRID: IMSR_JAX:014548). Mice were housed with free access to food and water on a 12:12 day-night cycle.

Method details

Quantification and Statistical Analysis

Data were analyzed with pClamp (Molecular Devices), MatLab (MathWorks), and IGOR Pro (Wavemetrics). Complex spikes and simple spikes were sorted offline either manually with pClamp software or with PCA in MatLab. Instantaneous firing rate of simple spikes was calculated as the inverse of each interspike interval, and instantaneous rate vs. time plots were digitized at 1 kHz. Whisker position data was up-sampled from 250 Hz to 1 kHz to facilitate comparison with the instantaneous rate trace. For puff-evoked whisks 3–90 trials were averaged (mean, 36 trials) and for spontaneous whisks, 11–184 trials were averaged (mean, 48 trials). Complex spike histograms were computed by finding the spike probability in a sliding 20-ms bin advanced in 1 ms steps. Spontaneous whisks from rest were included in the analysis only if they were >10° in amplitude and if the mean position for the preceding 250 ms was less than −90° and a standard deviation less than 2.5°, to confirm a non-moving, resting position. Data are presented as mean ± SEM. Statistics were assessed based on an assumption of normality with paired Student’s t-tests, with a Bonferroni-Holm correction for multiple comparisons as needed. Significance was taken as p<0.05. For all comparisons, p-values and the associated test are reported in the text.