elife eLife eLife 2050-084X eLife Sciences Publications, Ltd 109183 10.7554/eLife.109183 10.7554/eLife.109183.2 reviewed preprint 1.3 Neuroscience Cerebellar climbing fibers impact experience-dependent plasticity in the mouse primary somatosensory cortex Silbaugh Abby 1 Koster Kevin P 1 Hansel Christian 1 chansel@bsd.uchicago.edu https://ror.org/024mw5h28Department of Neurobiology and Neuroscience Institute, University of Chicago, Chicago, United States Ito-Ishida Aya Reviewing Editor https://ror.org/04j1n1c04 RIKEN Center for Brain Science Saitama Japan Chen Lu Senior Editor https://ror.org/00f54p054Stanford University Stanford United States

Competing interests: No competing interests declared

 24 10 2025 04 12 2025 14 RP109183 20 09 2025 Preprint posted 22 09 2025 Reviewed preprint v1 24 10 2025 eLife Assessment Reviewer #1 (Public review): Reviewer #2 (Public review): Reviewer #3 (Public review): © 2025, Silbaugh et al 2025 Silbaugh et al https://creativecommons.org/licenses/by/4.0/ This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

In the cerebellum, climbing fibers (CFs) provide instructive signals for supervised learning at parallel fiber to Purkinje cell synapses. It has not been tested so far whether CF signaling may also influence plasticity in other brain areas. Here, we show that optogenetic CF activation suppresses potentiation of whisker responses in L2/3 pyramidal cells in primary somatosensory cortex (S1) of awake mice that is observed after repeated whisker stimulation. Using two-photon imaging and chemogenetics, we find that CFs control plasticity by modulating SST- and VIP-positive interneurons in S1 cortex. Transsynaptic labeling identifies zona incerta (ZI) to thalamic posterior medial nucleus projections as a pathway for cerebellar output reaching S1 cortex. Chemogenetic inhibition of PV-positive neurons in the ZI prevents CF co-activation effects, identifying the ZI as a critical relay. Our findings demonstrate that CFs impact sensory signal processing and plasticity in S1 cortex and thus may convey instructive signals.

 Keywords Calcium Instructive signal Parvalbumin-positive interneuron Posterior medial thalamic nucleus Purkinje cell Pyramidal cell Somatostatin-position interneuron Supervised learning Vasoactive intestinal polypeptide-positive interneuron Zona incerta http://dx.doi.org/10.13039/100000065 National Institute of Neurological Disorders and Stroke (NINDS) R21NS136954 Hansel Christian http://dx.doi.org/10.13039/100000065 National Institute of Neurological Disorders and Stroke (NINDS) NS094184 Koster Kevin P publishing-route prc Summary of Updates:

This is a revision to address comments of three reviewers at eLife.

 Introduction

Sensory experience and learning are associated with plasticity of multiple brain regions. Neocortical plasticity canonically depends on the activity of its direct input structures1. For example, plasticity in the barrel field of the primary somatosensory cortex (S1 cortex) of rodents is evoked upon repeated whisker stimulation. Receptive field maps may reorganize after whisker clipping2–4 or a history of rhythmic whisker stimulation5–7 that necessitates adaptive plasticity of synaptic weights, e.g. via long-term potentiation (LTP) and/or long-term depression (LTD) at synaptic inputs onto pyramidal neurons8, 9. The high degree of plasticity in S1 somatosensory circuits enables adaptation to altered inputs10. Though experience-dependent plasticity in neocortex is highly responsive to input features, modulating influences from other brain regions may play a role.

The cerebellum is increasingly assigned functions that were previously attributed to the neocortex11, 12. It projects via the thalamus to motor cortex and non-motor areas, including the prefrontal cortex13, 14, sensory, and associative areas15–17. A plausible possibility is therefore that cerebellar output impacts cortical signaling. It has indeed been shown that modulation of cerebellar outputs influences activity in the frontal cortex18, coherence of neuronal oscillations between neocortical areas19–21, and plasticity in parietal-motor cortex connections22. However, the cellular mechanisms of these interactions have not been described.

Here, we ask the specific question of whether cerebellar climbing fiber (CF) signals have the capacity to impact neocortical plasticity. Our motivation to study CF signaling is rooted in the importance of their activity for normal cerebellar function, where CF-controlled plasticity represents a classic example of supervised learning in the brain1. Depending on dendritic architecture, Purkinje cells (PCs) receive input from one or more climbing fibers (CFs)23 that originate in the contralateral inferior olive (IO). As originally predicted by Marr in his Theory of Cerebellar Cortex, CF signaling supervises plasticity at parallel fiber (PF) – PC synapses, providing contexts for learning24. It was later experimentally demonstrated that PF and CF co-stimulation initiates LTD at PF synapses25–28, and that the CF – via evoked calcium transients – tightly controls plasticity at these synapses29–34 and provides instructive signals for cerebellum-controlled learning35–38. CFs provide instructive error signals39, convey sensory input across modalities40–44, and may even signal the absence of expected sensory signals45. CFs may also carry reward or reward prediction-related signals46, 47. Contexts in which CFs are recruited continue to be identified, underscoring the importance of establishing whether they may influence neocortical activity and plasticity.

To determine whether CF signaling can impact plasticity of S1 neurons, we use the mouse brain to give us experimental access at the cellular level. Our observation that CF signaling regulates plasticity of L2/3 pyramidal cells in S1 cortex shows that activity in the olivo-cerebellar system has distinct consequences for input processing and plasticity in S1 cortex, and significantly expands what is known about the basic interaction between these regions. This mechanism could be used to provide signals to neocortex that are instructive in nature, depending on the context in which CFs are recruited, similar to those signals that play a role in cerebellar supervised learning.

Activity-dependent plasticity in S1 cortex is regulated by optogenetic CF activation

To study the effects of CF activity on S1 plasticity, we expressed GCaMP6f in neurons in S1 cortex and used two-photon microscopy in awake mice to measure responses to air puff stimulation of the whisker field in layer 2/3 (L2/3) neurons (Figure 1A, D). For optogenetic CF activation, channelrhodopsin-2 (ChR2) was expressed in neurons of the inferior olive (IO), which give rise to CFs terminating in the contralateral cerebellar cortex (Figure 1B, E; Figure S1A, B). An LED was used to deliver 470nm light pulses over an optical window centered on crus I/II, an area of the cerebellar cortex responsive to stimulation of the ipsilateral whisker field40. The efficacy of optogenetic activation was tested by expressing GCaMP6f in PCs of crus I/II and measuring calcium transients evoked by LED light pulses, which we varied in frequency and duration to achieve optogenetically evoked signals resembling spontaneous CF-induced calcium transients (Figure 1C; Figure S1F-I). Single light pulses lasting 50ms evoked calcium transients which were not significantly different in amplitude or peak latency from spontaneous CF-induced events recorded in the same PCs (Figure S1H, I)48. For plasticity induction in S1 cortex, we used a multi-whisker stimulation protocol (‘rhythmic whisker stimulation’; RWS) similar to those previously used to study sensory-driven synaptic plasticity in vivo7, 9, 49, 50. Here, RWS consisted of an application of 100ms air puffs (8psi) to the contralateral whisker field for 5min at 8Hz, a natural whisking frequency for mice that are sampling the environment51. RWS stimulation resulted in an increase in whisker-evoked calcium responses (ΔF/F over the range of 0-700ms post-stimulus onset) that we observed in morphologically identified L2/3 pyramidal neurons (PNs; Figure 1F) for the duration of the recording (60min post-RWS; Figure 1G-I; Figure S2D). These somatic calcium responses measured with GCaMP6 correlate with neuronal spike rates52. L2/3 pyramidal neurons respond to stimulation of the corresponding principal whisker but may also respond to stimulation of surround whiskers53. Therefore, the plasticity observed here in response to multi-whisker stimulation may reflect both a potentiation of synapses already conveying principal and surround whisker input as well as an expansion of the neuron’s receptive field to include additional surround whiskers. Given increases in the activity of excitatory neurons may be coupled with decreased activity of local inhibitory neurons in sensory-evoked plasticity54, we also assessed the responses of neurons that were not identified as pyramidal cells (i.e. putative interneurons; INs) in the same mice. Indeed, INs collectively responded to RWS stimulation with a significant depression in the amplitude of evoked calcium events (Figure 1J-L; Figure S2E). As parvalbumin (PV)-positive interneurons, somatostatin (SST)-positive interneurons, and vasoactive intestinal polypeptide (VIP)-positive interneurons make up ∼40%, ∼30%, and ∼12% of the neocortical interneuron population55, 56, respectively, this plasticity effect is likely comprised of a substantial portion of SST and PV interneurons. Using transgenic mice to tag SST and PV interneurons with tdTomato, we demonstrate RWS causes a significant depression in SST and PV interneurons (Figure S2F, G) – consistent with our observations in putative interneurons.

Optogenetic CF co-activation prevents adaptive L2/3 pyramidal cell potentiation in S1 cortex.

(A) Experimental schematic for (D – L). (B and C) ChR2 expression in the inferior olive (IO) and optogenetic CF activation evokes calcium transients in PCs. (B) Calbindin staining of PCs marks the cerebellum (shown here in the same plane as the IO; white arrowhead). Scale bar: 1mm. (C) Two-photon field of view of PC dendrites with calcium responses evoked by 470nm, 50ms LED pulses. Scale bar: 100µm. Right: Distribution of response amplitudes (box: 25th and 75th percentiles; line: median). (D) GCaMP6f expression in S1 cortex. Scale bar: 1mm. Trace scale bar: 0.2 ΔF/F; 0.5s. (E) Sagittal view of ChR2-expressing CF terminals in the contralateral cerebellar cortex that are activated in optogenetic experiments. Scale bar: 100µm. (F) Left: Example of the morphological identification of L2/3 pyramidal cells by apical dendrites (green arrowheads) and INs (white arrowheads) in a two-photon field of view. Scale bar: 100µm. Right: Pyramidal cell at higher magnification. Scale bar: 15µm. (G) Pyramidal cell plasticity after RWS (N=12 mice) and after pairing with CF activation (RWS+CF; N=11 mice). Right: Quantification of evoked responses across neurons. (H) Data in (G) binned across time. Blue shading: RWS/RWS+CF. (I) Quantification across mice, normalized to the pre-plasticity response. (J-L) Analyses for INs in the same mice. Box: 25th and 75th percentiles; line: median; whiskers: range. Trace scale bars: 0.1 ΔF/F; 0.5s.

To determine whether CF co-activation with whisker stimulation influences this form of experience-dependent plasticity, 50ms light pulses were delivered at 1Hz during the 5min period of repeated whisker stimulation (RWS+CF), which is the rate of spontaneous CF discharge39.

Each light pulse was delivered with a delay of 45ms with respect to the onset of whisker stimuli, mimicking the natural latency of CF responses to whisker stimulation40. RWS+CF blocked the potentiation of L2/3 pyramidal cell responses in S1 cortex and depression was observed instead (Figure 1G-I; Figure S2D). The depression of IN responses observed after RWS was absent upon optogenetic CF co-activation (Figure 1J-L, Figure S2E), and recordings in mice with tdTomato-tagged SST and PV interneurons further demonstrated the absence of inhibitory interneuron depression (Figure S2F, G). Importantly, both the S1 plasticity and optogenetic CF activation effects were observed when trials with active movement of the mouse were included (Figure 1; Figures S2) or when analysis was restricted to trials during which the mice were resting both before stimulus onset and throughout the entire response period (Figure S3). Note movements of the vibrissae were absent during rest.

To test the robustness of these effects, we reproduced these findings using an alternate methodology that is well suited to measure receptive fields, including barrels in S1 cortex. Intrinsic optical imaging57 was used to measure plasticity effects at low spatial resolution in anesthetized mice to facilitate measurement of single whisker responses. We slipped a glass pipette over an individual untrimmed whisker and moved it back and forth at 10Hz for a period of 5min. In the barrel field of contralateral S1 cortex, we confirmed the activation of the corresponding barrel and recorded responses to this passive sensory experience for 20min before and 30min after repeated stimulation. We observed an increase in the responsive area beyond the barrel (Figure S4). Plasticity was not observed in S1 cortex when ChR2-expressing CFs in crus I/II of the cerebellum were optogenetically activated at 1Hz for 5min during the period of repeated stimulation or when CFs alone were optogenetically activated at 1Hz for 5 min (Figure S4). Thus, intrinsic optical imaging demonstrated sensory-evoked receptive field plasticity of barrels can be regulated by cerebellar signaling.

Taken together, our findings demonstrate that whisker-induced response potentiation in S1 cortex results from L2/3 pyramidal cell plasticity. Local inhibitory interneurons show the opposite direction of plasticity in the same animals. The observation that this IN plasticity is of lower amplitude and does not perfectly mirror pyramidal cell plasticity suggests that the former supports the latter and may help to control it, but that pyramidal cell LTP may take place even in the absence of LTD in IN neurons. Note that ‘LTP’ and ‘LTD’ here reflect target neuron responsiveness – as measured by neural calcium signals – and are not specifically shown to rest on synaptic gain changes. CF activity blocks the whisker-induced response potentiation in PNs and depression in INs, thus regulating this form of whisker map plasticity.

 CF activation modulates inhibitory interneurons in S1 cortex

Local inhibitory interneurons in the neocortex are known to gate plasticity of pyramidal neurons, a mechanism studied extensively in the rodent barrel cortex56. Given our observation that CF activation exerts a gating effect on pyramidal neuron plasticity, it is plausible that neocortical inhibitory interneurons ultimately mediate the effects of CF signaling on S1 plasticity. To test this, we first examined whether CF activation causes acute modulation of the responses of inhibitory interneurons to whisker stimulation (Figure 2A). Paired whisker and CF stimulation indeed caused an acute increase in the whisker-evoked response of INs (Figure 2C). In the same mice, CF co-stimulation caused a concomitant decrease in the whisker-evoked response of PNs (Figure 2B). The acute response increase observed in the nonspecific population of INs here is likely driven by SST and PV neurons, which together make up ∼70% of the S1 inhibitory interneuron population. As such, we further characterized the responses of tdTomato-tagged SST and PV neurons (Figure 2E-F). Matching the results of the IN population, the acute responses of SST and PV neurons to whisker stimulation were enhanced upon CF co-activation (Figure 2H-I). VIP interneurons – which primarily inhibit other interneurons including SSTs and PVs56 – showed no consistent change in their acute responses to whisker stimulation upon CF co-activation in the default analysis window (0-700ms; Data S2). However, VIPs demonstrated significant suppression in a later analysis window of 650-850ms (corresponding to the early component of the post-peak response; Figure 2G). These findings suggest that CF co-activation causes a range of response changes in the population of L2/3 VIP neurons, but that an overall suppressive effect emerges in the late response phase. Taken together, these results demonstrate CFs have the capacity to regulate the activity of excitatory and inhibitory neurons in S1 by differentially modulating their acute responses to sensory input.

Optogenetic CF co-activation differentially modulates basic responses to whisker stimulation in different types of neocortical neurons.

(A) Schematic of the recording configuration (left) and S1 cortex microcircuit (right). (B) CF co-activation with whisker stimulation significantly alters evoked responses of morphologically identified L2/3 pyramidal neurons. Left: averaged traces. Middle: analysis by neuron (AUC from 0-700 ms). Right: analysis by mouse. Blue line indicates mean. (C) CF co-activation significantly enhances responses in putative interneurons (INs; n=2014 cells; N=53 mice). (D – F) Sample two-photon field of view (left) and post-hoc verification of expression in S1 using confocal microscopy (right). Arrowheads indicate neurons co-expressing GCaMP6f and tdTomato in VIP interneurons (D), SST interneurons (E), or PV interneurons (F). Scale bars: 100µm (left); 500µm (right). (G) CF co-activation with whisker stimulation significantly reduces VIP population responses (n=129 cells, N=7 mice), but enhances responses in SST interneurons (n=102 cells; N=12 mice; H) and PV interneurons (n=177 cells; N=18 mice; I). Note VIP responses are reduced in a late analysis window (650-850 ms), while SST and PV responses are enhanced in the default 0-700 ms analysis window. Middle plots show analysis by neuron, right plots analysis by mouse (blue line indicates mean). Scale bars: 0.1 ΔF/F; 0.5s. Box: 25th and 75th percentiles; line: median; whiskers: range. Trace scale bars: 0.1 ΔF/F; 0.5s.

 Opposing roles of SST and VIP interneurons in CF-mediated control of S1 plasticity

SST interneurons contact the dendrites of L2/3 pyramidal cells (Figure 3A, E), where inhibition can impact local synaptic plasticity processes. As we observed CF co-activation with whisker stimulation causes an acute increase in the activity of SST interneurons, we tested their necessity and sufficiency in mediating the effects of CF activity on S1 plasticity using chemogenetic approaches. We first expressed the excitatory hM3D(Gq) DREADD in neocortical SST interneurons (Figure 3B). Administration of the DREADD agonist deschloroclozapine (DCZ) by intraperitoneal injection caused an increase in SST interneuron responses (Figure 3C). Under these conditions, RWS did not potentiate L2/3 pyramidal cell responses and calcium signals were depressed instead (Figure 3D), demonstrating an increase in SST activity is sufficient to mimic the suppressive effect of CF signaling on PNs (Figure 1G-I). In contrast, inhibitory hM4D(Gi) DREADD expression in SST interneurons (Figure 3F) and acute administration of DCZ caused a significant reduction in SST interneuron responses (Figure 3G). This rescued potentiation of L2/3 pyramidal cell responses when RWS was paired with CF co-activation (Figure 3H), demonstrating SST activity is required to mediate the suppressive effect of CF signaling on PNs. These findings confirm that neocortical SST interneurons are directly positioned to control plasticity of L2/3 pyramidal neurons9 and suggest that this mechanism is recruited by CF co-activation.

SST interneurons are recruited to prevent L2/3 pyramidal cell plasticity.

(A) Circuit diagram highlighting the chemogenetic activation of SST interneurons and recording configuration. (B) Expression of activating hM3D(Gq) receptors is localized to neocortical SST interneurons. Scale bar: 100µm (top) and 500µm (bottom). (C) DCZ application enhances responses in SST interneurons (n=12 cells; N=7 mice). (D) In the presence of DCZ, RWS stimulation suppresses L2/3 pyramidal cell potentiation, and even causes depression, even when the CF is not co-activated (n=255 cells; N=4 mice). Left: average traces; middle: AUC analysis by cell; right: maximum response amplitude by cell. (E) Circuit diagram highlighting the chemogenetic inhibition of SST interneurons and recording configuration. (F) Expression of inhibiting hM4D(Gi) receptors is localized to neocortical SST interneurons. Scale bars as in (B). (G) DCZ application reduces responses in SST interneurons (n=53 cells; N=6 mice). (H) In the presence of DCZ, pyramidal cell potentiation is observed, despite CF-co-activation (n=183 cells; N=3 mice). Data analysis presented as in (D). Box: 25th and 75th percentiles; line: median; whiskers: range. Trace scale bars: 0.2 ΔF/F; 0.5s.

SST interneurons receive relatively weak input from cortical and subcortical regions providing input to S1 but are strongly inhibited by local VIP interneurons (Figure 4A, E)58, 59. Activation of VIP interneurons may suppress SST activity during RWS, thus enabling pyramidal neuron potentiation9, 60. We have shown above that optogenetic CF activation causes an overall reduction in whisker-evoked responses of VIP interneurons (Figure 2G), which may be a mechanism through which CFs block pyramidal neuron potentiation. To examine whether this VIP activity regulation is critical for RWS+CF-induced suppression of L2/3 pyramidal cell plasticity, we used chemogenetic approaches to manipulate the activity of VIP interneurons. We first observed that expression of inhibitory hM4D (Gi) DREADDs in VIP interneurons (Figure 4B) and acute DCZ administration significantly reduced activity in VIP interneurons (Figure 4C) and prevented S1 potentiation upon RWS stimulation alone (Figure 4D). This manipulation was similar to the effect of RWS+CF (Figure 1G-I) or SST activation during RWS (Figure 3D). Conversely, expression of excitatory hM3D(Gq) DREADDs in VIP interneurons (Figure 4F) and acute DCZ administration enhanced VIP interneuron responses (Figure 4G) and rescued S1 plasticity upon paired RWS and optogenetic CF activation (RWS+CF; Figure 4H), mimicking the effect of RWS (Figure 1G-I) or suppression of SST interneuron activity during RWS+CF (Figure 3H).

VIP interneurons mediate the effects of optogenetic CF activation.

(A) Circuit diagram highlighting the chemogenetic inhibition of VIP interneurons and recording configuration. (B) Expression of inhibiting hM4D(Gi) receptors is localized to neocortical VIP interneurons. Scale bars: 100µm (top) and 500µm (bottom). (C) DCZ application reduces responses in VIP interneurons (n=10 cells; N=3 mice). (D) In the presence of DCZ, RWS stimulation suppresses L2/3 pyramidal cell potentiation, and even causes depression, even when the CF is not co-activated (n=189 cells; N=3 mice). Left: averaged traces; middle: AUC analysis by cell; right: maximum response amplitude by cell. (E) Circuit diagram highlighting the chemogenetic activation of VIP interneurons and recording configuration. (F) Expression of activating hM3D(Gq) receptors is localized to neocortical VIP interneurons. Scale bars as in (B). (G) DCZ application enhances responses in VIP interneurons (n=14 cells; N=3 mice). (H) In the presence of DCZ, pyramidal cell potentiation is observed, despite CF co-activation (n=202 cells; N=4 mice). Data analysis presented as in (D). Box: 25th and 75th percentiles; line: median; whiskers: range. Trace scale bars: 0.1 ΔF/F; 0.5s.

To control for the continued presence of DCZ, identical experiments for each condition were performed without the administration of RWS (Figure S5A, D) or RWS+CF (Figure S5G, J, M). When comparing control and plasticity experiments, both the SST activation and VIP inactivation experiments successfully show SST activation is sufficient to block RWS-mediated PN potentiation (Figure S5C), and VIP activity is required for RWS-mediated PN potentiation (Figure S5F). The VIP activation experiments also successfully showed VIP activation rescues PN potentiation after RWS+CF (Figure S5O), and that this effect and was not due to sustained increases in responsivity of pyramidal neurons caused by the presence of DCZ in these conditions (Figure S5N). We additionally tested the potential contribution of PV interneurons to the CF-mediated control of S1 plasticity. Expression of inhibitory hM4D (Gi) DREADDs in PV interneurons indeed rescued PN potentiation after RWS+CF when compared to controls (Figure S5L), a robust effect given that PV interneurons are usually basket cells with strong recurrent connectivity to pyramidal neurons56.

To further support our interpretation that VIP interneurons control a disinhibitory network in S1 cortex, we analyzed the IN populations (which consist of putative SSTs and PVs) in the VIP chemogenetic manipulation experiments (Figure 5). VIP inactivation during RWS, which blocked PN potentiation (Figure 4D), indeed caused a corresponding increase in IN activity (Figure 5C) to mimic the effects of RWS+CF (Fig. 1G-L) even without CF activity. VIP activation during RWS+CF, which rescued PN potentiation (Figure 4H), caused a corresponding decrease in IN activity (Figure 5F) to mimic the effects of RWS (Figure 1G-L) even when CFs were active. These findings confirm that VIP interneurons orchestrate the inhibitory network in S1 cortex and show that this mechanism is recruited by optogenetic CF activation. Taken together, the findings presented thus far show that the critical circuit for RWS-driven plasticity is located in S1 cortex itself and identify VIP and SST interneurons as effectors that are sufficient to mediate the consequences of CF activation. Here, the notion of sufficiency does not exclude potential effects of plasticity processes elsewhere that might well modulate effector activation in this context and others not yet tested.

VIP interneurons orchestrate the inhibitory network in S1.

(A) Circuit diagram highlighting the chemogenetic inhibition of VIP interneurons and recording configuration. (B) Inhibiting hM4D(Gi) receptors are expressed in neocortical VIP interneurons. Note A-B are identical to Fig. 4.A-B. (C) In the presence of DCZ, RWS stimulation activates L2/3 interneurons, even when the CF is not co-activated (n=27 cells; N=3 mice). Left: averaged traces; middle: AUC analysis by cell; right: maximum response amplitude by cell. (D) Circuit diagram highlighting the chemogenetic activation of VIP interneurons and recording configuration. (E) Activating hM3D(Gq) receptors are expressed in neocortical VIP interneurons. Note D-E are identical to Fig. 4E-F. (F) In the presence of DCZ, IN depression is observed, despite CF co-activation (n=105 cells; N=4 mice). Data analysis presented as in (C). Box: 25th and 75th percentiles; line: median; whiskers: range. Trace scale bars: 0.2 ΔF/F; 0.5s.

 Zona incerta to the thalamic posterior medial nucleus (POm) as a multi-synaptic pathway from the cerebellar nuclei to S1 cortex

Cerebellar climbing fibers must exert their effects on the neocortex via the deep cerebellar nuclei (DCN), beyond which there may be several potential pathways to the primary somatosensory cortex. The barrel cortex in mice receives whisker sensory information from two thalamic sources, the ventral posterior medial thalamic nucleus (VPM), and the posterior medial thalamic nucleus (POm). Whereas the VPM projects densely to the middle layers of S1 cortex, POm inputs terminate largely in L5 as well as L1, where, in addition to pyramidal cells, they synapse on VIP interneurons58, 59. POm-to-VIP inputs have previously been shown to gate RWS-mediated plasticity of PNs in barrel cortex50. However, because the DCN send only weak direct projections to POm (and VPM), we focused on other output regions of the cerebellum that could influence the activity of sensory thalamic nuclei: the thalamic reticular nucleus (TRN) and zona incerta (ZI). The ZI represents a more plausible pathway through which the CF could exert its effects on S1 given its robust innervation by the cerebellar nuclei17, 61, 62 and projections to the POm63. To rule out a pathway to S1 through the thalamic reticular nucleus, whose GABAergic neurons primarily inhibit VPM64, we used retrograde fluorogold labeling from the somatosensory thalamus (which included VPM and POm) and anterograde labeling from the contralateral DCN. We found no overlap in the TRN (Figure S6). Next, we determined the primary output pathways of ZI cells that receive cerebellar input by using a dual-injection approach which leveraged the transsynaptic transport properties of AAV165: AAV1-Flp recombinase (pAAV-EF1a-Flpo) was delivered to the deep cerebellar nuclei, followed by a Flp-dependent label (Frt-ChR2-EYFP) to contralateral ZI (Figure 6A). Abundant EYFP-expressing neurons were found in the ZI (Figure 6B, C; Figure S7), with their axon terminals visible in both POm (Figure 6B, C; Figure S7) and S1 cortex (Figure 6B, C). These results suggest that the ZI-POm pathway, and not the TRN-VPM pathway, constitutes a critical thalamic route for cerebellar signals reaching S1 cortex (Figure S6C).

A pathway from the cerebellar nuclei via the zona incerta and thalamic nucleus POm to S1 cortex.

(A) Schematic of dual-injection strategy to label outputs of zona incerta (ZI) neurons receiving input from contralateral cerebellar nuclei. (B) EYFP expression in the left hemisphere (left; scale bar: 1mm) with higher magnification images taken in the same slice. Center: EYFP expression in the ZI. Right: labeled axons in the posterior medial nucleus (POm; bottom; scale bar: 20µm) and S1 cortex (top; scale bar: 10µm). (C) Quantification of cerebellar nuclei projections to the ZI. Curves show terminal density distribution from dorsal ZI (dZI) to ventral ZI (vZI). Color of curves correspond to images taken along the rostro-caudal axis (green: top (rostral) image; navy: center image; light blue: bottom (caudal) image. Scale bars: 500µm. (D) Labeling strategy for electrophysiological recordings shown in (E-F). Scale bar: 1mm. (E) Whole-cell patch-clamp recordings from POm-projecting ZI neurons receiving input from cerebellar nuclei. Top: responses to depolarizing current pulses in the absence (left) and presence (right) of TTX (1µM). Bottom: Responses in ZI cells to photostimulation of ChR2-expressing terminals from cerebellar nuclei neurons in the absence (let) and presence (right) of TTX. (F) Amplitude (left) and latency (right) of photostimulation-evoked EPSPs before (n=5) and after (n=3) wash-in of TTX. (G) Experimental configuration for recordings from mice expressing inhibiting hM4D(Gi) receptors in PV-expressing ZI interneurons. (H) In the presence of DCZ, the suppressive effect of CF co-activation on acute whisker responses of L2/3 pyramidal neurons is blocked (n=80 cells; N=3 mice). (I) Bar graphs: analysis by neuron. Box: 25th and 75th percentiles; line: median; whiskers: range. Trace scale bars: 0.1 ΔF/F; 0.5s.

To assess whether there is a monosynaptic connection from the cerebellar nuclei to neurons in the ZI that project to POm, we expressed ChR2 expression in the cerebellar nuclei and performed in vitro whole-cell patch-clamp recordings from POm-projecting ZI neurons (i.e. cells that were retrogradely labeled by fluoro-Ruby injected into POm) and tested them for ChR2-evoked responses (Figure 6D, Figure S8). Optogenetic activation of ChR2+ terminals at 10Hz elicited EPSPs that persisted in the presence of bath-applied TTX (1 µM; Figure 6E, F), despite significant amplitude reduction. This TTX-insensitive component represents the ChR2-evoked EPSP at the synapses between cerebellar nuclei neurons and ZI neurons and demonstrates the monosynaptic nature of this pathway66.

It has previously been demonstrated that most POm-projecting neurons in ZI are PV neurons67. To test whether these neurons convey cerebellar instructive signals to S1 cortex, we chemogenetically inhibited PV neurons in ZI during basic whisker and whisker plus optogenetic CF activation conditions and measured responses in L2/3 pyramidal neurons in S1 cortex. Expression of inhibitory hM4D (Gi) DREADDs in ZI-residing PV neurons (Figure 6G) and DCZ application prevented the suppressive effect of CF co-activation on pyramidal cell responses (Figure 6H, I). These data demonstrate that thalamus-projecting PV neurons in ZI convey CF signals to S1 cortex.

 Discussion

Multiple functions have been assigned to cerebellar CFs. As in our recordings, optogenetic CF co-activation suppresses S1 pyramidal neuron plasticity in an experimental paradigm lacking behavioral reward, this behaviorally neutral signal acts like an error signal would. However, it remains possible that CF recruitment and regulation of S1 plasticity may happen in a reward / reward omission context as well. CF activation might be driven by reward / reward omission in sensory-association training. This might be the reason why sensory stimulation in such training context drives cortical plasticity less well than passive sensory stimulation alone68. Plasticity in S1 cortex results from an update in sensory input – here exposure to a new sensory experience, prolonged rhythmic whisker activation at 8Hz – that necessitates circuit adaptation. CF signaling may get recruited when an obvious disruption or violation of a sensory prediction occurs, or simply when no stability in sensory input is reached. The absence of sameness may be detected by inferior olive neurons as an error in the sensory environment, and subsequently CFs prevent the otherwise unfolding potentiation. Such function provides an example for a “sensory role” of the olivo-cerebellum69,70, in this case by controlling adaptive plasticity in a primary sensory cortex, without immediate effects on the motor domain.

In rodents, direct trigemino-olivary connections convey whisker-related information to the olive71. Passive whisker activation evokes CF-mediated complex spike firing in PCs in ipsilateral crus I/II40. These evoked spikes occur on top of a spontaneous complex spike firing rate of ∼1Hz39. What distinguishes CF activity that suppresses plasticity in the cerebral cortex from these more common activity patterns? First, prolonged low-frequency air puff stimulation to sensitive skin areas on the snout or wrist of rodents evokes potentiation, and not depression, of PC responses72, 73. This finding is in line with the current understanding of CF signaling: novelty activates it39; when stimuli are repeatedly presented (producing/inhabiting a state of sameness), evoked complex spike firing weakens. Thus, RWS stimulation likely will not produce continuous elevated CF activation. For this to happen under natural conditions, novel input features have to emerge. Second, the synchronization of CF responses might be critical. Complex spike synchrony is absent or weak during spontaneous activity and might increase with movement initiation74, 75 or sensory stimulation76. Synchrony in a sufficiently high number of PCs is needed to evoke inhibition and subsequent rebound excitation in target neurons in the cerebellar nuclei77,78. Parallel fibers (PFs) likely cannot generate synchronous activity in enough PCs due to the spatial arrangement of PF input79. A threshold level of synchronization will be reached with optogenetic CF activation but may also naturally be reached with specific qualities of the sensory input signal, e.g. those that make it an error signal. It is also conceivable that synchronization to a threshold level is driven by an error signal that does not originate from whisker stimulation but has another sensory or non-sensory origin.

Whisker-related sensory information in rodents is conveyed by the trigeminal ganglion via thalamic nuclei VPM and POm to the barrel cortex; information from specific whiskers is projected via VPM to cortical layer 480. The observation that cerebellar signals impact S1 cortex via ZI and POm fits the description of a more modulatory pathway that does not interfere with thalamic input in layer 4, but with synaptic signal integration in the more superficial as well as deeper layers. Both two-photon imaging and chemogenetics identify SST interneurons in S1 cortex as the ultimate effectors of the activation of this modulatory pathway (note that a similar mechanism might underlie a phenomenon clinically known as CBI or ‘cerebellar-brain inhibition’, where electrical or magnetic stimulation over the cerebellum in human subjects causes a reduction in excitability of the motor cortex81, 82). SST interneurons, in turn, are regulated by VIP interneurons that receive activation from POm59. We identified PV neurons in ZI as neurons that are critically involved in this pathway as well, likely reducing the POm-mediated activation of VIP interneurons in S1 cortex. Our findings demonstrate that cerebellar output may recruit PV neurons in ZI sufficiently well to ultimately cause a removal of VIP interneuron-mediated inhibition of SST interneurons and consequently enhance dendritic inhibition of L2/3 pyramidal neurons. This activation sequence likely requires an initial activation of excitatory neurons in the cerebellar nuclei by output from the cerebellar cortex, which may result from rebound excitation following synchronous PC activity78. Another context where PV neurons in the ZI become activated – via glutamatergic input from S1 cortex – is self-grooming in the orofacial area83. It is plausible that under such conditions of predicted sensory input, whisker-related L2/3 pyramidal neuron plasticity is suppressed as well. While our findings identify the cerebellar nuclei to ZI to POm projection as critical to conveying CF signals to S1 cortex, it is likely that more projections that are not tested here can provide links from the cerebellum to S1 cortex. This is suggested by the distinct multi-channel output from the cerebellar nuclei84. For example, we cannot exclude a participation of direct projections from the cerebellar nuclei to VPM or POm85 (note, however, the prominence of projections that reach the ZI and originate in the cerebellar nuclei, as shown in Figure 6A-C; Figure S7). Moreover, it remains possible that cerebellar output reaches S1 cortex via M1 cortex, as both cortices show strong reciprocal connections86 and S1-projecting pyramidal neurons in M1 target VIP interneurons in S1 cortex87.

Our data show that activity of cerebellar CFs has an impact on S1 cortex, preventing plasticity of L2/3 pyramidal neurons via the recruitment of local SST interneurons. In our experiments, CFs were artificially (optogenetically) stimulated, and definitive conclusions about CF engagement and its consequences under natural conditions remain open. However, our study demonstrates both the capacity for such impact on cortical signaling and a pathway that enables this impact. These are requirements for CF signals to act as instructive signals in supervised learning88. A condition for such an effect likely is a synchronization of CF responses in groups of PCs77, 78, a scenario that provides a qualifying distinction from other CF activity patterns74–76. The observed external influence enables regulatory control that might safeguard the cerebral cortex from maladaptation, albeit at the cost of exposing it to pathological dysfunction of the olivo-cerebellar system, e.g. sensory over-responsivity that has been described in syndromic autism48.

 Materials and methods Animals

All animal experiments were approved and conducted in accordance with the regulations and guidelines of the Institutional Animal Care and Use Committee of the University of Chicago (IACUC 72496). Mice were housed on a 12-hour light/dark cycle and fed a standard rodent diet. Animals of either sex were used in all experiments and no sex-dependent differences were observed in any reported measures. Strains included wildtype mice (C57BL/6J; Jax: 000664), PVCre mice (B6.129P2-Pvalbtm1(cre)Arbr/J; Jax: 017320), SSTCre mice (STOCK Ssttm2.1(cre)Zjh/J; Jax: 013044), VIPCre mice (B6J.Cg-Viptm1(cre)Zjh/AreckJ; Jax: 031628), and Ai9 mice (B6.Cg-Gt(ROSA)26Sortm9(CAG-tdTomato)Hze/J; Jax: 007909). Homozygous SSTCre, PVCre, and VIPCre animals were bred with wild type animals to generate heterozygotes. To generate SST and VIP reporter lines (SSTCretdTom and VIPCretdTom), homozygous SSTCre and VIPCre mice were bred with homozygous Ai9 mice. Only heterozygous animals were used for experiments given evidence that endogenous somatostatin (SST) and vasoactive intestinal polypeptide (VIP) expression may be reduced in homozygous SSTCre and VIPCre animals, respectively89, 90.

 Stereotaxic injections

A list of all viruses with titers is provided in Data S10.

For electrophysiological experiments

Stereotaxic virus injections for electrophysiological experiments were performed as previously described using C57BL/6J mice between postnatal day 23-2891. For optogenetics experiments in which cerebellar inputs to zona incerta were measured, a 250 nL (40 nL/min) infusion of AAV5-hSyn-hChR2(H134R)-EYFP (UNC Vector Core) was made into the right deep cerebellar nuclei (-1.5 ML; -5.8 AP; -3.2 DV92) using a 1 mL Hamilton syringe (catalog: 65458-02) with a flat needle tip. Next, a small infusion (40 nL total, at 7nL/min) of fluoro-Ruby (ThermoFisher, catalog: D1817) was made into left posterior medial thalamus using a 0.5 mL Hamilton syringe (catalog: 65457-01). Following a two-week incubation, animals were sacrificed for recordings in acute slices (see Slice electrophysiology).

 For tracing experiments

For anterograde transsynaptic labeling experiments, a dual-injection approach was utilized to specifically label the outputs of neurons in the zona incerta that receive input from the contralateral cerebellar nuclei. Flpo-recombinase (AAV1-EF1a-Flpo, item#: 55637, Addgene) was injected into left deep cerebellar nuclei as above (+1.5 ML; -5.8 AP; -3.2 DV), followed by an infusion of AAV1-Flp-on ChR2-EYFP (AAV-hSyn-Coff/Fon-hChR2(H134R)-EYFP-WPRE, UNC Vector core) into right zona incerta (-1.8 ML; -2.0 AP; -4.0 DV). Following a 5-week incubation – the period required for anterograde transport of the AAV1 virus injected into zona incerta – animals were transcardially perfused for histology (see Tissue preparation for fluorescence microscopy and immunohistochemistry).

To visualize potential overlap between cerebellar terminals and cells in thalamic nuclei (i.e. the thalamic reticular nucleus) that project to somatosensory thalamus (ventral posteromedial nucleus and posterior medial thalamus), an orthograde virus (AAV5-hSyn-hChR2(H134R)EYFP (UNC Vector core)) was injected into the right deep cerebellar nuclei (-1.5 ML; -5.8 AP; 3.2 DV) and fluorogold (2% w/v in sterile saline, Fluorochrome) was injected into the left somatosensory thalamus (primarily posterior medial thalamus). Following a 2-week incubation, animals were transcardially perfused for histology (see Tissue preparation for fluorescence microscopy and immunohistochemistry).

 For two-photon calcium imaging experiments

Viral injections were performed to express channelrhodopsin-2 in cerebellar climbing fibers; fluorescent calcium indicators in cerebellar Purkinje cells or primary somatosensory (S1; barrel) cortex neurons; tdTomato, activating DREADDs, or inactivating DREADDs in S1 cortex; and inactivating DREADDs in interneurons in zona incerta. With respect to the inferior olive injection site (conferring viral expression in climbing fibers terminating in the contralateral cerebellar hemisphere), injections into cerebellar cortex were performed contralaterally, and injections into zona incerta and somatosensory cortex were performed ipsilaterally.

To express channelrhodopsin-2 in cerebellar climbing fibers, ChR2-EYFP (AAV9-CaMKIIahChR2(H134R)-EYFP; Addgene #26969-AAV9) or ChR2-mCherry (pAAV9-CaMKIIahChR2(H134R)-mCherry; Addgene #26975-AAV9) was injected into the left inferior olive as follows: after surgical preparation (see Two-photon calcium imaging: Cranial window surgeries), tissues attached to the occipital bone were surgically detached, followed by opening of the posterior atlanto-occipital membrane to expose the dorsal medulla. A 5μL Hamilton Neuros syringe (catalog: 65460-03) with a 33-gauge beveled needle tip (catalog: 65461-02) was inserted into the left medulla at an angle of 58-59° to the perpendicular line (Figure S1A), 0.75 mm laterally and 1.70 mm deep from the surface of the medulla as previously described93. A total volume of 400 nL was infused at a rate of 80 nL/min, and the syringe was held in place for five minutes before removal. ChR2-EYFP or ChR2-mCherry was chosen based on the excitation and emission spectra of other fluorophores expressed in the same animal to maximize signal separation when conducting post-hoc confocal imaging.

Stereotaxic virus injections for imaging of Purkinje cell dendrites were administered as previously described using P80-120 adult C57BL/6J mice73. A virus mixture was prepared with 0.5% L7-Cre (AAV1.sL7.Cre.HA.WPRE.hGH.pA, Princeton Neuroscience Institute Viral Core Facility, kindly provided by the lab of Dr. Samuel Wang at Princeton University), and 20% Cre-dependent GCaMP6f (AAV1.CAG.Flex.GCaMP6f.WPRE.SV40; Addgene, #100835) in saline solution. A glass pipette with a tip diameter of ∼300 μm was prepared by a puller (model P-97, Sutter Instrument Co.) and used to inject a total volume of 1800 nL into two separate injection sites: 900 nL were injected into right medial Crus I (-1.8 ML; -7 AP; -1.8 DV), followed by a 900 nL injection into right medial Crus II (-2.5 ML; -7 AP; -2.2 DV). The pipette carrying the AAV virus was held in place for five minutes before withdrawal.

For DREADD suppression of PV interneurons in zona incerta, a 1 mL Hamilton syringe was used to infuse Cre-dependent hm4D(Gi) (AAV5-hSyn-DIO-hM4D(Gi)-mCherry; Addgene #44362-AAV5) into the left zona incerta (1.8 ML; -2.0 AP; -4.0 DV) of PVCre animals.

To express GCaMP6f in primary somatosensory (barrel) cortex neurons in C57BL/6J, SSTCretdTom, and VIPCretdTom animals, a 5μL Hamilton Neuros syringe with a 33-gauge beveled needle tip was used to infuse a total volume of 800 nL of GCaMP6f (AAV1.Syn.GCaMP6f.WPRE.SV40; Addgene #100837-AAV) at two separate injection sites: 400 nL were injected at -1.2 AP; 2.5 ML and 400 nL at -1.5 AP; 3 ML. At each injection site, 200 nL were first injected 300 μm below the pial surface, followed by 200 nL at 150 μm below the pial surface. The injections were performed at a rate of 100nL/minute, and the syringe was held in place for five minutes before removal. To express GCaMP6f in barrel cortex neurons with co-expression of tdTomato in parvalbumin (PV)-, SST-, and VIP-expressing interneurons in PVCre, SSTCre, or VIPCre animals, respectively, GCaMP6f (Addgene #100837-AAV1) and Cre-dependent tdTomato (AAV9-FLEX-tdTomato; Addgene #28306-AAV9) were co-injected using the same technique. To optimize co-expression of the viruses, GCaMP6f with tdTomato was mixed at a 3:1 volume ratio. For DREADD activation or suppression of VIP, SST, or PV interneurons, Cre-dependent hM3D(Gq) (AAV5-hSyn-DIO-hM3D(Gq)-mCherry; Addgene #44361-AAV5) or Cre-dependent hM4D(Gi) (AAV5-hSyn-DIO-hM4D(Gi)-mCherry; Addgene #44362-AAV5) were used in place of tdTomato, respectively.

For all two-photon imaging experiments, recordings began after a 4-week incubation period after the final virus injection (see Two-photon calcium imaging: cranial window surgeries for detail).

 Tissue preparation for fluorescence microscopy and immunohistochemistry

After completion of two-photon experiments, selected animals (all those used in DREADD and cerebellar imaging experiments, and a subset of animals in all other experimental cohorts) were sacrificed for post-hoc characterization of viral expression. Animals were transcardially perfused with ice-cold 4% paraformaldehyde in phosphate-buffered saline (25mL), pH 7.4. The brain was extracted and postfixed in 4% paraformaldehyde overnight at 4°C before being transferred to a cold 30% sucrose (in phosphate-buffered saline) solution for >48 hours. Brains were then cryosectioned at 50 μm thickness, with an approximately equal number of brains sectioned coronally or sagittally per experimental cohort. For tracing experiments, animals were perfused with ice-cold phosphate-buffered saline (25mL) prior to perfusion with paraformaldehyde and cryosectioned coronally at 150 μm thickness. Slices were stored in phosphate-buffered saline prior to immunohistochemistry or confocal imaging.

For post-hoc visualization of Purkinje cells in mice expressing ChR2-EYFP in climbing fibers (i.e. in animals without GCaMP6f expression in Purkinje cells), calbindin immunohistochemical staining was performed. Tissue slices were washed in a 0.01M phosphate-buffered saline solution containing 200 mM glycine for two hours at 4°C, then incubated in a 0.01M phosphate-buffered saline solution containing 10 mM sodium citrate and 0.1% Tween-20 at ∼60°C for 30 minutes using a heated water bath. After cooling to room temperature, the slices were washed three times in a 0.01M phosphate-buffered saline solution containing 0.5% Tween-20 (PBS-Tween) at room temperature for a period of 15 minutes per wash. Tissue was permeabilized with 200 mM glycine in PBS-Tween for 15 minutes at room temperature. Blocking was done for two hours at room temperature with a PBS-Tween solution containing 5% bovine serum albumin and 10% normal donkey serum, followed by overnight incubation (12-15 hours) at 4°C in a PBS-Tween solution containing 1% normal donkey serum and guinea pig anti-calbindin primary antibody (1:500; Synaptic Systems Cat# 214 004, RRID:AB_10550535). After washing the tissue three times for fifteen minutes in PBS-Tween at 4°C, slices were incubated for two hours at 4°C in PBS-Tween containing 1% normal donkey serum and donkey anti-guinea pig Cy3 (555) secondary antibody (1:200; Jackson ImmunoResearch Labs Cat# 706-165-148, RRID:AB_2340460). Finally, the slices were washed three times in PBS-Tween (ten minutes per wash) before mounting with Vectashield (Vector Laboratories, Inc.). The mounted slices were allowed to set overnight before imaging.

To confirm the channelrhodopsin-2 expression observed in cerebellar cortex was indeed expressed in climbing fiber terminals, a subset of tissue was additionally labeled with VGluT2. VGluT2 (rabbit anti-VGluT2, 1:500; ThermoFisher Scientific Cat# 42-7800, RRID:AB_2533537) was added to the primary antibody solution and donkey anti-rabbit AF647 (Jackson ImmunoResearch Labs Cat# 711-605-152, RRID:AB_2492288) was added to the secondary antibody solution.

 Confocal imaging Image acquisition

Z-stack confocal images were taken at 5x (Olympus MPlan N 0.1NA, air), 10x (Zeiss Achroplan 0.25NA, air), 20x (Olympus UMPlanFL N 0.5NA, water), 40x (Zeiss EC Plan-Neofluar 1.3NA, oil immersion), and 63x (Zeiss W Plan-Apochromat, water) magnifications with a Zeiss LSM 900 Axio Examiner.Z1 scope. A subset of images (Figure 1E and Figure S1B) were taken with a Zeiss LSM 5 Exciter, Axioskop 2 (using the 20x Olympus UMPlanFL and 40x Zeiss EC Plan-Neofluar objectives, respectively). The images in Figure 6D (shown with channel separation in Figure S8B) and Figure S6 were taken with a second Zeiss LSM 900 confocal microscope. To facilitate direct comparison, images shown in series (as in Figure 6C and Figure S7D, E) were taken with identical hardware and software configurations, available in the image metadata.

 Post-hoc image processing

To stitch multiple images (as in Figures 1B, 1D, 2D-F (right), 4B, 4F, 6B, 6C, S1A (top), S1E, S6A, S6B (top and bottom left), and S7C-E), a series of pairwise stiches were performed using Fiji (NIH). Figure 1B and Figure S1A (top) each consisted of 4 images (each 5x magnification, 1024 x 1024 pixels). Figure 1D consisted of 12 images (each 5x magnification, 512 x 512 pixels). Figure 2D (right) consisted of 10 images (each 10x magnification, 1024 x 1024 pixels). Figures 2E and 2F (right) each consisted of 8 images (each 10x magnification, 512 x 512 pixels). Figures 4B and 4F each consisted of 2 images (each 20x magnification, 512 x 512 pixels). Figure 6B (also shown as the first image in Figure S7D), and all images in S7D and S7E each consisted of 6 images (each 5x magnification, 512 x 512 pixels). Images in Figure 6C each consisted of 9 images (each 10x magnification, 512 x 512 pixels). Figure S1E consisted of 8 images (each 5x magnification, 1024 x 1024 pixels). S7C consisted of 10 images (each 40x magnification, 512 x 512 pixels). Note a subset of the final stitched images displayed in figures are cropped to show relevant detail.

Channel colors set during image acquisition followed conventional emission wavelength assignments (i.e. red to indicate tdTomato, green to indicate GCaMP6f, etc.). To illustrate interneuron subtypes and continuity across experiments (e.g., cyan to indicate ChR2 expression across all optogenetic stimulation experiments), channel colors were specified post-imaging using Fiji by separating channels and selecting the appropriate color when merging, ignoring source LUTs. When necessary, post-hoc brightness and contrast adjustments were applied linearly to an entire image – after stitching, if applicable – using Fiji. To facilitate direct comparison, identical linear brightness/contrast adjustments were made across all images shown in series (as in Figure 6C, Figure S7D, and Figure S7E).

To measure terminal density across the sections shown in Figure 6c, Fiji was used to rotate each image such that the axis separating ventral from dorsal zona incerta was horizontal. A 400 µm x 1 mm region of interest (ROI) was drawn around the zona incerta, and each row of pixels within this ROI was separated to facilitate calculation of the integrated density across each row of pixels. A similar method was used to quantify the terminal density of cerebellar inputs to zona incerta as in Figure S8. Briefly, images of sections matched for their rostro-caudal location were rotated as before. Then, five line-ROIs oriented vertically, each being 200 µm wide, were positioned side-by-side across the mediolateral axis of the zona incerta, and the spatial pattern of fluorescence along these lines was measured by calculating the mean grey value along their length (which was roughly 400 µm from the dorsal boundary to the ventral boundary). These were averaged for each animal used. Shading represents SEM for both measurements.

 Slice electrophysiology Acute slice preparation and whole cell recordings

For recordings in acute slice, injected animals (see Stereotaxic injections: For electrophysiological experiments) were deeply anesthetized using the isoflurane drop method in a bell jar (with a raised platform for the animal) and immediately transcardially perfused with 8-10 mL of ice cold oxygenated (95% O2, 5%CO2) artificial cerebrospinal fluid, which contained the following (in mM): 125 NaCl, 25 NaHCO3, 3 KCl, 1.25 NaH2PO4, 1 MgCl2, 2 CaCl2, and 25 glucose. The brain was then extracted, glue-mounted on a vibratome (Leica) platform and blocked for coronal slices using an agarose (5%) cube and sliced in the same solution (ice-cold). Slices were cut to 365 µm thickness. The brain slices were then transferred to 32-34 °C oxygenated artificial cerebrospinal fluid that was allowed to return to room temperature over the course of one hour, which constituted the slice recovery period.

Slices containing the zona incerta and terminals from the cerebellar nuclei inputs were visualized using differential interference contrast with an Axioskop 2FS microscope (Carl Zeiss). Fluorescence from ChR2-EYFP expression and from retrograde fluoro-Ruby labeling was confirmed using the 5x air objective and guided recording locations in zona incerta. Recordings were made with a Multiclamp 700B amplifier and pCLAMP software (Molecular Devices). Recording glass pipettes with 7–9 MΩ resistance were filled with an internal solution as follows (in mM): 117 K-gluconate, 13 KCl, 1 MgCl2, 0.07 CaCl2, 10 Hepes, 0.1 EGTA, 2 Na2-ATP, 0.4 Na-GTP, pH 7.3, 290 mOsm. Incertal cells whose morphology was clearly revealed by strong red fluorescence, indicating retrograde labeling from posterior medial thalamus, were targeted for recordings. Bath application of TTX (1 µM) was performed in a subset of recorded cells to verify the monosynaptic nature of the cerebellar input to zona incerta.

Optogenetic stimulation was delivered using a 355 nm laser (DPSS: 3505–100), controlled with galvanometer mirrors (Cambridge Technology) focused on the slice through a 5x air objective using custom software in MATLAB (MathWorks). Focal photostimulation of the ChR2expressing synaptic terminals in zona incerta was performed at 10Hz (4 pulses of 1-ms duration at 100 ms interstimulus interval).

 Electrophysiological data analysis

Electrophysiological data were collected using custom MATLAB software and analyzed using Graphpad Prism (v7.0). The amplitude of excitatory responses (i.e., the first response in the stimulus train) to stimulation pulses was measured by subtracting the average value for 20 ms before the delivery of a pulse (baseline) from the maximum value of the peak in current clamp at resting membrane potential. Latency of the optogenetic response was calculated by measuring the delay in milliseconds from the onset of the first optogenetic stimulus to the moment the resulting response reached >10% of its maximum amplitude.

 Two-photon calcium imaging Cranial window surgeries for imaging of Purkinje cell dendrites

To minimize overall recovery burden, surgeries were performed in two stages: (1) inferior olive virus injection and headframe installation, and (2) cerebellar virus injection and cerebellar cranial window installation as previously described73. In the first stage, mice (P60-90) were deeply anesthetized using 1-2% isoflurane and clamped by ear bars at the external acoustic foramen. The line between the clamping point and maxilla was set parallel to the horizontal plane. After trimming the fur on top of the skull with clippers, the surgical site was prepared by applying betadine and 70% ethanol three times in an alternating fashion. Meloxicam (2 mg/kg), extended-release buprenorphine (0.1 mg/kg), and 0.5 mL saline were administered subcutaneously, and depth of anesthesia was confirmed via tactile stimulation of the toe before making a 20-25 mm incision in the skin to reveal the occipital bone and attached tissues. After opening of the posterior atlantooccipital membrane and channelrhodopsin-2 virus injection into the left inferior olive (see Stereotaxic injections: For two-photon calcium imaging experiments), tissues were reattached to the occipital bone using instant adhesive, ensuring adequate space for later implantation of the cranial window over the right cerebellar cortex. A custom titanium headframe (H. E. Parmer Company, Nashville, TN, USA) was secured with dental cement (Stoelting Co., Wood Dale, IL, USA), completing the first stage.

After monitoring recovery for a period of 4 days, mice underwent the second surgery. A circular craniotomy over right cerebellar cortex with a diameter of 4mm, centered at -2.7ML; -6.9 AP, was performed using a dental drill. The dura was carefully removed to expose lobules Crus I and anterior Crus II before administering GCaMP6f injections (see Stereotaxic injections: For two-photon calcium imaging experiments). Following injection, a two-layer glass window was installed using C&B Metabond dental cement (Patterson Dental Company, Saint Paul, MN, USA), completing the second stage. The glass windows consisted of a 4 mm glass window (Tower Optical Corp, Boynton Beach, FL, # 4540-0495) adhered to a 5 mm glass window (Warner Instruments, Holliston, MA, USA, # CS-5R) using ultraviolet light-activated glue (Norland Optical Adhesive 71; Norland Products Inc., Jamesburg, NJ). Experiments were conducted 4.5 weeks after inferior olive injection and 4 weeks after GCaMP6f injection.

 Cranial window surgeries for imaging of barrel cortex neurons

Surgeries were again separated into two stages: (1) inferior olive virus injection, headframe installation, and cerebellar cranial window installation; and (2) S1 cranial window installation and virus injection. First, channelrhodopsin-2 was injected into the left inferior olive as above (see Cranial window surgeries for imaging of Purkinje cell dendrites and Stereotaxic injections: For two-photon calcium imaging). After reattaching tissues to the occipital bone using instant adhesive, a dental drill was used to perform a circular craniotomy over right cerebellar cortex (lobules Crus I and II), followed by removal of the dura and installation of the glass window as above. Finally, the custom titanium headframe was secured using dental cement, completing the first stage.

After monitoring recovery for a period of 4 days, mice underwent the second surgery. After surgically detaching a small portion of the left temporalis muscle attached to the parietal bone, a second circular craniotomy over left primary somatosensory cortex with a diameter of 4 mm (centered at -1.5 AP; 3 ML) was performed. The dura was removed to expose the neocortical surface before administering the appropriate viral injections (see Stereotaxic injections: For two-photon calcium imaging experiments). Following injection, a glass window was installed over the craniotomy using dental cement. To ensure optical isolation during recordings, a custom sheath was used to couple a Thorlabs patch cable (used for LED stimulation of cerebellar climbing fibers; see Optogenetic and tactile stimulation) to the cerebellar window. This sheath was adhered over the cerebellar window using Metabond dental cement, completing the second stage. For experiments involving DREADD manipulation of neurons in zona incerta, viral injections into left zona incerta were administered three days prior to the first surgery. Thus, experiments were conducted approximately 5 weeks after zona incerta injection, 4.5 weeks after inferior olive injection, and 4 weeks after primary somatosensory cortex injections.

 Habituation

After completion of surgeries, postoperative care was provided for 5-7 days. Once animals exhibited exploratory behavior without signs of pain or distress, habituation began with handling of mice for 15 minutes and transportation from the animal facility to the laboratory. After two days, anxiety during handling typically diminished, and the handling period increased to 30 minutes in the vicinity of the recording area. Following two days of handling, the mice were introduced to a free-running treadmill and head restraint via clamping of the head frame on left and right sides. The duration of time spent on the treadmill increased from 10 minutes to two hours per day depending on the mouse’s comfort. The relative time spent running on the treadmill typically decreased after two days. To habituate mice to air puff stimulation during this time, receptive field mapping of the field of view was conducted by delivering 8psi air puffs through a capillary tube while observing epifluorescence signals and calcium responses in real time. Air puffs were delivered at irregular frequencies, with at least 30 seconds between stimuli, to avoid influencing subsequent plasticity experiments. In total, mice were habituated for a period of 1-2 weeks until they showed no aversive responses to air puff stimulation, spent the majority of time in the recording environment at rest, and reduced movement initiation (whisking/running) with air puff stimulation. Animals used for DREADD manipulation experiments were familiarized with the DCZ injection procedure by receiving DMSO/saline injections during habituation for two days prior to recording.

 Optogenetic and tactile stimulation

To optogenetically stimulate cerebellar climbing fibers, the tip of a Ø200 µm core, 0.39 NA rotary joint patch cable (Thorlabs catalog: RJPSL2) was inserted into the sheath attached to the cranial window over the right cerebellar cortex (see Two-photon calcium imaging: Cranial window surgeries). For experiments in which Purkinje cell responses to optogenetic climbing fiber stimulation were also measured, the patch cable was inserted into a custom light shield that fit over both the objective and the cable. The tip of the patch cable was centered over Crus II in all optogenetic stimulation experiments (-2.7ML; -6.9 AP). Blue light pulses were delivered with a 470 nm Fiber-Coupled LED (Thorlabs catalog: M470F3). The absolute power output was adjusted to 2.5 mW. Given the tip of the LED was positioned 1mm over the surface of the brain (0.7 mm over the surface of the optical window), the estimated spot size was 0.6 mm and the estimated power density was 8 mW/mm2 at the pial surface. To deliver whisker stimuli, a 0.86 mm diameter glass capillary tube was positioned over the right C-row whiskers 5 mm from the right whisker pad at an angle of 45° in the X and Y planes (adjusted slightly during receptive field mapping for each animal). Air puffs were delivered at 8 psi (Picospritzer III, Parker Hannifin). For all imaging sessions, calcium activity was measured in trials lasting 20 seconds, with a stimulus delivered once per trial. Trials began at random intervals, with an interval of at least 5 seconds between trials. When testing Purkinje cell responses to optogenetic climbing fiber stimulation, 470 nm pulses were delivered with the following durations: 1 x 20ms, 1 x 50ms, 3 x 15 ms at 8 Hz, and (3 x 15 ms at 8 Hz)*2 with a 250 ms interval between epochs. The 1 x 50 ms light pulse was selected for all subsequent experiments given this stimulation duration evoked responses resembling spontaneous events in the same Purkinje cells (Figure S1). To test the basic transmission of the climbing fiber signal to S1 (i.e., the immediate influence of optogenetic stimulation of climbing fibers on acute responses to whisker stimulation shown in Figure 2), 100 ms air puff test pulses were delivered with or without optogenetic co-stimulation of climbing fibers. For paired test pulse stimuli, the 50 ms blue light stimulus was delivered with a delay of 45 ms with respect to the onset of the whisker stimulus. This delay was chosen to mimic the natural latency of climbing fiber responses to whisker stimulation, which peaks approximately 50 ms after stimulus onset40. For plasticity experiments, whisker test pulses were delivered during a baseline period of at least 40 minutes, followed by a 5-minute period of rhythmic whisker stimulation (RWS) at 8 Hz – the speed at which mice naturally sample objects51. For plasticity experiments with optogenetic co-activation of climbing fibers (RWS+CF), the same protocol was followed with the addition of 50 ms blue light pulses at 1 Hz during plasticity induction. The 8 Hz whisker stimulation and 1 Hz CF stimulation occurred at a consistent time relative to one another, with a blue light pulse always occurring at a 45 ms delay relative to the onset of a whisker stimulus (the same delay used for the paired test pulses). After RWS or RWS+CF, whisker test pulses were delivered for a period of at least 60 minutes. To coordinate precise stimulus delivery, a Cygnus Digital Stimulator (catalog: PG4000A, Cygnus Technology, Inc.) was interfaced with an Arduino Uno to trigger both the LED driver and picospritzer.

 Cerebellar imaging protocol

Calcium imaging of the genetically encoded indicator GCaMP6f was conducted in Crus I/II of the right hemisphere of awake, head-fixed mice using a laser scanning two-photon microscope (Neurolabware, Los Angeles, CA, USA) and Scanbox software (Scanbox, Los Angeles, CA). Calcium images were obtained at a frame rate of 30.98 fps with a pixel dimension of 512 x 796, using an 8KHz resonant scanning mirror with bidirectional scanning. A Mai Tai® DeepSee (Spectra-Physics, Milpitas, CA, USA) laser source was used to excite GCaMP6f at 920 nm. The fluorescence emission was collected through a 16x water immersion objective (Nikon LWD 0.8NA, 3 mm WD) using a GaAsP PMT (Hamamatsu Photonics, Shizuoka, Japan). A 2.0 or 2.8x digital zoom was applied during imaging, generating a field of view of 760 x 613 or 551 x 444 μm, respectively. To minimize background noise originating from ambient light, a custom light shield was fitted around the brain window and objective. This shield had a small opening such that the Thorlabs patch cable used for LED stimulation of climbing fibers could be positioned over the cerebellar window. The laser power was set to 1% with a PMT gain between 0.79 and 0.81, allowing for prolonged recording while minimizing phototoxicity.

 S1 imaging protocol

Calcium imaging of GCaMP6f was conducted in barrel cortex of the left hemisphere using the same hardware as above. Images were obtained at a frame rate of 30.98 Hz with a pixel dimension of 512 x 796. A 2.4x digital zoom was applied during imaging, generating a field of view of 638 x 515 μm. Two of 11 RWS+CF experiments and one of 10 RWS experiments were conducted using a 2.0x digital zoom (generating a field of view of 760 x 613 μm) and were added to the experimental cohorts, as results were not significantly different across any reported measures. A custom light shield was fitted around the brain window and objective to reduce background noise originating from ambient light. Note further isolation of the LED light used for optogenetic stimulation of climbing fibers in these recordings was provided by the sheath attached to the cerebellar window (see Cranial window surgeries for imaging of barrel cortex neurons). While observing the epifluorescence signal, small vessels on the putative surface of the brain were put into focus in the z plane and matched to a field of view selected during receptive field mapping (see Habituation). After adjusting the focus 250 μm below the putative pial surface, small adjustments were made in the x-y planes while observing calcium signal in real time to maximize responsivity to whisker stimulation and overlap between GCaMP6f and tdTomato or mCherry in the field of view. GCaMP6f was excited at 920 nm and tdTomato or mCherry (for DREADD experiments) at 1040 nm, with emitted fluorescence collected using two GaAsP PMTs. The laser power was set to 2%, and adjustments to the PMT gain were made due to slight variance in the level of GCaMP6f and tdTomato/mCherry signal across animals. PMT gain in the green channel was 0.73 (SD: 0.07) while PMT gain in the red channel was 0.83 (SD: 0.14), which were not significantly different across experimental groups. Responsivity to whisker stimulation was determined by intermittently delivering 8 psi air puffs at random intervals, with a minimum of 30 seconds between stimuli, to avoid influencing plasticity experiments. If more than ten minutes were spent determining an optimal field of view on the day of recording, the experiment was aborted and instead conducted the following day. Mice were occasionally used for multiple recording configurations but were sacrificed immediately after performing plasticity experiments, and selected animals were used for histological characterization (see Tissue preparation for fluorescence microscopy and immunohistochemistry). Note DREADD expression was histologically verified in all animals to ensure expression was restricted to neocortex.

 DREADD experimental protocol

To activate the hM4D(Gi) and hM3D(Gq) receptors, we used deschloroclozapine dihydrochloride (DCZ, MedChemExpress), which was chosen over clozapine N-oxide (CNO) as an agonist due to its heightened selectivity for hM4D(Gi) and hM3D(Gq) receptors over endogenous receptors, a more rapid onset than CNO, and significantly increased potency.94 DCZ was dissolved in DMSO at a 0.02 mg/mL concentration and stored at −80 °C. On the day of recording, DCZ aliquots were thawed to room temperature and diluted to 0.01 mg/mL with DMSO/saline. Two-photon calcium imaging experiments began with recordings of baseline responses to whisker stimulation for at least 40 minutes (with and without optogenetic stimulation of climbing fibers in a subset of experiments), followed by an injection of DCZ (0.1 mg/kg) intraperitoneally. For DREADD inhibition of PV neurons, a dose of 0.02 mg/kg was used instead, as the 0.1 mg/kg dose caused significant increases in neocortical excitability (note PV animals that had received a dose of 0.1 mg/kg were not used for any subsequent experiments or in any of the datasets shown here). Injections were carefully administered to ensure the field of view would not be disturbed; experiments were aborted if this occurred, as responses in each period could no longer be directly compared. Activation of receptors was allowed to take effect for fifteen minutes prior to resumption of recording, after which a second period of baseline activity was recorded. This period lasted between fifteen and thirty minutes, such that a period of fifteen minutes of stable activity was recorded prior to plasticity induction. Plasticity induction protocols were followed as above (see Optogenetic and tactile stimulation), and whisker test pulses were delivered for at least 40 minutes after plasticity induction. To ensure observed plasticity phenomena were not the result of prolonged changes in activity related to hM4D(Gi) or hM4D(Gq) activation, control experiments lasting the same duration as the plasticity experiments were conducted similarly, except without induction of plasticity (see Figure S5). To ensure DREADD expression was localized to the neocortex, viral spread was histologically characterized in all animals (see Tissue preparation for fluorescence microscopy and immunohistochemistry).

 Characterization of neurons

Calcium images taken in the green (PMT0) and red (PMT1) channels were first merged, trial-averaged, and concatenated using Fiji. The average image was then calculated, followed by identification of all cells expressing tdTomato or mCherry (applicable to all recordings except the zona incerta manipulation experiments). ROIs were manually drawn around each cell co-expressing GCaMP6f and tdTomato or mCherry. All images across a session were then concatenated and trial-averaged before calculating the maximum image. ROIs were manually drawn around all putative pyramidal neurons, which were identified with the following criteria: a triangular or pyramidal soma, the presence of a single apical dendrite extending in a similar orientation to other pyramidal cells in the same imaging session, and the presence of basal dendrites smaller in diameter than the apical dendrite. In a subset of experiments in which apical dendrites could not be clearly distinguished in the imaging plane, z-stack images were taken at the end of the imaging session, averaged, and referenced to visualize apical dendrites. After identification of fluorescently tagged interneurons and pyramidal neurons, relevant putative interneurons were morphologically identified and compiled (e.g., putative VIP and SST neurons in PV-tagged mice). Putative interneurons were identified by the presence of an elongated soma and bipolar dendritic arbors (likely VIP), round somas smaller in diameter than pyramidal cells, and neurons with multiple perisomatic dendrites (usually highly branched; likely PV). The morphology of untagged interneurons was visually compared to the morphology of tagged interneurons in relevant mice (e.g., morphologically identified interneurons in PVCre mice were compared to fluorescently tagged interneurons in SSTCre and VIPCre mice). The average (11.07µm, SD 1.64µm) and maximum (18.14µm, SD 2.74µm) soma diameter for morphologically identified pyramidal neurons (n = 3358) were significantly higher than in morphologically identified interneurons in the same mice (n = 3522; average: 9.90µm, SD 1.96; maximum: 16.37µm, SD: 3.39µm; p < 0.001 for both values, unpaired t-test) and tagged VIP, SST, and PV neurons (n = 543; average: 8.96µm, SD 1.87µm; maximum: 14.98µm, SD 3.51µm; p < 0.001 for both values, unpaired t-test).

 Intrinsic Signal Imaging Image acquisition

Mice underwent the same surgical procedures described in Two-Photon calcium imaging: Cranial window surgeries for imaging of barrel cortex neurons. To facilitate manipulation of single whiskers, mice were lightly anesthetized with 10 mg/kg ketamine and placed on a stereotaxic frame. Using a custom GUI written in MATLAB, small vessels were visualized under a green LED (525 nm wavelength) with a high-speed camera (Teledyne DALSA Genie Nano GigE). Focus in the z-plane was adjusted approximately 250 μm below the putative pial surface before taking a reference image of the vasculature. For measurement of intrinsic signals, the field of view was illuminated with red light (625 nm). An Arduino Uno and MATLAB were used to trigger trial onset, the device used to mechanically deflect whiskers (a glass capillary tube attached to a rotary motor), a blue LED to stimulate climbing fibers, and recording of intrinsic signals at 30 Hz. Trials consisted of a 1-second baseline period, followed by a 1-second test pulse in which whiskers contralateral to the optical window were mechanically deflected at 8 Hz (RWS). This was repeated with an inter-trial interval of 18 seconds for 30 minutes prior to plasticity induction, after which whiskers were stimulated at 8 Hz for 5 minutes. Test pulses were delivered for 40 minutes after plasticity induction. For RWS+CF and control experiments, 50 ms blue light pulses were delivered at 1 Hz during plasticity induction with or without 8 Hz co-stimulation of whiskers, respectively.

 Image processing

Evoked signals were calculated similarly to Vasquez et al 95. Briefly, images were smoothed with a Gaussian filter and spatially downsampled by a factor of four. For each trial, an average baseline reflectance image (R0) was generated by calculating the average across the 1-second (30 frame) period before stimulus onset. Post-stimulus reflectance images were temporally averaged across 200 ms bins for 600 ms total, starting 400 ms after stimulus onset. The change in reflectance, ΔR/R, was calculated by subtracting R0 from each post-stimulus image and dividing the result by R0. Each of the three bins (400-600 ms, 600-800 ms, and 800 ms-1 s post stimulus onset) were averaged across a minimum of 30 trials and finally summed to yield a single total stimulus-evoked ΔR/R image. The reference image taken with green light illumination was used to draw a region of interest around the cranial window. Pixels outside this ROI mask were set to zero. Binary images were generated by calculating z-scores of the ΔR/R image and thresholding values below a z-score of -1.5 (note negative changes in reflectance indicate an increase in the presence of deoxyhemoglobin caused by increased oxygen consumption by active neurons). Binarized images were overlaid onto reference images. To calculate the areal extent of single whisker-evoked responses, clusters of active pixels were identified by applying a median filter with a 5 x 5-pixel neighborhood size, and the number of pixels above threshold was calculated. To reduce the impact of residual noise (i.e. detected pixels outside the area reliably activated by a single whisker), values were normalized by the area of the cranial window.

 Two-photon image processing Processing of cerebellar data

Processing of Purkinje cell calcium traces was conducted as previously described23, 73. Individual trials from each session were concatenated along the z-axis, representing a time series across the entire recording session, to ensure consistency among trials after motion correction. The concatenated file underwent motion correction using a custom MATLAB script (MATLAB R2017b, MathWorks, Natick, MA, USA) based on whole frame cross-correlation (provided by the lab of Mark Sheffield, University of Chicago). Cellular regions of interest were manually selected in Fiji based on the average field of view across trials. Trials were then separated for further processing. In a subset of trials, optogenetic stimulus artifacts were visible in a portion of the field of view in 1-3 nonconsecutive frames, depending on the stimulus duration. These frames were manually identified and excluded using Fiji before further processing.

Using custom MATLAB script, ΔF/F values were calculated from the raw trace using the following equation: (Ft − F0)/F0, where Ft represented the raw calcium intensity of the time series, and the baseline fluorescence (F0) was set as the 20th percentile of the fluorescence trace from each trial. ΔF/F values were then subjected to low-pass filtering using a five-frame moving window smoothing function. Finally, traces were normalized to a pre-stimulus baseline period by subtracting the average ΔF/F value from the 5-frame period prior to stimulus onset from the entire trace. To visualize population average responses to optogenetic stimulation of climbing fibers, calcium signals from each ROI were first trial-averaged, followed by concatenation and averaging of the signal across cells and mice. To calculate maximum amplitude of responses in the evoked period, the maximum in a window 0-700 ms after stimulus onset was calculated for each ROI’s trial-averaged trace, then concatenated across cells and mice. Area under the curve (AUC) was calculated similarly by integrating the ΔF/F values over the same window. To identify spontaneous events, frame indices in which the first and second derivatives of the ΔF/F signal reached 0.04 were identified (Fidx). Peaks were detected using a peak prominence threshold of 0.1 ΔF/F, with spontaneous events defined as peaks occurring within 200 ms of Fidx. The onset of each spontaneous event was determined as the first frame in which the first derivative of the ΔF/F dropped below 0.05 (working backward from frame index of the detected peak). A 1-second window following event onset was extracted. Finally, after averaging all detected spontaneous events within ROIs, spontaneous events were compiled across neurons to generate the population average trace.

 Processing of S1 data

Motion correction and ROI curation were performed as above (see Processing of cerebellar data and Two-photon calcium imaging: Characterization of neurons). Residual optogenetic stimulus artifacts were observed in a single frame in a subset of trials, typically at or immediately after stimulus offset. These artifacts were detected using the findpeaks function in MATLAB and corrected by replacing the artifact frame’s value with a linear interpolation of the two adjacent frames before further processing. Background subtraction was performed by calculating the bottom 1st percentile of the raw fluorescence trace collected from an ROI that did not contain cell bodies or neuropil, then subtracting this value from all trials. Smoothing, expression of raw signals as ΔF/F, calculation of baseline noise (σ), and calculation of the signal-to-noise ratio (SNR) were performed as in Ayaz et. al.96. Smoothing was performed with a 51-point 1st-order Savitsky-Golay filter, followed by calculation of the relative percent change of fluorescence ((Ft − F0)/F0) using the 1st percentile of the smoothed trace as F0. Given background subtraction prior to calculation of F0 may generate negative F0 values and subsequent signal inversion, ROIs with F0 < 0 were rejected from further analyses (44/7423 cells). To calculate σ for each neuron, a 5-second (155 frame) sliding window was first used to calculate the standard deviation of the fluorescence change during each 5-second period within a session. After compiling all values, the 1st percentile value was taken as σ. For each neuron, the SNR was defined as the 95th percentile of ΔF/F signals recorded across all trials divided by the baseline noise.

 Evoked event detection

To detect whisker–evoked events, we first applied OASIS deconvolution to the ΔF/F traces using an event detection threshold of 3σ to generate a binary trace containing event times. An autoregressive model with order p=1 was used to ensure the number of events (“spikes”) for each calcium transient varied with the amplitude of the calcium transient as previously observed52. Next, for each frame, the sum of all events was calculated and divided by the number of trials to determine the proportion of trials with an event in each frame (Pevent). Pevent was averaged across frames, and consecutive frames with Pevent 3 standard deviations above the mean were selected. This generated an evoked event window of 387 ms (Figure S2B). Finally, for each neuron, the calcium traces for all trials containing an event within the evoked event window were compiled for further analysis. To visualize population average traces for all experiments, evoked calcium signals for each neuron were first averaged across trials, followed by averaging across neurons. For visualizations between the same neural populations in different conditions, only those cells that responded to both conditions, i.e. having an evoked event in each condition, were included (following the definition of “Persistent cells” used in Williams et. al.60). Shaded regions in population average traces represent SEM.

 Calculation of amplitude and area under the curve in S1 recordings

For all calculations of AUC in S1 recordings, the ΔF/F signal was integrated over a 700 ms window after stimulus onset unless otherwise specified (details below). Note that AUC is equal to the average response in the 700 ms window, multiplied by the 700 ms window duration (thus, AUC is directly proportional to the mean). We choose to report AUC, a descriptive statistic, rather than the mean within this 700 ms window.

To quantify plasticity phenomena caused by RWS or RWS+CF, all evoked events measured within 50 minutes prior to plasticity induction (“Pre”) or 60 minutes after plasticity induction (“Post”) were extracted for each cell and trial-averaged before calculating AUC. Only the neurons with evoked events both pre- and post-plasticity induction were included in subsequent analyses. AUC were averaged both across all neurons or within animal to perform statistics across neurons or mice (see Statistics and quantifications). To visualize the duration of plasticity effects, evoked events were binned into 10-minute epochs and trial-averaged within cells, followed by calculation of the AUC and averaging across neurons. To verify the robustness of plasticity effects, AUC and time course measurements were also analyzed across all cells and trials (i.e., those that did not have stimulus-evoked events; Figure S2) and in the absence of animal movement (Figure S3; see Determination of active and rest trials). To quantify plasticity phenomena in DREADD manipulation experiments, response amplitude was calculated in addition to AUC before and after plasticity induction. Response amplitude was taken as the maximum ΔF/F value in the same 700 ms window used for AUC calculations.

To quantify the immediate effect of climbing fiber activation on basic whisker responses in S1 cortex, AUC was calculated similarly for whisker stimulation (W) or whisker and CF coactivation (W+CF) trials. An additional measurement was taken 650-850 ms after stimulus onset for pyramidal neurons and VIP interneurons, as the effect of CF modulation was present in the later phase of the calcium response. Note that while W and W+CF test pulses were delivered prior to DCZ administration in DREADD experiments to verify functional opsin expression, these mice were excluded from this analysis due to an insufficient number of trials in each animal.

To verify that hM4D(Gi) and hM3D(Gq) expression in SST neurons indeed suppressed or increased their activity, respectively, AUC was calculated before and after DCZ administration in a window 1750 ms after stimulus onset. This larger window was used to capture the significantly increased duration of whisker-evoked calcium responses in SST neurons expressing hM3D(Gq). To verify that hM4D(Gi) and hM3D(Gq) expression in VIP neurons indeed suppressed or increased their activity, respectively, AUC was calculated before and after DCZ administration in a window 650-850 ms after stimulus onset. Given VIP neurons are modulated by climbing fiber stimulation in this phase of the calcium response (Figure 2G), this later window was chosen to ensure DREADD manipulation of VIP neurons suppressed or enhanced VIP responses in the physiologically relevant response period.

 Determination of active and rest trials

Activity of mice was monitored at 30.98 Hz using a DALSA M640 CCD camera (Teledyne Technologies, Thousand Oaks, CA, USA) focused on the animal’s face and a portion of the running wheel. To determine the frames in which mice were active, custom MATLAB script was used to generate a binary trace that indicated the frames in which movement between frames occurred (based on a comparison of pixel values between consecutive frames). For each recording session, movement traces and corresponding videos from ten randomly selected trials were manually inspected to ensure all frames in which mice were whisking or running were indicated. Note whisking occurred independently of running, but running seldom occurred without whisking. Rest trials were defined as any trial with no movement detected at least 400 ms before and after stimulus onset – longer than the evoked event window, and the same as the window chosen in Ayaz et. al.96 to distinguish rest from running onset. All other trials were considered active trials.

Statistics and quantifications

All statistics and quantifications were performed using MATLAB. To assess normality, we applied the Lilliefors test to each dataset before performing statistical calculations. For paired samples, a paired t-test was used when data were normally distributed, and a Wilcoxon signed rank test was used for non-normal data. For unpaired comparisons, an unpaired t-test was used for normally distributed data, a Welch’s t-test for normally distributed data with unequal variances, and a Wilcoxon rank-sum test for non-normal distributions. Categorical variables were compared using the Chi-squared test. Two-way ANOVA was used for multi-group comparisons (i.e., to determine whether plasticity effects across time were significantly different between groups). To perform statistics across all animals, average responses were first normalized by dividing the response in the relevant experimental condition by the baseline response. A one-sample t-test was then performed against a standard value of 1. Statistical significance was set to p < 0.05, with p values presented in figures as follows: p < 0.05: *; p < 0.01: **; p < 0.001: ***; p > 0.05: not significant (n.s.). Bar graphs represent mean ± SEM. Box plots contain the interquartile range (25th and 75th percentiles), median, and range.

 Data availability

All data and code have been deposited on Github: https://github.com/abbysilbaugh/climbingfiber

 Supplementary figures and tables Optogenetic CF activation causes GCaMP6f-encoded calcium transients in Purkinje cells.

(A) Schematic of IO injection. Top left: Cerebellum (calbindin-staining of PCs) and IO (as in Fig 1B). Bottom: ChR2 expression in the principal IO (IOPr) and dorsal IO (IOD). Scale bars: 1mm (top) and 100µm (bottom). (B) Overlap of VGluT2-stained excitatory terminals with ChR2 expression in the contralateral cerebellar cortex. Scale bar: 40µm. (C) Post-hoc verification of ChR2 expression in CF terminals in crus I and II. Scale bars: 50µm. (D) Measure of distance from the pial surface to the cerebellar nuclei. While the IO innervates both the cerebellar cortex and nuclei directly, it is unlikely that CF terminals other than the superficial ones innervating PCs are optogenetically activated as the intensity of blue light decreases exponentially in brain tissue. Scale bar: 300µm. Dashed line: 2mm. (E) Experimental schematic for F-I. (F) Test of various stimulus protocols: 3×15ms, 2×20ms, 1×50ms (interval 50ms) and (3×15ms)*2 (interval 50ms). Traces show response averages (n=102 cells from one mouse). (G) Bar graphs showing response peaks (left) and integrals (right) from the corresponding data in F. Data: mean ± SEM. (H) Calcium transients evoked by 470nm, 50ms LED light pulses in PCs in crus I/II (n=131 cells; N=2 mice) in comparison to spontaneous events (left). Scale bars: 0.1 ΔF/F; 0.5s. (I) Amplitude and latency to peak measures from the corresponding data in H. Data: mean ± SEM. Scale bars: 0.1 ΔF/F; 0.5s.

 Plasticity in different neuron types.

(A) Experimental schematic. (B) Event probability around stimulus onset with the window for detection of stimulus-evoked events (orange; three standard deviations above the mean event probability, calculated across all frames collected during the 20-second trial periods). (C) Pie chart indicating the percentage of L2/3 pyramidal cells recruited or suppressed by RWS/RWS+CF activity (only responsive to whisker stimulation in post- or pre- periods, respectively) and how many remained entirely unresponsive. (D) Left: RWS-evoked response changes in L2/3 pyramidal neurons (n=329 cells; N=12 mice). Right: RWS+CF-evoked response changes in L2/3 pyramidal neurons (n=597 cells; N=11 mice). Box plots quantify calcium responses across all cells and trials (i.e. including unresponsive neurons and trials without whisker-evoked responses). (E) The same as D in INs in the same mice (n=616 cells for RWS; n=590 cells for RWS+CF). (F, G) Same as (D, E) in tdTomato-tagged SST interneurons (F; n=41 cells and N=3 mice for RWS; n=31 cells and N=3 mice for RWS+CF) and PV interneurons (G; n=83 cells; N=5 mice for RWS; n=113 cells; N=5 mice for RWS+CF). Box: 25th and 75th percentiles; line: median; whiskers: range. Trace scale bars: 0.05 ΔF/F; 0.5s.

 Impact of RWS- and RWS+CF-stimulation on animal motion.

(A) Experimental schematic. Mice are awake and head-fixed but can walk/run on the treadmill. (B) Average cross-correlation coefficient between neural calcium signals and movement activity data, segregated by neuron type. The numbers below the box charts indicate the number of recorded cells. (C) Left: probability distribution of animal movement in Rest trials, demonstrating the absence of movement both before and after stimulus onset. Right: probability distribution of animal movement in Active trials. (D) Pie charts indicating the percentage of Rest, Active and Stop trials pre-vs post-RWS (left) or RWS+CF (right). (E) Average probability of animal movement in Rest trials pre- and post-RWS (left) or RWS+CF (right) over the same time scale as the neural traces shown below. (F) and (H) RWS-evoked potentiation in L2/3 pyramidal neurons (n=124; N=12) and RWS+CF-mediated suppression of potentiation (n=388; N=11) in the absence of animal movement (i.e. in Rest trials). (G) and (I) Analysis for INs in the same mice (n=212 cells for RWS; n=296 cells for RWS+CF). Note motion activity traces capture both whisking and running activity. Box: 25th and 75th percentiles; line: median; whiskers: range. Trace scale bars: 0.2 ΔF/F; 0.5s.

 Intrinsic optical imaging confirms CF-mediated suppression of S1 plasticity.

(A) Experimental schematic. Intrinsic optical imaging is performed using a high-speed camera. The field of view is illuminated with red light (625 nm). A decrease in ΔR/R indicates increased oxygen consumption by neurons activated by single whisker stimulation (blue arrow). (B) Top: Example of S1 plasticity evoked by stimulation of the C2 whisker at 10Hz. Bottom: Example of suppression of response potentiation when whisker stimulation is paired with optogenetic CF co-activation at 1Hz. (C) Quantification of RWS-evoked potentiation of the single whisker response area (blue; N=4 mice) and RWS+CF-mediated suppression of response potentiation (red; N=4 mice). No response potentiation is observed in the absence of repeated whisker stimulation or CF activation (grey; N=4 mice). Data are normalized to the pre-plasticity response area. Light blue line indicates mean.

 Plasticity effects in L2/3 PNs vs controls in chemogenetic manipulation experiments.

(A) Chemogenetic activation of SSTs. Left: Experimental schematic. Middle: circuit diagram. Right: DCZ injection activates SSTs (yellow). Top: Control protocol. ‘Post’ marks the period equivalent in time to the period for the measurement of plasticity effects after RWS (bottom). (B) Responses in L2/3 PNs during Pre, DCZ wash-in, and Post periods in control experiments. (C) The difference in response (ΔAUC) of L2/3 pyramidal cells to whisker stimulation between the DCZ wash-in and Post periods; (control: n=218 cells; N=3 mice; plasticity: n=253 cells; N=4 mice); negative ΔAUC values indicate AUC values in the Post period are lower than in the DCZ wash-in period. The ΔAUC in control and plasticity experiments are normalized to the absolute ΔAUC in control experiments to demonstrate changes in the plasticity experiments relative to controls. Note the absence of PN potentiation in the plasticity experiment. (D) – (F) Equivalent measures during chemogenetic VIP inhibition (red; control: n=173 cells; N=3 mice; plasticity: n=164 cells; N=3 mice). Note the absence of PN potentiation in the plasticity experiment. (G) – (I) Equivalent measures during chemogenetic SST inhibition (control: n=175 cells; N= 3 mice; plasticity: n=181 cells; N=3 mice). (J) – (L) Equivalent measures during chemogenetic PV inhibition (control: n=368 cells; N=4 mice; plasticity: n=258 cells; N=3 mice). PNs are significantly potentiated after plasticity experiments beyond changes in controls. (m) – (o) Equivalent measures during chemogenetic VIP activation (control: 145 cells; N=3 mice; plasticity: n=197 cells; N=4 mice). PNs are significantly potentiated after plasticity experiments beyond changes in controls. Optogenetic CF co-activation is applied during RWS in the experiments shown in (G) – (O). Data: mean ± SEM.

 Cerebellar nuclei do not strongly project to thalamic reticular nucleus.

(A) Schematic of labeling approach (top) and injection site of retrograde tracer (bottom). Scale bar: 1mm. (B) Image of retrograde label from the somatosensory thalamus (POm and VPM; cyan) and orthograde label from the contralateral cerebellar nuclei (red) taken approximately -1.0mm posterior from bregma, demonstrating lack of overlap between neurons in the thalamic reticular nucleus (TRN) and cerebellar nuclei terminals in the adjacent thalamic nuclei. Representative images are shown from two separate mice. Scale bars: 1mm (left images) and 250µm (right images). (C) Projection diagram. DCN: deep cerebellar nuclei; ZI: zona incerta; POm: posterior medial thalamic nucleus; TRN: thalamic reticular nucleus; VPM: ventral posterior medial nucleus; VPL: ventral posterior lateral nucleus; VA: ventral anterior nucleus; VL: ventral lateral nucleus.

 Transsynaptic labeling identifies projections from the cerebellar nuclei to POm.

(A) Schematic of dual-injection approach to label outputs of ZI neurons receiving input from the contralateral CBN. (B) Top left: EYFP-expressing terminals and cell bodies within ZI (dashed lines) and axonal projections out of the ZI (arrowheads). Scale bar: 100µm. Bottom left: projections from the ZI terminate in POm (arrowheads). Scale bar: 50 µm. Right: zoomed-out view showing projections from ZI to POm in the same plane. Scale bar: 200 µm. (C) ZI neurons receiving cerebellar input project to S1. Synaptic terminals are observed in L1 and L5a. Scale bar: 100 µm. (D) Series of coronal sections showing details of the projections from the contralateral CBN to ZI and POm. Scale bars: 1mm. (E) The same as D in a second mouse, showing robust EYFP expression across the rostro-caudal extent of ZI. Scale bars: 1mm. POt = posterior thalamic nucleus (triangular part). Colored dots mark images taken from the same sections.

 Whole-cell patch-clamp recordings are performed from neurons in the ZI that project to POm and receive input from cerebellar nuclei.

(A) Left: Differential interference contrast (DIC) image showing the ZI (white dashed circle) at low magnification (top; scale bar: 500µm), and a neuron selected for patching at higher magnification (bottom; scale bar: 100µm). Middle left: same images showing ChR2-expression. Middle right: same images showing retrograde fluoro-Ruby label originating in the POm. Right: merged images (boxed area with yellow dashed line marks the recordings area shown in the images below). (B) Post-hoc histology of an acute slice used for recordings demonstrating the details of cerebellar innervation of the ZI, separated by dorsal ZI (dZI) and ventral ZI (vZI; image corresponds to Fig. 6D, separated by channel). Scale bar: 500µm. (C) Quantification of the terminal fluorescence segregated by dZI and vZI. Data: mean ± SEM.

 Data and statistical tests corresponding to <xref rid="fig1" ref-type="fig">Figure 1</xref>. Data and statistical tests corresponding to <xref rid="fig2" ref-type="fig">Figure 2</xref>. Data and statistical tests corresponding to <xref rid="fig3" ref-type="fig">Figures 3</xref>, <xref rid="fig4" ref-type="fig">4</xref>, and <xref rid="fig5" ref-type="fig">5</xref>. Data and statistical tests corresponding to <xref rid="fig6" ref-type="fig">Figure 6</xref>. Data and statistical tests corresponding to Supplementary Figure 1. Data and statistical tests corresponding to Supplementary Figure 2. Data and statistical tests corresponding to Supplementary Figure 3. Data and statistical tests corresponding to Supplementary Figure 4. Data and statistical tests corresponding to Supplementary Figure 5. List of virus titers. Acknowledgements

We thank Drs. M. Brecht (Humboldt University, Berlin) as well as A.M. Oswald and S.M. Sherman (both University of Chicago) for invaluable feedback on the manuscript. We thank S.R. Postlewaite and current and former Hansel lab members S.E. Busch, D. Huang and T.F. Lin for insightful discussions and comments on the manuscript. This work was supported by National Institutes of Health (NINDS) grants R21NS136954 (to C.H.) and NS094184 (to K.P.K. in the lab 940 of S.M. Sherman).

 References K. Doya, Complementary roles of basal ganglia and cerebellum in learning and motor control. Curr. Opin. Neurobiol. 10, 732–739 (2000). M. E. Diamond, M. Armstrong-James, F. F. Ebner, Experience-dependent plasticity in adult rat barrel cortex. Proc. Natl. Acad. Sci. USA 90, 2082–2086 (1993). DOI:10.1073/pnas.905.2082. S. Glazewski, A. L. Barth, H. Wallace, M. McKenna, A. Silva, K. Fox, Impaired experience-dependent plasticity in barrel cortex of mice lacking the alpha and delta isoforms of CREB. Cereb. Cortex 9, 249–256 (1999). E.A. Stern, M. Maravall, K. Svoboda, Rapid development and plasticity of layer 2/3 maps in rat barrel cortex in vivo. Neuron 31, 305–315 (2001). Doi: 10.1016/s0896-6273(01)00360-9. J. Delacour, O. Houcine, J. C. Costa, Modifications of the responses of barrel field neurons to vibrissal stimulation during theta in the awake and undrugged rat. Neuroscience 37, 237–243 (1990). P. Megevand, E. Troncoso, C. Quairiaux, D. Muller, C. M. Michel, J. Z. Kiss, Long-term plasticity in mouse sensorimotor circuits after rhythmic whisker stimulation. J. Neurosci. 29, 5326–5335 (2009). F. Gambino, S. Pages, V. Kehayas, D. Baptista, R. Tatti, A. Carleton, A. Holtmaat, Sensory-evoked LTP driven by dendritic plateau potentials in vivo. Nature 515, 116–119 (2014). D. E. Feldman, Timing-based LTP and LTD at vertical inputs to layer II/III pyramidal cells in rat barrel cortex. Neuron 27, 45–56 (2000). L. E. Williams, A. Holtmaat, Higher-order thalamocortical inputs gate synaptic long-term potentiation via disinhibition. Neuron 101, 91–102 (2019). D. E. Feldman, M. Brecht, Map plasticity in somatosensory cortex. Science 310, 810–815 (2005). J. D. Schmahmann, J. C. Sherman, The cerebellar cognitive affective syndrome. Brain 121, 561579 (1998). C. J. Stoodley, P. T. Tsai, Adaptive prediction for social contexts: the cerebellar contribution to typical and atypical social behaviors. Annu. Rev. Neurosci. 44, 475–493 (2021). F. A. Middleton and P. L. Strick, Cerebellar projections to the prefrontal cortex of the primate. J. Neurosci. 21, 700–712 (2001). P. L. Strick, R. P. Dum, J. A. Fiez, Cerebellum and non-motor function. Annu. Rev. Neurosci., 32, 413–434 (2009). A. Badura, J. L. Verpeut, J. W. Metzger, T. D. Pereira, T. J. Pisano, B. Deverett, D. E. Bakshinskaya, S. S. Wang, Normal cognitive and social development require posterior cerebellar activity. eLife 7, e36401 (2018). 10.7554/eLife.36401 E. Kelly et al., Regulation of autism-relevant behaviors by cerebellar-prefrontal cortical circuits. Nat. Neurosci. 23, 1102–1110 (2020). T. J. Pisano et al., Homologous organization of cerebellar pathways to sensory, motor, and associative forebrain. Cell Rep. 36, 109721 (2021). Z. Gao, C. Davis, A.M. Thomas, M.N. Economo, A.M. Abrego, K. Svoboda, C.I. De Zeeuw, N. Li. A cortico-cerebellar loop for motor planning. Nature 563, 113–116. 2018 D. Popa, M. Spolidoro, R. D. Proville, N. Guyon, L. Belliveau, C. Lena, Functional role of the cerebellum in gamma-band synchronization of the sensory and motor cortices. J. Neurosci. 33, 6552–6556 (2013). S. McAfee, Y. Liu, R. V. Sillitoe, D. H. Heck, Cerebellar lobulus simplex and Crus I differentially represent phase and phase difference of prefrontal cortical and hippocampal oscillations. Cell Rep. 27, 2328–2334 (2019). S. Lindeman, S. Hong, L. Kros, J. F. Mejias, V. Romano, R. Oostenveld, M. Negrello, L. W. Bosman, C. I. De Zeeuw, Cerebellar Purkinje cells can differentially modulate coherence between sensory and motor cortex depending on region and behavior. Proc. Natl. Acad. Sci. USA 118, e2015292118 (2021). E. R. Goldenkoff, J.A. Brissenden, T.G. Lee, K.J. Michon, M. Vesia. Cerebellar activity affects distal cortical physiology and synaptic plasticity in a human parietal-motor pathway associated with motor actions. J. Neurosci. 45, e0404252025. 2025 S. E. Busch, C. Hansel, Climbing fiber multi-innervation of mouse Purkinje dendrites with arborization common to human. Science 381, 420–427 (2023). D. Marr, A theory of cerebellar cortex. J. Physiol. 202, 437–470 (1969). J. S. Albus, A theory of cerebellar function. Math. Biosci. 10, 25–61 (1971). M. Ito, M. Yoshida. The cerebellar-evoked monosynaptic inhibition of Deiters’ neurons. Experientia 20, 515–516 (1964). M. Ito, M. Kano, Long-lasting depression of parallel fiber-Purkinje cell transmission induced by conjunctive activation of parallel fibers and climbing fibers in the cerebellar cortex. Neurosci. Lett. 33, 253–258 (1982). M. Ito, M. Sakurai, P. Tongroach, Climbing fibre induced depression of both mossy fibre responsiveness and glutamate sensitivity of cerebellar Purkinje cells. J. Physiol. 265, 833–854 (1982). S. S. Wang, W. Denk, M. Häusser, Coincidence detection in single dendritic spines mediated by calcium release. Nat. Neurosci. 3, 1266–1273 (2000). V. Lev-Ram, S. T. Wong, D. R. Storm, R. Y. Tsien, A new form of cerebellar potentiation is postsynaptic and depends on nitric oxide but not cAMP. Proc. Natl. Acad. Sci. USA 99, 83898393 (2002). M. Coesmans, J. T. Weber, C. I. De Zeeuw, C. Hansel, Bidirectional parallel fiber plasticity in the cerebellum under climbing fiber control. Neuron 44, 691–700 (2004). C. Piochon, H. K. Titley, D. H. Simmons, G. Grasselli, Y. Elgersma, C. Hansel, Calcium threshold shift enables frequency-independent control of plasticity by an instructive signal. Proc. Natl. Acad. Sci. USA 113, 13221–13226 (2016). A. Suvrathan, H. L. Payne, J. L. Raymond, Timing rules for synaptic plasticity matched to behavioral function. Neuron 92, 959–967 (2016). H. K. Titley, M. Kislin, D. H. Simmons, S. S. Wang, C. Hansel, Complex spike clusters and false-positive rejection in a cerebellar supervised learning rule. J. Physiol. 597(16), 4387–4406 (2019). J. F. Medina, J. C. Repa, M. D. Mauk, J. E. LeDoux, Parallels between cerebellum- and amygdala-dependent conditioning. Nat. Rev. Neurosci. 3, 122–131 (2002). Y. Yang, S. G. Lisberger, Interaction of plasticity and circuit organization during the acquisition of cerebellum-dependent motor learning. eLife 2, e01574 (2013). 10.7554/eLife.01574 Y. Yang, S. G. Lisberger, Purkinje-cell plasticity and cerebellar motor learning are graded by complex-spike duration. Nature 510, 529–532 (2014). N. T. Silva, J. Ramirez-Buritica, D. L. Pritchett, M. R. Carey, Climbing fibers provide essential instructive signals for associative learning. Nat. Neurosci. 27, 940–951 (2024). J. I. Simpson, D. R. Wylie, C. I. De Zeeuw, On climbing fiber signals and their consequence(s). Behav. Brain Sci. 19, 384–398 (1996). L. W. Bosman et al., Encoding of whisker input by cerebellar Purkinje cells. J. Physiol. 588, 3757–3783 (2010). D. J. Herzfeld, R. Shadmehr, Cerebellum estimates the sensory state of the body. Trends Cogn. Neurosci. 18, 66–67 (2014). F. Najafi, A. Giovannucci, S. S. Wang, J. F. Medina, Sensory-driven enhancement of calcium signals in individual Purkinje dendrites of awake mice. Cell Rep. 6, 792–798 (2014a). F. Najafi, A. Giovannucci, S. S. Wang, J. F. Medina, Coding of stimulus strength via analog calcium signals in Purkinje cell dendrites of awake mice. eLife 3, e03663 (2014b). 10.7554/eLife.03663 M. A. Gaffield, A. Bonnan, J. M. Christie, Conversion of graded presynaptic climbing fiber activity into graded postsynaptic Ca2+ signals by Purkinje cell dendrites. Neuron 102, 762–769 (2019). S. Ohmae, J. F. Medina, Climbing fibers encode a temporal-difference prediction error during cerebellar learning in mice. Nat. Neurosci. 18, 1798–1803 (2015). W. Heffley, C. Hull, Classical conditioning drives learned reward prediction signals in climbing fibers across the lateral cerebellum. eLife 8, e46764 (2019). 10.7554/eLife.46764 D. Kostadinov, M. Beau, M. Blano-Pozo, M. Häusser, Predictive and reactive reward signals conveyed by climbing fiber inputs to cerebellar Purkinje cells. Nat. Neurosci. 22, 950–962 (2019). D. H. Simmons, S. E. Busch, H. K. Titley, G. Grasselli, J. Shi, X. Du, C. Wei, C. M. Gomez, C. Piochon, C. Hansel, Sensory over-responsivity and aberrant plasticity in cerebellar cortex in a mouse model of syndromic autism. Biol. Psychiatry: GOS 2, 450–459 (2022). J. S. Jouhanneau, L. Ferrarese, L. Estebanez, N. J. Audette, M. Brecht, A. L. Barth, J. F. Poulet, Cortical fos-GFP expression reveals broad receptive field excitatory neurons targeted by Pom. Neuron 84, 1065–1078 (2014). N. J. Audette, S. M. Bernhard, A. Ray, L. T. Stewart, A. L. Barth, Rapid plasticity of higher-order thalamocortical inputs during sensory learning. Neuron 103, 277–291 (2019). M. E. Diamond, M. von Heimendahl, P. M. Knutsen, D. Kleinfeld, E. Ahissar, ‘Where’ and ‘what’ in the whisker sensorimotor system. Nature Rev. Neurosci. 9, 601–612 (2008). L. Theis et al., Benchmarking spike rate inference in population calcium imaging. Neuron 90, 471–482 (2016). M. Brecht, A. Roth, B. Sakmann, Dynamic receptive fields of reconstructed pyramidal cells in layers 2 and 3 of rat somatosensory barrel cortex. J. Physiol. 553, 243–265 (2003). J. Heiss, Y. Katz, E. Ganmor, I. Lampl, Shift in the balance between excitation and inhibition during sensory adaptation of S1 neurons. J. Neurosci. 28, 13320–13330 (2008). J.F. Staiger, C.C.H. Petersen, Neuronal circuits in barrel cortex for whisker sensory perception. Physiol. Rev. 101, 353–415 (2021). R. Tremblay, S. Lee, B. Rudy, GABAergic interneurons in the neocortex: from cellular properties to circuits. Neuron 91, 260–292 (2016). T. Bonhoeffer, A. Grinvald, Iso-orientation domains in cat visual cortex are arranged in pinwheel-like patterns. Nature 353, 429–431 (1991). B. S. Sermet, P. Truschow, M. Feyerabend, J. M. Mayrhofer, T. B. Oram, O. Yizhar, J. F. Staiger, C. C. H. Petersen, Pathway-layer- and cell-type-specific thalamic input to mouse barrel cortex. eLife 8, e52665 (2019). 10.7554/eLife.52665 N. J. Audette, J. Urban-Cieko, M. Matsushita, A. L. Barth, POm thalamocortical input drives layer-specific microcircuits in somatosensory cortex. Cereb. Cortex 28, 1312–1328 (2018). L. E. Williams, L. Küffer, T. Bawa, E. Husi, S. Pagès, A. Holtmaat, Repetitive sensory stimulation potentiates and recruits sensory-evoked cortical population activity. J. Neurosci. 45, e2189232024 (2025). M. Novello, L.W. Bosman, C. I. De Zeeuw, A systematic review of direct outputs from the cerebellum to the brainstem and diencephalon in mammals. Cerebellum 23, 210–239 (2022). Y. Zhao, J. Wu, J. Feng, X. Cai, X. Wang, L. Wang, W. Xie, Y. Gu, J. Liu, W. Chen, L. Zhou, Y. Shen, Dual and plasticity-dependent regulation of cerebello-zona incerta circuits on anxiety-like behaviors. Nature Comm. 16, 3339 (2025). P. Bartho, T. F. Freund, L. Acsady, Selective GABAergic innervation of thalamic nuclei from zona incerta. Eur. J. Neurosci. 16, 999–1014 (2002). R. I. Martinez-Garcia, B. Voelcker, J. B. Zaltsman, S. L. Patrick, T. R. Stevens, B.W. Connors, S. J. Cruikshank, Two dynamically distinct circuits driving inhibition in sensory thalamus. Nature 22, 813–818 (2020). B. Zingg, X.L. Chou, Z.G. Zhang, L. Mesik, F. Liang, H.W. Tao, L.I. Zhang, AAV-mediated anterograde transsynaptic tagging: mapping cortico-collicular input-defined neural pathways for defense behaviors. Neuron 93, 33–47 (2017). J.H. Cho, K. Deisseroth, V.Y. Bolshakov, Synaptic encoding of fear extinction in mPFC-amygdala circuits. Neuron 80, 1491–1507 (2013). J. C. Trageser, K. A. Burke, R. Masri, Y. Li, L. Sellers, A. Keller, State-dependent gating of sensory inputs by zona incerta. J. Neurophysiol. 96, 1456–1463 (2006). M. Zhu, S. J. Kuhlman, A. L. Barth, Transient enhancement of stimulus-evoked activity in neocortex during sensory learning. Learn. Mem. 31, a053870.123 (2024). J.M. Bower, Is the cerebellum sensory for motor’s sake or motor for sensory’s sake: the view from the whiskers of a rat? Prog. Brain Res. 114, 483–516 (1997). O. Baumann et al., Consensus paper: the role of the cerebellum in perceptual processes. Cerebellum 14, 197–220 (2015). L.W. Bosman et al., Anatomical pathways involved in generating and sensing rhythmic whisker movements. Front. Integr. Neurosci. 5, 53 (2011). K.B. Ramakrishnan, K. Voges, L. De Propris, C.I. De Zeeuw, E. D’Angelo, Tactile stimulation evokes long-lasting potentiation of Purkinje cell discharge in vivo. Front. Cell. Neurosci. 10, 36 (2016). T.F. Lin, S.E. Busch, C. Hansel, Intrinsic and synaptic determinants of receptive field plasticity in Purkinje cells of the mouse cerebellum. Nat. Commun. 15, 4645 (2024). J. P. Welsh, Functional significance of climbing-fiber synchrony. A population coding and behavioral analysis. Ann. N. Y. Acad. Sci. 978, 188–204 (2002). S. Kitazawa, D. M. Wolpert, Rhythmicity, randomness and synchrony in climbing fiber signals. Trends Neurosci. 28, 611–619 (2005). S. R. Schultz, K. Kitamura, A. Post-Uiterweer, J. Krupic, M. Häusser, Spatial pattern coding of sensory information by climbing fiber-evoked calcium signals in networks of neighboring cerebellar Purkinje cells. J. Neurosci. 29, 8005–8015 (2009). V. Gauck, D. Jaeger, The control of rate and timing of spikes in the deep cerebellar nuclei by inhibition. J. Neurosci. 20, 3006–3016 (2000). A. L. Person, I. M. Raman, Purkinje neuron synchrony elicits time-locked spiking in the cerebellar nuclei. Nature 481, 502–506 (2012). T. Tang, T.A. Blenkinsop, E.J. Lang, Complex spike synchrony dependent modulation of rat deep cerebellar nuclear activity. eLife 8, e40101 (2019). 10.7554/eLife.40101 D. Feldmeyer, M. Brecht, F. Helmchen, C.C.H. Petersen, J. F. Poulet, J. F. Staiger, H. J. Luhmann, C. Schwarz, Barrel cortex formation. Prog. Neurobiol. 103, 3–27 (2013). Y. Ugawa, B.L. Day, J.C. Rothwell, P.D. Thompson, P.A. Merton, C.D. Marsden. Modulation of motor cortical excitability by electrical stimulation over the cerebellum in man. J. Physiology. 441, 57–72 (1991). Y. Ugawa, Y. Uesaka, Y. Terao, R.H. Hanajima, I. Kanazawa. Magnetic stimulation over the cerebellum in humans. Ann. Neurology 37, 691–821 (1995). J. Ge et al., Ventral zona incerta parvalbumin neurons modulate sensory-induced and stress-induced self-grooming via input-dependent mechanisms in mice. iScience 27, 110165 (2024). J. M. Kebschull et al. Cerebellar nuclei evolved by repeatedly duplicating a conserved cell-type set. Science 370, eabd5059 (2020). J. Beckinghausen et al. The cerebellum contributes to generalized seizures by altering activity in the ventral posteromedial nucleus. Comm. Biol. 6, 731 (2023). T. Mao, D. Kusefoglu, B.M. Hooks, D. Huber, L. Petreanu, K. Svoboda. Long-range neuronal circuits underlying the interaction between sensory and motor cortex. Neuron 72, 111–132 (2011). S. Lee, I. Krugilov, Z.J. Huang, G. Fishell, B. Rudy. A disinhibitory circuit mediates motor integration in the somatosensory cortex. Nat. Neurosci. 11, 1662–1670 (2013). E. I. Knudsen, Supervised learning in the brain. J. Neurosci. 14, 3985–3997 (1994). C. Viollet, A. Simon, V. Tolle, A. Labarthe, D. Grouselle, Y. Loe-Mie, M. Simonneau, G. Martel, J. Epelbaum. Somatostatin-IRES-Cre Mice: Between Knockout and Wild-Type? Front Endocrinol 8 (2017). A. H. Cheng, S. W. Fung, H. M. Cheng. Limitations of the Avp-IRES2-Cre (JAX #023530) and Vip-IRES-Cre (JAX #010908) Models for Chronobiological Investigations. J Biol Rhythms 34, 634–644 (2019). K. P. Koster, S. M. Sherman. Convergence of inputs from the basal ganglia with layer 5 of motor cortex and cerebellum in mouse motor thalamus. eLife 13, e97489 (2024). 10.7554/eLife.97489 G. Paxinos, K. B. J. Franklin. Paxinos and Franklin’s The Mouse Brain in Stereotaxic Coordinates (Elsevier, ed. 5, 2019). T. Miyazaki, M. Watanabe. Development of an anatomical technique for visualizing the mode of climbing fiber innervation in Purkinje cells and its application to mutant mice lacking GluRδ2 and Ca(v)2.1. Anat Sci Int 86 (2011). Y. Nagai, et al. Deschloroclozapine, a potent and selective chemogenetic actuator enables rapid neuronal and behavioral modulations in mice and monkeys. Nature neuroscience 23 (2020). B. Vasquez, B. Campos, A. Cao, A. T. Theint, W. Zeiger. High-Sensitivity Intrinsic Optical Signal Imaging Through Flexible, Low-Cost Adaptations of an Upright Microscope. eNeuro 10 (2023). A. Ayaz, A. Stäuble, M. Hamada, M. A. Wulf, A. B. Saleem, F. Helmchen. Layer-specific integration of locomotion and sensory information in mouse barrel cortex. Nature Commun 10, 2585 (2019). Abby Silbaugh (2025) Climbing fiber. GitHub. https://github.com/abbysilbaugh/climbingfiber 10.7554/eLife.109183.2.sa4 eLife Assessment Ito-Ishida Aya Reviewing Editor https://ror.org/04j1n1c04 RIKEN Center for Brain Science Saitama Japan Convincing Fundamental

This study presents a fundamental discovery of how cerebellar climbing fibers modulate plastic changes in the somatosensory cortex by identifying both the responsible cortical circuit and the anatomical pathways. The evidence supporting the conclusions is convincing and well supported by modern neuroscience methodologies. Overall, this work represents a significant contribution that will be of broad interest to neuroscientists, especially those studying the long-distance cerebellar influence on non-motor brain functions.

 10.7554/eLife.109183.2.sa3 Reviewer #1 (Public review): Reviewer

Summary:

Silbaugh, Koster and Hansel investigated how the cerebellar climbing fiber (CF) signals influence neuronal activity and plasticity in mouse primary somatosensory (S1) cortex. They found that optogenetic activation of CFs in the cerebellum modulates responses of cortical neurons to whisker stimulation in a cell-type-specific manner and suppresses potentiation of layer 2/3 pyramidal neurons induced by repeated whisker stimulation. This suppression of plasticity by CF activation is mediated through modulation of VIP- and SST-positive interneurons. Using transsynaptic tracing and chemogenetic approaches, the authors identified a pathway from the cerebellum through the zona incerta and the thalamic posterior medial (POm) nucleus to the S1 cortex, which underlies this functional modulation.

The authors have addressed all the necessary points.

 10.7554/eLife.109183.2.sa2 Reviewer #2 (Public review): Reviewer

Summary:

The authors examined long-distance influence of climbing fiber (CF) signaling in the somatosensory cortex by manipulating whiskers through stimulation. Also, they examined CF signaling using two-photon imaging and mapped projections from the cerebellum to somatosensory cortex using transsynaptic tracing. As a final manipulation, they used chemogenetics to perturb parvalbumin positive neurons in the zona incerta and recorded from climbing fibers.

Strengths:

There are several strengths to this paper. The recordings were carefully performed and AAVs used were selective and specific for the cell-types and pathways being analyzed. In addition, the authors used multiple approaches that support climbing fiber pathways to distal regions of the brain. This work will impact the field and describes nice methods to target difficult to reach brain regions, such as the inferior olive.

No weaknesses noted.

 10.7554/eLife.109183.2.sa1 Reviewer #3 (Public review): Reviewer

Summary:

The authors developed an interesting novel paradigm to probe the effects of cerebellar climbing fiber activation on short-term adaptation of somatosensory neocortical activity during repetitive whisker stimulation. Normally, RWS potentiated whisker responses in pyramidal cells and weakly suppressed them in interneruons, lasting for at least 1h. Crusii Optogenetic climbing fiber activation during RWS reduced or inverted these adaptive changes. This effect was generally mimicked or blocked with chemogenetic SST or VIP activation/suppression as predicted based on their "sign" in the circuit.

Strengths:

The central finding about CF modulation of S1 response adaptation is interesting, important, and convincing, and provides a jumping-off point for the field to start to think carefully about cerebellar modulation of neocortical plasticity.

Weaknesses:

The SST and VIP results appeared slightly weaker statistically, but I do not personally think this detracts from the importance of the initial finding (if there are multiple underlying mechanisms, modulating one may reproduce only a fraction of the effect size). I found the suggestion that zona incerta may be responsible for the cerebellar effects on S1 to be a more speculative result (it is not so easy with existing technology to effectively modulate this type of polysynaptic pathway), but this may be an interesting topic for the authors to follow up on in more detail in the future.

Comments on revisions:

The authors have appropriately addressed my comments.

 10.7554/eLife.109183.2.sa0 Author response: Silbaugh Abby Author Koster Kevin P Author Hansel Christian Author

The following is the authors’ response to the original reviews

Public Reviews:

Reviewer #1 (Public review):

Summary:

Silbaugh, Koster, and Hansel investigated how the cerebellar climbing fiber (CF) signals influence neuronal activity and plasticity in mouse primary somatosensory (S1) cortex. They found that optogenetic activation of CFs in the cerebellum modulates responses of cortical neurons to whisker stimulation in a cell-type-specific manner and suppresses potentiation of layer 2/3 pyramidal neurons induced by repeated whisker stimulation. This suppression of plasticity by CF activation is mediated through modulation of VIP- and SST-positive interneurons. Using transsynaptic tracing and chemogenetic approaches, the authors identified a pathway from the cerebellum through the zona incerta and the thalamic posterior medial (POm) nucleus to the S1 cortex, which underlies this functional modulation.

Strengths:

This study employed a combination of modern neuroscientific techniques, including two-photon imaging, opto- and chemo-genetic approaches, and transsynaptic tracing. The experiments were thoroughly conducted, and the results were clearly and systematically described. The interplay between the cerebellum and other brain regions - and its functional implications - is one of the major topics in this field. This study provides solid evidence for an instructive role of the cerebellum in experience-dependent plasticity in the S1 cortex.

Weaknesses:

There may be some methodological limitations, and the physiological relevance of the CFinduced plasticity modulation in the S1 cortex remains unclear. In particular, it has not been elucidated how CF activity influences the firing patterns of downstream neurons along the pathway to the S1 cortex during stimulation.

Our study addresses the important question of whether CF signaling can influence the activity and plasticity of neurons outside the olivocerebellar system, and further identifies the mechanism through which this indeed occurs. We provide a detailed description of the involvement of specific neuron subtypes and how they are modulated by climbing fiber activation to impact S1 plasticity. We also identify at least one critical pathway from the cerebellar output to the S1 circuit. It is indeed correct that we did not investigate how the specific firing patterns of all of these downstream neurons are affected, or the natural behaviors in which this mechanism is involved. Now that it is established that CF signaling can impact activity and plasticity outside the olivocerebellar system -- and even in the primary somatosensory cortex -- these questions will be important to further investigate in future studies.

(1) Optogenetic stimulation may have activated a large population of CFs synchronously, potentially leading to strong suppression followed by massive activation in numerous cerebellar nuclear (CN) neurons. Given that there is no quantitative estimation of the stimulated area or number of activated CFs, observed effects are difficult to interpret directly. The authors should at least provide the basic stimulation parameters (coordinates of stim location, power density, spot size, estimated number of Purkinje cells included, etc.).

As discussed in the paper, we indeed expect that synchronous CF activation is needed to allow for an effect on S1 circuits under natural or optogenetic activation conditions. The basic optogenetic stimulation parameters (also stated in the methods) are as follows: 470 nm LED; Ø200 µm core, 0.39 NA rotary joint patch cable; absolute power output of 2.5 mW; spot size at the surface of the cortex 0.6 mm; estimated power density 8 mW/mm2. A serious estimate of the number of Purkinje cells that are activated is difficult to provide, in particular as ‘activation’ would refer to climbing fiber inputs, not Purkinje cells directly.

(2) There are CF collaterals directly innervating CN (PMID:10982464). Therefore, antidromic spikes induced by optogenetic stimulation may directly activate CN neurons. On the other hand, a previous study reported that CN neurons exhibit only weak responses to CF collateral inputs (PMID: 27047344). The authors should discuss these possibilities and the potential influence of CF collaterals on the interpretation of the results.

A direct activation of CN neurons by antidromic spikes in CF collaterals cannot be ruled out. However, we believe that this effect will not be substantial. The activation of the multi-synaptic pathway that we describe in this study is more likely to require a strong nudge as resulting from synchronized Purkinje cell input and subsequent rebound activation in CN neurons (PMID: 22198670), rather than small-amplitude input provided by CF collaterals (PMID: 27047344). A requirement for CF/PC synchronization would also set a threshold for activation of this suppressive pathway.

(3) The rationale behind the plasticity induction protocol for RWS+CF (50 ms light pulses at 1 Hz during 5 min of RWS, with a 45 ms delay relative to the onset of whisker stimulation) is unclear.

a) The authors state that 1 Hz was chosen to match the spontaneous CF firing rate (line 107); however, they also introduced a delay to mimic the CF response to whisker stimulation (line 108). This is confusing, and requires further clarification, specifically, whether the protocol was designed to reproduce spontaneous or sensory-evoked CF activity.

This protocol was designed to mimic sensory-evoked CF activity as reported in Bosman et al (J. Physiol. 588, 2010; PMID: 20724365).

b) Was the timing of delivering light pulses constant or random? Given the stochastic nature of CF firing, randomly timed light pulses with an average rate of 1Hz would be more physiologically relevant. At the very least, the authors should provide a clear explanation of how the stimulation timing was implemented.

Light pulses were delivered at a constant 1 Hz. Our goal was to isolate synchrony as the variable distinguishing sensory-evoked from spontaneous CF activity; additionally varying stochasticity, rate, or amplitude would have confounded this. Future studies could explore how these additional parameters shape S1 responses.

(4) CF activation modulates inhibitory interneurons in the S1 cortex (Figure 2): responses of interneurons in S1 to whisker stimulation were enhanced upon CF coactivation (Figure 2C), and these neurons were predominantly SST- and PV-positive interneurons (Figure 2H, I). In contrast, VIP-positive neurons were suppressed only in the late time window of 650-850 ms (Figure 2G). If the authors' hypothesis-that the activity of VIP neurons regulates SST- and PVneuron activity during RWS+CF-is correct, then the activity of SST- and PV-neurons should also be increased during this late time window. The authors should clarify whether such temporal dynamics were observed or could be inferred from their data.

Yes, we see a significant activity increase in PV neurons in this late time window (see updates to Data S2). Activity was also increased in SST neurons, though this did not reach statistical significance (Data S2). One reason might be that – given the small effect size overall – such an effect would only be seen in paired recordings. Chemogenetic activity modulation in VIP neurons, which provides a more crude test, shows, however, that SST- and PV-positive interneurons are indeed regulated via inhibition from VIP-positive interneurons (Fig. 5).

(5) Transsynaptic tracing from CN nicely identified zona incerta (ZI) neurons and their axon terminals in both POm and S1 (Figure 6 and Figure S7).

a) Which part of the CN (medial, interposed, or lateral) is involved in this pathway is unclear.

We used a dual-injection transsynaptic tracing approach to specifically label the outputs of ZI neurons that receive input from the deep cerebellar nuclei. The anterograde viral vector injected into the CN is unlabeled (no fluorophore) and therefore, it is not possible to reliably assess the extent of viral spread in those experiments as performed. However, we have previously performed similar injections into the deep cerebellar nuclei and post hoc histology suggest all three nuclei will have at least some viral expression (Koster and Sherman, 2024). Due to size and injection location, we will mostly have reached the lateral (dentate) nuclei, but cannot exclude partial transsynaptic tracing from the interposed and medial nuclei.

b) Were the electrophysiological properties of these ZI neurons consistent with those of PV neurons?

Although most recorded cells demonstrated electrophysiological properties consistent with PV+ interneurons in other brain regions (i.e. fast spiking, narrow spike width, non-adapting; see Tremblay et al., 2016), interneuron subtypes in the ZI have been incompletely characterized, with SST+ cells showing similar features to those typically associated with PV+ cells (if interested, compare Fig. 4 in DOI: 10.1126/sciadv.abf6709 vs. Fig. S10 in https://doi.org/10.1016/j.neuron.2020.04.027). Therefore, we did not attempt to delineate cell identity based on these characteristics.

c) There appears to be a considerable number of axons of these ZI neurons projecting to the S1 cortex (Figure S7C). Would it be possible to estimate the relative density of axons projecting to the POm versus those projecting to S1? In addition, the authors should discuss the potential functional role of this direct pathway from the ZI to the S1 cortex.

An absolute quantification is difficult to provide based on the images that we obtained. However, any crude estimate would indicate the relative density of projections to POm is higher than the density of projections to S1 (this is apparent from the images themselves). While the anatomical and functional connections from POm to S1 have been described in detail (Audette et al., 2018), this is not the case for the direct projections to ZI. A direct ZI to S1 projection would potentially involve a different recruitment of neurons in the S1 circuit. Any discussion on the specific consequences of the activation of this direct pathway would be purely speculative.

Reviewer #2 (Public review):

Summary:

The authors examined long-distance influence of climbing fiber (CF) signaling in the somatosensory cortex by manipulating whiskers through stimulation. Also, they examined CF signaling using two-photon imaging and mapped projections from the cerebellum to the somatosensory cortex using transsynaptic tracing. As a final manipulation, they used chemogenetics to perturb parvalbumin-positive neurons in the zona incerta and recorded from climbing fibers.

Strengths:

There are several strengths to this paper. The recordings were carefully performed, and AAVs used were selective and specific for the cell types and pathways being analyzed. In addition, the authors used multiple approaches that support climbing fiber pathways to distal regions of the brain. This work will impact the field and describes nice methods to target difficult-to-reach brain regions, such as the inferior olive.

Weaknesses:

There are some details in the methods that could be explained further. The discussion was very short and could connect the findings in a broader way.

In the revised manuscript, we provide more methodological details, as requested. We provided as simple as possible explanations in the discussion, so as not to bias further investigations into this novel phenomenon. In particular, we avoid an extended discussion of the gating effect of CF activity on S1 plasticity. While this is the effect on plasticity specifically observed here, we believe that the consequences of CF signaling on S1 activity may entirely depend on the contexts in which CF signals are naturally recruited, the ongoing activity of other brain regions, and behavioral state. Our key finding is that such modulation of neocortical plasticity can occur. How CF signaling controls plasticity of the neocortex in all contexts remains unknown, but needs to be thoughtfully tested in the future.

Reviewer #3 (Public review):

Summary:

The authors developed an interesting novel paradigm to probe the effects of cerebellar climbing fiber activation on short-term adaptation of somatosensory neocortical activity during repetitive whisker stimulation. Normally, RWS potentiated whisker responses in pyramidal cells and weakly suppressed them in interneurons, lasting for at least 1h. Crusii Optogenetic climbing fiber activation during RWS reduced or inverted these adaptive changes. This effect was generally mimicked or blocked with chemogenetic SST or VIP activation/suppression as predicted based on their "sign" in the circuit.

Strengths:

The central finding about CF modulation of S1 response adaptation is interesting, important, and convincing, and provides a jumping-off point for the field to start to think carefully about cerebellar modulation of neocortical plasticity.

Weaknesses:

The SST and VIP results appeared slightly weaker statistically, but I do not personally think this detracts from the importance of the initial finding (if there are multiple underlying mechanisms, modulating one may reproduce only a fraction of the effect size). I found the suggestion that zona incerta may be responsible for the cerebellar effects on S1 to be a more speculative result (it is not so easy with existing technology to effectively modulate this type of polysynaptic pathway), but this may be an interesting topic for the authors to follow up on in more detail in the future.

Our interpretation of the anatomical and physiological findings is that a pathway via the ZI is indeed critical for the observed effects. This pathway also represents perhaps the most direct pathway (i.e. least number of synapses connecting the cerebellar nuclei to S1). However, several other direct and indirect pathways are plausible as well and we expect distinct activation requirements and consequences for neurons in the S1 circuit. These are indeed interesting topics for future investigation.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

(1) Line 77: "CF transients" is not a standard or widely recognized term. Please use a more precise expression, such as "CF-induced calcium transients."

We now avoid the use of the term “CF transients” and replaced it with “CF-induced calcium transients.”

(2) Titer of AAVs injected should be provided.

AAV titers have been included in an additional data table (Data S9).

(3) Several citations to the figures are incorrect (for example, "Supplementary Data 2a (Line 398)" does not exist).

We apologize for the mistakes in this version of the article. Incorrect citations to the figures have been corrected.

(4) Line 627-628: "The tip of the patch cable was centered over Crus II in all optogenetic stimulation experiments." The stereotaxic coordinate of the tip position should be provided.

The stereotaxic coordinate of the tip position has been provided in the methods.

(5) Line 629: "Blue light pulses were delivered with a 470 nm Fiber-Coupled LED (Thorlabs catalog: M470F3)." The size of the light stim and estimated power density (W/mm^2) at the surface of the cortex should be provided.

The spot size and estimated power density at the surface of the cortex has been provided in the methods.

(6) Line 702-706: References for DCZ should be cited.

We now cited Nagai et al, Nat. Neurosci. 23 (2020) as the original reference.

(7) Two-photon image processing (Line 807-809): The rationale for normalizing ∆F/F traces to a pre-stimulus baseline is unclear because ∆F/F is, by definition, already normalized to baseline fluorescence: (Ft-F0)/F0. The authors should clarify why this additional normalization step was necessary and how it affected the interpretation of the data.

A single baseline fluorescence value (F₀) was computed for each neuron across the entire recording session, which lasted ~120-minutes. However, some S1 neurons exhibit fluctuations in baseline fluorescence over time—often related to locomotive activity or spontaneous network oscillations—which can obscure stimulus-evoked changes. To isolate fluorescence changes specifically attributable to whisker stimulation, we normalized each ∆F/F trace to the prestimulus baseline for that trial. This additional normalization allowed us to quantify potentiation or depression of sensory responses themselves, independently of spontaneous oscillations or locomotion-related changes in the ongoing neural activity.

Reviewer #2 (Recommendations for the authors):

(1) Did the climbing fiber stimulation for Figure 1 result in any changes to motor activity? Can you make any additional comments on other behaviors that were observed during these manipulations?

Acute CF stimulation did not cause any changes in locomotive or whisking activity. The CF stimulation also did not influence the overall level of locomotion or whisking during plasticity induction.

(2) Figure 3B and F- it is very difficult to see the SST+ neurons. Can this be enhanced?

We linearly adjusted the brightness and contrast for the bottom images in Figure 3B and F to improve visualization of SST+ neurons. Note the expression of both hM3D(Gq) and hM4D(Gi) in SST+ neurons is sparse, which was necessary to avoid off-target effects.

(3) Can you be more specific about the subregions of cerebellar nuclei and cell types that are targeted in the tracing studies? Discussions of the cerebellar nuclei subregions are missing and would be interesting, as others have shown discrete pathways between cerebellar nuclei subregions and long-distance projections.

See our response to comment 5a from Reviewer 1 (copied again here): we used a dual-injection transsynaptic tracing approach to specifically label the outputs of ZI neurons that receive input from the deep cerebellar nuclei. The anterograde viral vector injected into the CN is unlabeled (no fluorophone) and therefore, it is not possible to reliably assess the extent of viral spread in those experiments as performed. However, we have previously performed similar injections into the deep cerebellar nuclei and post hoc histology suggest all three nuclei will have at least some viral expression (Koster and Sherman, 2024). Due to size and injection location, we will mostly have reached the lateral (dentate) nuclei, but cannot exclude partial transsynaptic tracing from the interposed and medial nuclei.

It would indeed be interesting to further investigate the effect of CFs residing in different cerebellar lobules, which preferentially target different cerebellar nuclei, on targets of these nuclei.

(4) Did you see any connection to the ventral tegmental area? Can you comment on whether dopamine pathways are influenced by CF and in your manipulations?

We did not specifically look at these pathways and thus are not able to comment on this.

(5) These are intensive surgeries, do you think glia could have influenced any results?

This was not tested and seems unlikely, but we cannot exclude such possibility.

(6) It is unclear in the methods how long animals were recorded for in each experiment. Can you add more detail?

Additional detail was added to the methods. Recordings for all experimental configurations did not last more than 120 minutes in total. All data were analyzed across identical time windows for each experiment.

(7) In the methods it was mentioned that recording length can differ between animals. Can this influence the results, and if so, how was that controlled for?

There was a variance in recording length within experimental groups, but no systematic difference between groups.

(8) I do not see any mention of animal sex throughout this manuscript. If animals were mixed groups, were sex differences considered? Would it be expected that CF activity would be different in male and female mice?

As mentioned in the Methods (Animals), mice of either sex were used. No sex-dependent differences were observed.

(9) Transsynaptic tracing results of the zona incerta are very interesting. The zona incerta is highly understudied, but has been linked to feeding, locomotion, arousal, and novelty seeking. Do you think this pathway would explain some of the behavioral results found through other studies of cerebellar lobule perturbations? Some discussion of how this brain region would be important as a cerebellar connection in animal behavior would be interesting.

Since the multi-synaptic pathway from the cerebellum to S1 involves several brain regions with their own inputs and modulatory influences, it seems plausible to assume that behaviors controlled by these regions or affecting signaling pathways that regulate them would show some level of interaction. Our study does not address these interactions, but this will be an interesting question to be addressed in future work.

Reviewer #3 (Recommendations for the authors):

General comments on the data presentation:

I'm not a huge fan of taking areas under curves ('AUC' throughout the study) when the integral of the quantity has no physical meaning - 'normalizing' the AUC (1I,L etc) is even stranger, because of course if you instead normalize the AUC by the # of data points, you literally just get the mean (which is probably what should be used instead).

Indeed, AUC is equal to the average response in the time window used, multiplied by the window duration (thus, AUC is directly proportional to the mean). We choose to report AUC, a descriptive statistic, rather than the mean within this window. In 1I and L, we normalize the AUC across animals, essentially removing the variability across animals in the ‘Pre’ condition for visualization. Note the significance of these comparisons are consistent whether or not we normalize to the ‘Pre’ condition (non-normalized RWS data in I shows a significant increase in PN activity, p = 0.0068, signrank test; non-normalized RWS+CF data in I shows a significant decrease in PN activity, p = 0.0135, paired t-test; non-normalized RWS data in L shows a significant decrease in IN activity, p <0.001, paired t-test; non-normalized RWS+CF data in L shows no significant change in IN activity, p = 0.7789, paired t-test).

I think unadorned bar charts are generally excluded from most journals now. Consider replacing these with something that shows the raw datapoints if not too many, or the distribution across points.

We have replaced bar charts with box plots and violin plots. We have avoided plotting individual data points due to the quantity of points.

In various places, the statistics produce various questionable outcomes that will draw unwanted reader scrutiny. Many of the examples below involve tiny differences in means with overlapping error bars that are "significant" or a few cases of nonoverlapping error bars that are "not significant." I think replacing the bar charts may help to resolve things here if we can see the whole distribution or the raw data points. As importantly, I think a big problem is that the statistical tests all seem to be nonparametric (they are ambiguously described in Table S3 as "Wilcoxon," which should be clarified, since there is an unpaired Wilcoxon test [rank sum] and a paired Wilcoxon test [sign rank]), and thus based on differences in the *median* whereas the bar charts are based on the *mean* (and SEM rather than MAD or IQR or other medianappropriate measure of spread). This should be fixed (either change the test or change the plots), which will hopefully allay many of the items below.

We thank the reviewer for this important point. As mentioned in the Statistics and quantification section, Wilcoxon signed rank tests were used for non-normal data. We have replaced the bar charts with box plots which show the IQR and median, which indeed allays may of the items below.

Here are some specific points on the statistics presentation:

(1) 1G, the test says that following RWS+CF, the decrease in PN response is not significant. In 1I, the same data, but now over time, shows a highly significant decrease. This probably means that either the first test should be reconsidered (was this a paired comparison, which would "build in" the normalization subsequently used automatically?) or the second test should be reconsidered. It's especially strange because the n value in G, if based on cells, would seem to be ~50-times higher than that in I if based on mice.

In Figure 1G, the analysis tests whether individual pyramidal neurons significantly changed their responses before vs. after RWS+CF stimulation. This is a paired comparison at the single-cell level, and here indicates that the average per-neuron response did not reliably decrease after RWS+CF when comparing each cell’s pre- and post-values directly. In contrast, Figure 1I examines the same dataset analyzed across time bins using a two-way ANOVA, which tests for effects of time, group (RWS vs. RWS+CF), and their interaction. The analysis showed a significant group effect (p < 0.001), indicating that the overall level of activity across all time points differed between RWS and RWS+CF conditions. The difference in significance between these two analyses arises because the first test (Fig. 1G) assesses within-neuron changes (paired), whereas the second test (Fig. 1I) assesses overall population-level differences between groups over time (independent groups). Thus, the tests address related but distinct questions—one about per-cell response changes, the other about how activity differs across experimental conditions.

(2) 1J RWS+CF then shows a much smaller difference with overlapping error bars than the ns difference with nonoverlapping errors in 1G, but J gets three asterisks (same n-values).

Bar graphs have been replaced with box plots.

(3) 1K, it is very unclear what is under the asterisk could possibly be significant here, since the black and white dots overlap and trade places multiple times.

See response to point 1. A significant group effect will exist if the aggregate difference across all time bins exceeds within-group variability. The asterisk therefore reflects a statistically significant main group effect (RWS versus RWS+CF) rather than differences at any single time point. Note, however, the very small effect size here.

(4) 2B, 2G, 2H, 2I, 3G, 3H, 5C etc, again, significance with overlapping error bars, see suggestions above.

Bar graphs have been replaced with box plots.

(5) Time windows: e.g., L149-153 / 2B - this section reads weirdly. I think it would be less offputting to show a time-varying significance, if you want to make this point (there are various approaches to this floating around), or a decay rate, or something else.

Here, we wanted to understand the overall direction of influence of CFs on VIP activity. We find that CFs exert a suppressive effect on VIP activity, which is statistically significant in this later time window. The specific effect of CF modulation on the activity of S1 neurons across multiple time points will be described in more detail in future investigations.

(6) 4G, 6I, these asterisks again seem impossible (as currently presented).

Bar graphs have been replaced with box plots.

The writing is in generally ok shape, but needs tightening/clarifying:

(1) L45 "mechanistic capacity" not clear.

We have simplified this term to “capacity.” We use the term here to express that the central question we pose is whether CF signals are able to impact S1 circuits. We demonstrate CF signals indeed influence S1 circuits and further describe the mechanism through which this occurs, but we do not yet know all of the natural conditions in which this may occur. We feel that “capacity” describes the question we pose -- and our findings -- very well.

(2) L48-58 there's a lot of material here, not clear how much is essential to the present study.

We would like to give an overview of the literature on instructive CF signaling within the cerebellum. Here, we feel it is important to describe how CFs supervise learning in the cerebellum via coincident activation of parallel fiber inputs and CF inputs. Our results demonstrate CFs have the capacity to supervise learning in the neocortex in a similar manner, as coincident CF activation with sensory input modulates plasticity of S1 neurons.

(3) L59 "has the capacity to" maybe just "can".

This has been adopted. We agree that “can” is a more straightforward way of saying “has the capacity to” here. In this sentence, “can” and “has the capacity to” both mean a general ability to do something, without explicit knowledge about the conditions of use.

(4) L61-62 some of this is circular "observation that CF regulates plasticity in S1..has consequences for plasticity in S1".

We now changed this to read “…consequences for input processing in S1.”

(5) L91 "already existing whisker input" although I get it, strictly speaking, not clear what this means.

This sentence has been reworded for clarity.

(6) L94 "this form of plasticity" what form?

Edited to read “sensory-evoked plasticity.”

(7) L119 should say "to test the".

This has been corrected.

(8) L120 should say "well-suited to measure receptive fields".

We agree; this wording has been adopted.

(9) L130 should say "optical imaging demonstrated that receptive field".

This has been adopted.

(10) L138, the disclaimer is helpful, but wouldn't it be less confusing to just pick a different set of terms? Response potentiation etc.

Perhaps, but we want to stress that components of LTP and LTD (traditionally tested using electrophysiological methods to specifically measure synaptic gain changes) can be optically measured as long as it is specified what is recorded.

(11) L140, this whole section is not very clear. What was the experiment? What was done and how?

The text in this section has been updated.

(12) L154, 156, 158, 160, 960, what is a "basic response"? Is this supposed to contrast with RWS? If so, I would just say "we measured the response to whisker stimulation without first performing RWS, and compared this to the whisker stimulation with simultaneous CF activation."

What we meant by “basic response” was the acute response of S1 neurons to a single 100 ms air puff. Here, we indeed measured the acute responses of S1 neurons to whisker stimulation (100 ms air puff) and compared them to whisker stimulation with simultaneous CF activation (100 ms air puff with a 50 ms light pulse; the light pulse was delayed 45 ms with respect to the air puff). This paragraph has been reworded for clarity.

(13) L156 "comprised of a majority" unclear. You mean most of the nonspecific IN group is either PV or SST?

Yes, that was meant here. This paragraph has been reworded for clarity.

(14) L165 tense. "are activated" "we tested" prob should be "were activated."

This sentence was reworded.

(15) L173 Not requesting additional experiments, but demonstrating that the effect is mimicked by directly activating SST or suppressing VIP questions the specificity of CF activation per se, versus presumably many other pathways upstream of the same mechanisms, which might be worth acknowledging in the text.

We indeed observe that directly activating SST or suppressing VIP neurons in S1 is sufficient to mediate the effect of CF activation on S1 pyramidal neurons, implicating SST and VIP neurons as the local effectors of CF signaling. In the text, we wrote “...the notion of sufficiency does not exclude potential effects of plasticity processes elsewhere that might well modulate effector activation in this context and others not yet tested.” Here, we mean that CFs are certainly not the only modulators of the inhibitory network in S1. One example we highlight in the discussion is that projections from M1 are known to modulate this disinhibitory VIP-to-SST-to-PN microcircuit in S1. We conclude from our chemogenetic manipulation experiments that CFs ultimately have the capacity to modulate S1 interneurons, which must occur indirectly (either through the thalamus or “upstream” regions as this reviewer points out). The fact that many other brain regions may also modulate the interneuron network in S1 -- or be modulated by CF activity themselves -- only expands the capacity of CFs to exert a variety of effects on S1 neurons in different contexts.

(16) L247 "induced ChR2" awkward.

We changed this to read “we expressed ChR2.”

(17) 6C, what are the three colors supposed to represent?

We apologize for the missing labels in this version of the manuscript. Figure 6C and the figure legend have been updated.