Adaptation of Whisker Movements Requires Cerebellar Potentiation

The ability to adapt while exploring the world is critical for survival, yet how it comes about is unclear. Here we show that a whisker protraction reflex can be elicited following rostro-caudal pad stimulation, and that short-term kinematics of this reflex are enhanced when coherent complex spike activity of cerebellar Purkinje cells occurs in crus 1. Instead, the long-term kinematics can be adapted by tetanic stimulation and this adaptation is largely controlled by changes in the simple spike activity of Purkinje cells in crus 2. Increases in whisker protraction correlate with preceding increases in simple spikes in trial-by-trial variation analysis, and both behavioral and spike adaptation are absent in independent mouse models in which postsynaptic long-term potentiation of Purkinje cells is blocked. These differentially distributed short-term and long-term cerebellum-dependent modulations of whisker movements may come into play during coordination and adaptation of natural behaviors like gap crossing and prey capture.


20
Active touch is important for exploring our environment, allowing us to assess the shape, 21 substance and movements of objects and organisms around us (Prescott et al., 2011). 22 Throughout the animal kingdom, various systems have evolved for this purpose; these include cellular and network mechanisms it may adjust such representations remains to be elucidated. 35 In the past, long-term depression (LTD) at the parallel fiber-to-Purkinje cell (PC) synapse, 36 guided by climbing fiber activity, has been considered to be the main cellular mechanism   Coherent complex spike firing is specifically enhanced by whisker pad stimulation 171 Next, we used two-photon Ca 2+ imaging to study the behavior of adjacent groups of PCs 172 around the moment of whisker pad air puff stimulation in awake mice. After injection of the 173 Ca 2+ -sensitive dye Cal-520 we could recognize the dendrites of PCs as parasagittal stripes 174 ( Figure 4A), each of which showed fluorescent transients at irregular intervals ( Figure 4B). 175 Previous studies identified these transients as the result of PC complex spike firing (Schultz et   Larger whisker protractions are preceded by complex spikes 192 In view of the strong convergence between PCs and cerebellar nuclei neurons, we expect that 193 complex spike coherence may contribute to driving motor activity (Mukamel et al., 2009;194 Hoogland et al., 2015). To study the impact of complex spike firing on touch-evoked whisker 195 protraction, we segregated the trials during which a complex spike was fired by a single PC 196 from those during which no complex spike was produced. It turned out that during the trials 197 with a complex spike, the protraction was significantly larger (see Figure 5A  that during the touch evoked whisker reflex simple spikes predominantly correlate with 228 whisker protraction and that this correlation is maximal without a clear time lead or lag, 229 unlike the complex spikes whose occurrence tended to precede the protraction. If an increase in simple spike firing correlates to stronger whisker protraction, then the simple 234 spike response to contralateral stimulation should be larger than that to ipsilateral stimulation, 235 as contralateral stimulation evokes a stronger protraction (Figures 2 and 7A). To test this we 236 recorded PC activity while stimulating the ipsi-and contralateral whiskers in a random 237 sequence ( Figure 7B). The change in maximal protraction was considerable (difference in 238 maximal protraction: 7.30 ± 1.24° (mean ± SEM); n = 9 mice; Figure 7C; cf. Figure 2E).

239
Possibly, the absence of direct impact of the air flow during contralateral stimulation can 240 explain part of this difference, which is also in line with the earlier onset of the protraction 241 during contralateral stimulation ( Figure 7C). However, in addition to this passive process, a 242 change in simple spikes may actively contribute to this difference as well, as the simple spikes 243 increased significantly more during contralateral stimulation (increase during first 60 ms after 244 air puff onset for contra-vs. ipsilateral stimulation: 13.7 ± 5.3%; mean ± SEM; p = 0.023; t = 245 2.413; df = 26; paired t test; n = 27 PCs; Figure 7E). Instead, the complex spike response was 246 reduced during contralateral stimulation (complex spike peak response: ipsilateral: 1.40%

14
This change in the amplitude of the touch-evoked whisker protraction was not 275 accompanied by any substantial change in the complex spike response to whisker pad 276 stimulation (p = 0.163; Wilcoxon matched pairs test; n = 55 PCs; Figure 8C; Table S2).

277
However, the rate of simple spike firing upon air puff stimulation was markedly increased 278 after 20 s of 4 Hz stimulation. This was especially clear during the first 60 ms after the air 279 puff (p = 0.003; Wilcoxon matched pairs test; n = 55 PCs; Figure 8D; Table S2). Overlaying pre-induction: 152.1 ± 18.1 ms; post-induction: 90.7 ± 9.4 ms (means ± SEM); p = 0.020; t = 286 2.664; df = 13; paired t test; n = 14 PCs; Figure 8F). Indeed, the maximum correlation was 287 found to be above the 45° line after the 4 Hz tetanic stimulation, implying that the changes in 288 simple spike firing now preceded the whisker movement, rather than being around the same 289 time as before the induction.   Figure 3H). We compared the location of this lateral crus 2 area to 316 that of the PCs with the strongest correlations between simple spike firing and whisker 317 protraction and we found these two crus 2 locations to match ( Figure 9B-C).

318
The impact of 4 Hz stimulation lasted as long as our recordings lasted; at least 30 min. 319 We separated between PCs with a weak complex spike response and those with a strong one.

320
In line with the correlation analysis shown in Figure 9A, the former PCs expressed a 321 potentiation of their sensory simple spike responses, whereas the latter ones did not (weak vs.     In line with the absence of increased touch-evoked whisker protraction, also the 348 increase in simple spike firing observed in WT mice was absent in L7-PP2B mice. As the 349 strong complex spike responders did not show plasticity (cf. Figure 9), we compared weak 350 complex spike responders of both genotypes. Simple spike responses were stably increased in 351 WT PCs with a weak complex spike response following tetanic stimulation (as shown in 352 Figure 9F), but not in those of L7-PP2B mice (effect of genotype: p = 0.003; F = 4.361; df = 353 4.137; Two-way repeated measures ANOVA with Greenhouse-Geyser correction; n = 8 WT 354 and n = 9 L7-PP2B PCs; Figure 10E).      Table   414 S3). In contrast, the L7-GluA3 KO PCs did not show any significant difference with the WT 415 PCs as far as these parameters were concerned, except for the peak complex spike response,  Table   417 S3). However, as the complex spike response negatively correlates with the occurrence of In this study we show for the first time that touch can evoke a reflexive whisker protraction 429 that is modifiable. As with learning of other reflexes, such as eyeblink conditioning or 430 adaptation of the vestibulo-ocular reflex, long-term modification of this touch-evoked whisker 431 protraction requires an intact cerebellum. We show here that a brief period of 4 Hz tetanic 432 sensory stimulation results in enduring amplification of a touch-evoked whisker protraction. 433 Extracellular recordings revealed that the simple spike rate of the PCs that are predominantly 434 located laterally in crus 2 correlates well with whisker movement and is congruently 435 increased with enhanced whisker protraction after tetanic stimulation. These PCs show a weak 436 complex spike response to whisker stimulation, which appears to act permissive for the 437 occurrence of parallel fiber-to-PC LTP. This plasticity mechanism is likely to be one of the 438 main mechanisms underlying this whisker reflex adaptation, as we found that the 4 Hz 439 induction protocol did result neither in more whisker protraction nor in stronger simple spike 440 responses in two independent genetic mouse models, both of which lack LTP induction at 441 their parallel fiber-to-PC synapse. By contrast, the PCs that show a strong complex spike 442 response to whisker stimulation and that are mainly located laterally in crus 1 did not manifest 443 a prominent regulatory role for their simple spike activity in long-term modification of their 444 whisker movements. While simple spike rates were readily modifiable and had a bilateral 445 relation with whisker movements, complex spike firing was more rigid and its coherence 446 correlated well with the strength of the short-term touch-evoked protraction reflex itself. The  Control of whisker movements 452 Although most mammals have whiskers, only a few species use their whiskers to actively 453 explore their environment by making fast, rhythmic whisker movements (Welker, 1964; 454 Woolsey et al., 1975;Ahl, 1986). In "whisking" animals, such as mice and rats, whisker      The authors declare no competing financial interests. In this study, we used two different mutant mouse lines, both on a C57Bl/6J background.   Table S4. All mice were socially housed until surgery and single-housed afterwards. The mice 616 were kept at a 12/12 h light/dark cycle and had not been used for any invasive procedure 617 (except genotyping shortly after birth) before the start of the experiment. All mice used were 618 specific-pathogen free (SPF). All experimental procedures were approved a priori by an      Data inclusion 708 We included all mice measured during this study, with the exception of one mouse where 709 video-analysis revealed that the air puff was delivered more to the nose than to the whisker

719
The impact of 4 Hz tetanic stimulation on air puff-triggered whisker movement was 720 quantified using a bootstrap method. First, we took the last 100 trials before induction and 721 divided these randomly in two series of 50. We calculated the differences in whisker position 722 between these two series, and repeated this 1000 times. From this distribution we derived the 723 expected variation after whisker pad air puff stimulation. We took the 99% confidence 724 interval as the threshold to which we compared the difference between 50 randomly chosen 725 trials after and 50 randomly chosen trials before induction ( Figure 10C). For some analyses, we discriminated between the sensory response period (0-60 ms 764 after stimulus onset) and inter-trial interval (500 to 200 ms before stimulus onset). We           Friedman's two-way ANOVA; see Table S1). Puffing from the contralateral whiskers or the    Figure 6D). Note that the strongest increase of simple spike 1333 responses after 4 Hz tetanic stimulation occurs in the region that also displayed the strongest Schematic drawing of the experimental layout. Air puffs lasting 30 ms were delivered from three different locations. In addition, some air puffs delivered ipsilaterally from the front were preceded by a brief air puff (2 ms) 100 ms before the actual air puff to test for pre-pulse inhibition (PPI). The four stimulus conditions were applied in a random order. B. For each of the 9 mice tested, we calculated the average whisker response (on the ipsilateral side) and represented these as summed line plots. The stacked line plots are scaled such that the brightest line (on top) depicts the average of all mice. The insets show the duration of the retraction (until the whiskers reached the baseline position again) comparing the 2 ms and the 30 ms pulses (left) and the maximal protraction amplitudes upon the pre-pulse compared to the pulse (right). The passive retraction upon the short pre-pulse was less intense, but the consecutive protractions were of similar amplitude, indicating the absence of pre-pulse inhibition (p = 0.0078 and p = 0.4961, respectively; Wilcoxon matched-pairs tests; significance level = 0.025 after Bonferroni correction for multiple comparisons). C. Overlay of averaged ipsilateral whisker responses with shaded areas indicating ± SEM. The three ipsilateral conditions resulted in similar amounts of protraction. Note that the puff from the back did not cause a retraction preceding the protraction and that the pre-pulse did not affect the size of the protraction following the second air puff. The brief pre-pulse induced a shorter retraction, but this had no effect on the protraction. Air puffs to the contralateral whisker pad caused stronger protractions than the ipsilateral stimuli. D. The maximum retraction was largest when the air puffer was in front of the ipsilateral whiskers. The shorter pre-pulse did cause a briefer retraction (see inset in B), but the amplitude was not significantly different from the retraction caused by the longer pulse (p = 0.268; Dunn's pair-wise post-hoc test after Friedman's two-way ANOVA; see Table S1). Puffing from the contralateral whiskers or the ipsilateral whiskers from the back caused the least retraction, indicating that the initial retraction is largely passive and caused by the air flow of the stimulator. E. The maximum protraction reached was similar for all conditions, except in case the contralateral whiskers were stimulated, which led to a stronger protraction on the ipsilateral side. n.s. p > 0.05; * p < 0.05; *** p < 0.001; *** p < 0.001. See also Source Data file.  Complex spike coherence is relatively rare during inter-trial intervals, but strongly enhanced following air puff stimulation. E. The same peri-stimulus histogram as in D, but with colors indicating the chance of occurrence of the level of coherence found based upon Poisson distribution of all complex spikes in this recording, emphasizing that coherence occurred more than expected, mainly during the sensory response. Indeed, during 1 Hz air puff stimulation, complex spikes were observed to be produced by large ensembles.

D E
In the absence of tactile stimulation, ensemble sizes tended to be smaller (F). The data presented in panels D-F come from the field of view shown in panel A. G. There was a shift from complex spikes fired by a single or a few Purkinje cells towards complex spikes fired by larger ensembles when introducing air puff stimulation. Presented are the median and the inter-quartile range of the differences between the two histograms as illustrated for an example experiment in panel F (n = 10). The increase in coherence directly after stimulation was highly significant (p = 0.001; Fr = 28.878; df = 9; Friedman's two-way ANOVA). * p < 0.05; ** p < 0.01, *** p < 0.001, **** p < 0.0001, ***** p < 0.00001.  Figure 2). C. Stacked line plots of the averaged whisker traces of 9 mice with the difference between the contralateral and ipsilateral stimulation depicted in the third column. D. Complex spike responses, on the other hand, were more prominent upon ipsilateral stimulation. E. The observation that increased simple spike firing correlates to enhanced whisker protraction (cf. Figure 6) was confirmed under these experimental conditions. * p < 0.05; *** p < 0.001. See also Data Source File.  Figure  6C-D) highlighting the anticipation of simple spike firing (F). The x-axis is based upon the instantaneous simple spike firing frequency and the y-axis upon whisker angle. After induction, the maximal correlation (R) between simple spikes and whisker angle (along dashed 45° line) shifts to an earlier time point after the air puff (3 rd column; shaded areas indicate SEM). In addition, a clear correlation is found with simple spike firing leading whisker movement (yellow/red area expands above the 45° line in 2 nd column as compared to the left column). Scale bar of the correlation matrices (left and middle) is at the right of the 3 rd column. ** p < 0.01; *** p < 0.001. See also Table S2 and Source Data File.  Figure  6D). Note that the strongest increase of simple spike responses after 4 Hz tetanic stimulation occurs in the region that also displayed the strongest correlation between instantaneous simple spike rate and whisker position. D. Example PSTHs of the simple spike response to whisker pad air puff stimulation of representative Purkinje cells and how they changed over time, depicted as heat maps of the instantaneous simple spike frequency (E; see scale bar in D). The left column displays the data from a representative Purkinje cell with a weak complex spike response, the right column of one with a strong complex spike response. The induction period is indicated with "4 Hz". F. The number of simple spikes following an air puff stimulation increased in weakly responding Purkinje cells and this increase remained elevated until the end of the recording (at least 30 min). In contrast, this increase was not found in Purkinje cells with strong complex spike responses. * p < 0.05; ** p < 0.01; *** p < 0.001.  2+ ] i ) that is largely determined by climbing fiber (CF) activity. Following CF activity, [Ca 2+ ] i raises rapidly and activates a phosphorylation cascade involving α-Ca 2+ /calmodulin-dependent protein kinase II (CaMKIIA) and several other proteins eventually leading to internalization of AMPA receptors and consequently to long-term depression (LTD). PF volleys in the absence of CF activity, on the other hand, result in a moderate increase in [Ca 2+ ] i , activating a protein cascade involving protein phosphatase 2B (PP2B) that promotes the insertion of new AMPA receptors into the postsynaptic density, thereby leading to long-term potentiation (LTP) of the PF-PC synapse. GluA3 subunits are part of the postsynaptic AMPA receptors. B. Example of a representative mouse with the averaged whisker movements before and after theta sensory stimulation, showing a stronger protraction afterwards, as evidenced by the differences between post-and pre-induction compared to a bootstrap analysis on the normal variation in whisker movements (C; shade: 99% confidence interval). Although also differences were observed in L7-PP2B mutants, these did generally not exceed the expected variations (right). D. In 14 out of 16 wild-type mice more protraction than expected was observed, against in only 3 out of 13 L7-PP2B mutant mice (pie charts). Stacked line plots of whisker movement differences between postand pre-induction for all mice. The plots are normalized so that the brightest line indicates the average per genotype. E. In contrast to simple spike responses in WT mice, those in L7-PP2B KO mice could not be potentiated by our 4 Hz tetanic stimulation protocol. This effect was stable, also during longer recordings. For this analysis, we selected those with weak complex spike responses, as the PCs with a weak complex spike response did not show increased simple spike firing after 4 Hz tetanic stimulation (see Figure 9A). * p < 0.05; ** p < 0.01; *** p < 0.001 Overall, 4 Hz tetanic stimulation did not result in stronger whisker protraction in L7-GluA3 mutant mice as observed in WT mice (see Figure 8). This is illustrated with a stacked line plot. C. Comparison of the average change in whisker angle over the 120 ms following the onset of the air puff shows enhanced protraction in WT (n = 16), but not in LTP-deficient miceneither in L7-PP2B (n = 13) nor in L7-GluA3 (n = 6) mutants, pointing towards a central role for parallel fiber-to-Purkinje cell LTP for the enhanced protraction in WT mice following a brief period of 4 Hz tetanic stimulation. The horizontal lines indicate the medians and the 1 st and 3 rd quartiles. The lack of change in whisker protraction following 4 Hz tetanic stimulation was reflected in the lack of change in simple spike responses as illustrated in three representative Purkinje cells (cf. Figure 9D-E). On top are the peri-stimulus time histograms (D) followed by heat maps illustrating the instantaneous firing rate over time (E). The induction period is indicated with "4 Hz". F. Overall, WT Purkinje cells (n = 35) showed increased simple spike firing after 4 Hz stimulation, while those in L7-PP2B (n = 21) or L7-GluA3 (n = 13) mutant mice did not. For this analysis, we restricted ourselves to the Purkinje cells with weak complex spike responses as the Purkinje cells with strong complex spike responses did not show potentiation in the WT mice (see Figure 9A) and to the first 100 trials after induction. As shown in Figure 11E, the increase in simple spiking firing developed over time (in WT, but not in L7-PP2B mice). Directly following the induction period, the increase in simple spike firing was not yet maximal. * p < 0.05; ** p < 0.01. See also Source Data File.