Climbing Fiber Input Shapes Reciprocity of Purkinje Cell Firing

Neuron

Article Climbing Fiber Input Shapes Reciprocity of Purkinje Cell Firing Aleksandra Badura,1,6 Martijn Schonewille,1,6 Kai Voges,1 Elisa Galliano,1 Nicolas Renier,3,4 Zhenyu Gao,1 Laurens Witter,2 Freek E. Hoebeek,1 Alain Che´dotal,3,4,5 and Chris I. De Zeeuw1,2,* 1Department

of Neuroscience, Erasmus MC, 3000 DR Rotterdam, The Netherlands Institute for Neuroscience, Royal Dutch Academy of Arts & Sciences (KNAW), 1105 BA Amsterdam, The Netherlands 3INSERM, U968, Paris F-75012, France 4UPMC Univ. Paris 06, UMR_S 968, Institut de la Vision, Paris F-75012, France 5CNRS, UMR_7210, Paris, F-75012, France 6These authors contributed equally to this work *Correspondence: [email protected] http://dx.doi.org/10.1016/j.neuron.2013.03.018 2Netherlands

SUMMARY

The cerebellum fine-tunes motor activity via its Purkinje cell output. Purkinje cells produce two different types of spikes, complex spikes and simple spikes, which often show reciprocal activity: a periodical increase in complex spikes is associated with a decrease in simple spikes, and vice versa. This reciprocal firing is thought to be essential for coordinated motor behavior, yet how it is accomplished is debated. Here, we show in Ptf1a::cre;Robo3lox/lox mice that selectively rerouting the climbing fibers from a contralateral to an ipsilateral projection reversed the complex-spike modulation during sensory stimulation. Strikingly, modulation of simple spikes, which is supposed to be controlled by mossy fibers, reversed as well. Climbing fibers enforce this reciprocity in part by influencing activity of inhibitory interneurons, because the phase of their activity was also converted. Ptf1a::cre;Robo3lox/lox mice showed severe ataxia highlighting that climbing fiber input and its impact on reciprocity of Purkinje cell firing play an important role in motor coordination. INTRODUCTION The cerebellum has been proposed to operate as a phase leadlag compensator with learning capabilities (De Zeeuw et al., 1995). Such operations, fine-tuning temporal processing, are relevant for many functions varying from motor coordination to affective control and cognition (Rizzolatti and Wolpert, 2005; Schmahmann, 2012; Timmann, 2012). Computations in the cerebellar cortex involve a variety of neurons including granule cells, Golgi cells and molecular layer interneurons, which ultimately relay the information they carry onto Purkinje cells. The Purkinje cells in turn influence both the excitatory and inhibitory neurons in the cerebellar and vestibular nuclei and thereby activity in the rest of the brain (Person and Raman, 2011). The excitatory loop controls directly activity in the brainstem and 700 Neuron 78, 700–713, May 22, 2013 ª2013 Elsevier Inc.

cerebral cortex and thereby motor behavior and cognition (Timmann, 2012), whereas the inhibitory loop controls activity in the inferior olive and thereby feedback control to the cerebellum (De Zeeuw et al., 2011). Unlike most other neurons in the brain (van Vreeswijk and Sompolinsky, 1996), Purkinje cells produce two different types of spikes: complex spikes and simple spikes (Medina and Lisberger, 2008; Welsh et al., 1995). The complex spikes reflect the activation of the climbing fibers, whereas the simple spikes can be triggered by the other main afferent input to the cerebellar cortex, the mossy fiber-parallel fiber pathway (Medina and Lisberger, 2008). The frequencies of these two types of spikes are often modulated in an antiphasic, i.e., reciprocal, manner: when complex-spike frequency increases, the simple-spike rate decreases and vice versa. Reciprocity is particularly evident during natural sensorimotor control (Graf et al., 1988; Yakhnitsa and Barmack, 2006). For example, during control of optokinetic responses (OKR) to sinusoidal stimulation, complex-spike and simple-spike activities in Purkinje cells of the vestibulocerebellum modulate out of phase with respect to each other (De Zeeuw et al., 1995; Simpson et al., 1996). It has been difficult to find out whether and to what extent the climbing fiber input controls this enigmatic reciprocity, because so far it has not been possible to manipulate the periodicity of the complex-spike activities without affecting their overall firing frequency, which induces changes in simple-spike firing by itself (Leonard and Simpson, 1986; Montarolo et al., 1982; Simpson et al., 1996). Here, we test the role of the climbing fiber system in reciprocity control by investigating a mouse mutant (Ptf1a::cre;Robo3lox/lox mice), in which climbing fibers derived from the inferior olive are not silenced but rerouted so that they project mainly to the ipsilateral, rather than the contralateral side (Renier et al., 2010). As described by Renier and colleagues (2010), the design of the Ptf1a::cre;Robo3lox/lox mice is based on the fact that the transmembrane receptor Robo3 is transiently expressed by commissural axons in the brainstem and hindbrain when they cross the ventral midline (Sabatier et al., 2004). In both mice and humans, commissural neurons deficient in Robo3 fail to cross the midline and extend their axons ipsilaterally (Jen et al., 2004). During normal development, inferior olivary neurons migrate tangentially from the rhombic lip toward the ventral floor

Neuron Climbing Fiber Dominance in Reciprocity

plate at which the cell bodies stop and only the axons, i.e., future climbing fibers, cross (Marillat et al., 2004). To generate mice lacking the interolivary commissure, we crossed Robo3lox/lox conditional knockout with mice in which Cre recombinase was knocked into the Ptf1a (Ptf1-p48) locus (Ptf1a::cre) (Kawaguchi et al., 2002), which encodes a bHLH transcription factor expressed by a majority of olivary neuron progenitors (for details of generation of Ptf1a::cre;Robo3lox/lox mice, see Renier et al., 2010). In these mice, Robo3 is absent in olivary axons and accordingly they are unable to cross the ventral midline, but still contact Purkinje cells located in the ipsilateral cerebellum (Renier et al., 2010; see also below). With this strategy (Figure 1A), we hypothesized, it should be possible to shift the modulation of complex-spike activities without affecting their firing frequency, because the modulation of the olivary neurons and that of their climbing fibers are still in place, yet in the opposite direction due to altered laterality. Thus, if any change will be observed in the modulation of the simple spikes in the Ptf1a::cre;Robo3lox/lox mice, it cannot be because of an overall removal of complex spikes, which are expected to occur at a relatively normal level. RESULTS Climbing Fiber Projections Are Selectively Rerouted in Ptf1a::cre;Robo3lox/lox Mice Unilateral injections with the tracers cholera toxin B-subunit (CTb) and/or gold-lectin conjugate of bovine serum albumin with wheat germ agglutinate and 10 nm gold sol (Gold-lectin) (Ruigrok and Apps, 2007) in the flocculus of the vestibulocerebellum in Ptf1a::cre;Robo3lox/lox mice (n = 9) revealed that on average 76% (±4%) of the labeled olivary neurons were located in the ipsilateral, dorsal cap of Kooy or ventrolateral outgrowth, i.e., the sources known to provide the visual climbing fiber input to the flocculus (Schonewille et al., 2006b; Figures 1B and 1C); the remaining labeled neurons (24% ± 5%) were located on the contralateral side residing in the same olivary subnuclei. In control mice (n = 6) retrogradely labeled neurons resided exclusively in the contralateral part of the inferior olive. Foxp2 immunostaining at P28 showed that the lamellation of the inferior olive was not altered in adult Ptf1a::cre;Robo3lox/lox mice (Figure 1D) and that the areas of their olivary subdivisions did not differ (p = 0.67; two-way ANOVA; n = 5 for both genotypes; Fujita and Sugihara, 2012). Moreover, there was no obvious mixing of olivary neurons among the subdivisions and their numbers were comparable (p = 0.86; two-way ANOVA); for quantification of cell numbers in dorsal cap and b subnucleus, the two main sources of climbing fibers to the vestibulocerebellum, see Figure 1D. However, the ventrolateral outgrowth was slightly elongated laterally with respect to control mice. Ipsilateral climbing fibers in Ptf1a::cre;Robo3lox/lox mice appropriately targeted Purkinje cells during their development and adulthood, and a subset, distributed in typical sagittal stripes, could be visualized with coimmunostaining for calcitonin gene-related peptide (CGRP) and calbindin (Morara et al., 2001; Peltier and Bishop, 1999; see Figure S1B available online). Quantification of mean width and position of CGRP+ climbing fiber zones in the nodulus of the vestibulocerebellum did not reveal any signif-

icant difference among Ptf1a::cre;Robo3lox/lox mice (n = 4) and controls (n = 3); for details see Table S1 and Renier et al. (2010). The vast majority of Purkinje cells in adult Ptf1a::cre; Robo3lox/lox mice showed a normal elimination of their surplus in climbing fiber connections (Figure S1A). Moreover, the expression patterns of heat-shock protein 25 (HSP25) (Schonewille et al., 2006b) and CGRP in the flocculus of adult Ptf1a::cre;Robo3lox/lox mice also remained unaffected (Figure S1B) confirming that the lack of the olivary commissure did not perturb the development and establishment of the olivocerebellar map. As expected from the lack of expression of Ptf1a in mossy fiber projection neurons (Yamada et al., 2007), the lateralization of the mossy-fiber projections to the flocculus was not affected in Ptf1a::cre;Robo3lox/lox mice (Figure 2). Unilateral injections with CTb and/or Gold-lectin in the granular layer of the flocculus of the Ptf1a::cre;Robo3lox/lox mice (n = 9) or controls (n = 6) resulted in similar proportions of cells retrogradely labeled in the ipsilateral and contralateral, medial vestibular nucleus, superior vestibular nucleus, nucleus prepositus hypoglossi and nucleus reticularis tegmenti pontis (Figures 2A and 2B). Likewise, immunocytochemistry for choline-acetyltransferase (ChAT) and vesicular transporters VGLUT1 and VGLUT2 in Ptf1a::cre;Robo3lox/lox mice and controls showed that the morphological characteristics and density of the termination of the mossy fibers in the flocculus were unaffected (Figure 2C). In addition, the compartimentation of mossy fibers in the anterior lobe as revealed by the patterned expression of vesicular glutamate transporters VGLUT1 and VGLUT2 was in line with previous studies (Gebre et al., 2012; Figure S2B). Moreover, unilateral injections of the anterograde tracer AF 594 Dextran into the medial vestibular nucleus of Ptf1a::cre;Robo3lox/lox mice resulted in both the ipsilateral and contralateral flocculus in labeling of mossy fiber terminals with morphological characteristics that were comparable to those in control mice (Figure 2D). Golgi-stained sections also indicated that the development of neurons in the cerebellar cortex and the termination pattern of Purkinje cells were normal. The cyto-architecture of the cerebellar cortex was normal in that the shape and spatial orientation of the dendrites and axons of Purkinje cells, stellate cells, Golgi cells, and mossy fiber rosettes did not differ between Ptf1a::cre;Robo3lox/lox and control mice (Figure S2A). Moreover, in sections of P10 Rig-1(Robo3)+/
::GFP knockin mice (Sabatier et al., 2004), calbindin immunolabeled Purkinje cells were neither positive for GFP nor for Robo3 immunostaining, indicating that Robo3 is not expressed by postnatal Purkinje cells and hence probably does not play a role in their synaptogenesis (Figure S2C). More generally, no cerebellar neurons were immunoreactive for Robo3 in the postnatal and adult cerebellum. Robo3 was transiently expressed at E13 by some commissural neurons in the cerebellar nuclei (Figure S2D; see also Tamada et al., 2008), but by E15 this signal was downregulated (after their axons had crossed the midline) and while a few cerebellar nuclei cells still expressed Robo3 mRNA the Robo3 immunolabeling was absent. Importantly, at E13, Ptf1a was not expressed by Robo3 immunoreactive cerebellar neurons as demonstrated by the lack of overlap between Robo3 and GFP in Ptf1a::cre;taumGFP mice and the persistence of Neuron 78, 700–713, May 22, 2013 ª2013 Elsevier Inc. 701

Neuron Climbing Fiber Dominance in Reciprocity

Figure 1. Rerouting of Olivary Axons in Ptf1a::cre;Robo3lox/lox Mice (A) Schematic illustration of the olivocerebellar circuitry in controls compared to Ptf1a::cre;Robo3lox/lox mice. Note that the climbing fibers (red thick line) in the mutants project predominantly to the Purkinje cells on the ipsilateral side, whereas those in control animals (blue line) project exclusively to the contralateral side. (B) Examples of injection sites of retrograde tracers (cholera toxin B-subunit and gold-lectin) in the flocculus of controls (left panel) and Ptf1a::cre;Robo3lox/lox mice (right panel). Control animals showed exclusively labeled neurons (arrow) in the contralateral inferior olive, whereas Ptf1a::cre;Robo3lox/lox mice predominantly showed labeled cell-bodies on the ipsilateral side. The retrogradely traced neurons appear more lateral than in controls (arrows), because the ventrolateral outgrowth (VLO) is somewhat flattened and extends more laterally (see also panel D). Note that labeled cells are not found in other olivary subdivisions. (legend continued on next page)

702 Neuron 78, 700–713, May 22, 2013 ª2013 Elsevier Inc.

Neuron Climbing Fiber Dominance in Reciprocity

Robo3 immunolabeling in the cerebellum of E13 Ptf1a::cre; Robo3lox/lox mice (Figure S2D). Although we cannot exclude secondary reconfigurations along the path of the olivary axons providing collaterals to the cerebellar and vestibular nuclei (De Zeeuw et al., 2011), these findings indicate that the climbing fiber projections in Ptf1a::cre;Robo3lox/lox mice are successfully selectively rerouted and innervate predominantly ipsilateral Purkinje cells with an otherwise normal topography, and that the lateralization of the mossy fibers is not affected and comparable to that reported in previous studies (Ruigrok, 2003; Voogd and Barmack, 2006). Floccular Purkinje Cells in Ptf1a::cre;Robo3lox/lox Mice Maintain Reciprocal Firing When we subjected head-fixed, awake Ptf1a::cre;Robo3lox/lox mice (n = 25) to sinusoidal vertical axis (VA) optokinetic stimulation and recorded extracellular single unit signals from Purkinje cells located in the VA zones in the flocculus (VA Purkinje cells) (Schonewille et al., 2006a, 2006b; Figure 3A), the complex spikes modulated robustly at all frequencies tested (0.1–0.8 Hz; Figure 3B) and their overall average firing frequency was not significantly different (p R 0.41 for all frequencies) from that in control animals (n = 20) (Figure S3). Whereas all VA Purkinje cells (n = 38) in controls showed an increased complex-spike frequency during visual stimulation toward the contralateral side (Schonewille et al., 2006b), the majority of the cells in Ptf1a::cre;Robo3lox/lox mice (48 out of 61 cells; i.e., 79%) modulated optimally in the opposite direction (p % 0.004 for all frequencies; one-way ANOVA; Table S2). Their average phase shift with respect to that of controls ranged from 103 to 168 (phase refers to the temporal shift in the periodic stimulation cycle set at 360 ; for details see Figure 3B). The complex-spike activities of a minority of the VA Purkinje cells (13 out of 61; 21%) in Ptf1a::cre;Robo3lox/lox mice responded in line with those of control mice; these cells presumably reflect the minority of cells that still receive a contralateral climbing fiber input, because the polarity of their modulation corresponded to that of an input from the contralateral inferior olive (dorsal cap) and the percentage closely resembled that of the amount of contralaterally projecting fibers in the tracing experiments described above (i.e., 25%) (Figures 1B and 4). Unlike the differences in phase values, the amplitude of complex-spike modulation in Ptf1a::cre;Robo3lox/lox mice was comparable to that in controls (Table S2). The main question is whether the phase of the simple-spike modulation in Ptf1a::cre;Robo3lox/lox mice remained normal, as one might expect from the fact that the lateralization of the mossy-fibers was unaffected. Alternatively, it could shift together with that of the complex-spike activities, as one may predict based on the notion that the climbing fibers rule many forms of plasticity in the cerebellar cortex in a synergistic fashion (Gao et al., 2012). Clearly, the simple-spike modu-

lation shifted together with that of their complex spikes at all frequencies demonstrating that the reciprocity stayed intact (Figures 3C and 4; Table S2); compared to the simple-spike modulation of controls the average phase shift in the mutants ranged from 131 to 153 (p < 0.0001 for all frequencies; oneway ANOVA). A minority of the Purkinje cells (13 out of 61) in the Ptf1a::cre;Robo3lox/lox mice showed a simple-spike phase that was comparable to that of controls. These simple-spike modulations were observed in exactly the same thirteen Purkinje cells that showed the control-like complex-spike modulation described above, which re-emphasizes the ubiquitous character of the reciprocity phenomenon in general (Figure 4A) and the dominant impact of the climbing fibers. One cell out of the 61 did not show reciprocity; in this cell the simple spikes were relatively in-phase (i.e., less than 90 , namely 84 ) with the complex spikes (see exceptional behavior of this particular cell in Figure 4A). The average peak-to-peak amplitude of simple-spike modulation in Ptf1a::cre;Robo3lox/lox mice was significantly higher than that in controls (p < 0.05 for all frequencies; one-way ANOVA); this difference may be related to a change in inhibition, because the average simple-spike firing frequency during the trough of the modulation in the mutants was consistently lower than that in controls (p < 0.03 for all frequencies; one-way ANOVA) (Figure 5A and Table S2). This potentially increased inhibition could also explain the increased irregularity (CV2, p < 0.05 for all frequencies; one-way ANOVA) (Figure S3D) and an increased phase locking coefficient r with respect to the stimulus (z value, p < 0.01; Rayleigh statistics) (Figure S4). In contrast, the average simple-spike firing frequency as well as the pause in simple-spike activity after complex spikes (i.e., climbing fiber pause) did not differ significantly between Ptf1a::cre;Robo3lox/lox and control mice (Figure S3). To find out whether the shift in Purkinje cell responses is not restricted to a particular category of cells (i.e., the VA cells), we also tested Purkinje cells that modulate optimally around the horizontal axis (HA cells) (Figure 4B). Again, the majority of the complex-spike and simple-spike activity in the Ptf1a::cre; Robo3lox/lox mice showed a phase opposite to that recorded in controls (p = 0.03; one-way ANOVA) (Figure S5 and Table S2), and all cells, including those that did and did not resemble the control cells, showed a clear reciprocity (Figures 4A and S5). In addition, the overall firing rate did not differ significantly (for simple spikes p = 0.10, for complex spikes p = 0.42; one-way ANOVA) and the simple-spike responses showed a higher amplitude of modulation (p = 0.006; oneway ANOVA) as well as an increase in the irregularity of spiking (CV2, p = 0.04; one-way ANOVA). Moreover, the average simple-spike firing frequency during the trough of the modulation in the Ptf1a::cre;Robo3lox/lox mice was also consistently lower than that in controls (p = 0.016; one-way ANOVA) (Figure 5A and Table S2).

(C) Quantitative representation of ipsi- and contralaterally stained olivary cells in the dorsal cap (DC) and VLO in controls (n = 6) and mutants (n = 9) following tracing experiments. (D) Coronal sections at the level of the inferior olive of P28 Robo3lox/lox and Ptf1A::cre;Robo3lox/lox mice immunolabeled with anti-Foxp2 antibody. Histograms show number of cells in the DC and b subnucleus and area measurements of inferior olive subnuclei for controls (n = 5, blue) and mutants (n = 5, red). Scale bars indicate 500 mm (B, injection sites), 100 mm (B, bottom left), 200 mm (B, bottom right) and 250 mm (D). Error bars denote SEM.

Neuron 78, 700–713, May 22, 2013 ª2013 Elsevier Inc. 703

Neuron Climbing Fiber Dominance in Reciprocity

A

Ptf1a::Cre;Robo3lox/lox IPSI CONTRA

Control IPSI

CONTRA

PRH

MVN

SVN

NRTP

C Control CONTRA Control IPSI Ptf1a::Cre;Robo3lox/lox CONTRA Ptf1a::Cre;Robo3lox/lox IPSI

80

VGLUT1

VGLUT2

Ptf1a::Cre;Robo3lox/lox ChAT VGLUT1

VGLUT2

MVN

60 40 20

0 100 Cell count %

PRH

D

SVN

NRTP

80 60 40

Ptf1a::Cre;Robo3lox/lox Control CHAT 12 10 8 6 4 2 0 100 VGLUT VGLUT 80 1 2 60

Density %

Cell count %

100

Control ChAT

Number of Rossetes

B

Control

Ptf1a::Cre;Robo3lox/lox VEN

40

20

20

0

0

Figure 2. Mossy Fiber Lateralization and Termination in Ptf1a::cre;Robo3lox/lox Mice Are Normal (A) Retrogradely labeled neurons in ipsilateral and contralateral nucleus prepositus hypoglossi (PRH), medial vestibular nucleus (MVN), superior vestibular nucleus (SVN), and nucleus reticularis tegmentis pontis (NRTP) following unilateral injection of cholera toxin B-subunit and/or gold-lectin in the flocculus of controls and Ptf1a::cre;Robo3lox/lox mice. (B) Quantification of the labeled cells on the ipsilateral and contralateral side of the brains of 6 controls and 9 Ptf1a::cre;Robo3lox/lox mice show that the lateralization of the mossy fiber projection is not affected; histograms depict averages. (C) Immunocytochemistry for choline-acetyltransferase (ChAT) and vesicular glutamate transporters VGLUT1 and VGLUT2 showed that the morphological characteristics and density of the termination of the mossy fibers in the flocculus were unaffected; CHAT is counted as number of rosettes per floccular area, VGLUT1 and VGLUT2 are represented as density (% of fluorescence) per floccular area. (D) Following anterograde tracer injections (insert) into the MVN we observed that the morphological characteristics of labeled mossy fibers in the flocculus of Ptf1a::cre;Robo3lox/lox mice were comparable to those in controls. Mossy fiber rosettes were found on both the ipsilateral and contralateral side; examples shown here are from the contralateral flocculus. Scale bars at (A) indicate 100 mm for PRH, MVN, and NRTP and 50 mm for SVN, at (C) 50 mm and at (D) 100 mm. Error bars denote SEM.

The Molecular Layer Interneurons May Contribute to the Reciprocal Firing of Purkinje Cells in Ptf1a::cre;Robo3lox/lox Mice If the climbing fibers shape the reciprocity in part by enhancing inhibition of Purkinje cells, it is possible that the molecular layer interneurons contribute to the phase-shift in simple-spike activ704 Neuron 78, 700–713, May 22, 2013 ª2013 Elsevier Inc.

ity (Mathews et al., 2012; Jo¨rntell et al., 2010; Szapiro and Barbour, 2007). This hypothesis is attractive, not only because these interneurons directly inhibit Purkinje cells (Barmack and Yakhnitsa, 2008; Ha¨usser and Clark, 1997; Mittmann et al., 2005), but also because the parallel fiber to molecular layer interneuron input may be potentiated by climbing fiber input (Ekerot

Neuron Climbing Fiber Dominance in Reciprocity

Optokinetic Stimulus (º)

A Optokinetic stimulus 0.2 Hz

0.1 Hz

0.8 Hz

0.4 Hz

10

5

0

0

0

0

-5

-2

-1

4 2 0

10 5 0

8 4 0

10 5 0

8 4 0

6 3 0

-10

1

2

CS Responses (Hz)

CS Responses (Hz)

B Complex Spikes Control

6 3 0

Ptf1a::cre;Robo3lox/lox

4 2 0

1

3

5

7

0.5

9

1.5

Time (s) 90 120

3.5

2.5

2

90

1.5

3

90 60

1

30

0

0

300

30

2

60

1.2

1

150

30

0.5

210

330

180

0

330

210

300

240

300

240

270

270

1

1.5

0

300

240

0.8

120

180

330

210

270

0.6

1

180

330

210

0.4

90

60

150

1

180

0.2

2

150

30

2

Time (s) 3

120

2

0.5

240

1

Time (s)

120

60

1.5 150

0.5

4.5

Time (s)

270

SS Responses (Hz)

C Simple Spikes 120 60 0

80 40 0

60 30 0

180 90 0

80 40 0

160 80 0

160 80 0

SS Responses (Hz)

80 40 0

1

3

5

7

9

0.5

1.5

Time (s) 90

100

120

80

90

1

1.5

90 120

60

0

210

330 300

0

330

210 300 270

0.2

0.4

0.8

90 60

60

100

120 30

40

1

1.2

80

60

60

150

30

40 20

180

0

210

330 300

240

0.6

Time (s)

20

180

240

100 80

150

30 50

180

2

Time (s)

150

30

20

270

0.5

4.5

100

40

240

3.5

150

120

60

60

150

2.5

Time (s)

180

0

210

330 300

240

270

270

Figure 3. Complex-Spike and Simple-Spike Reciprocity of VA Purkinje Cells Is Maintained in Ptf1a::cre;Robo3lox/lox Mice

(A) Optokinetic stimulation was provided at 0.1–0.8 Hz (8 /s peak velocity). (B) In vivo extracellular recordings from VA floccular Purkinje cells show complex-spike modulation during VA stimulation. Peristimulus time histograms (PSTHs) depict complex-spike modulation in mutants (red) and controls (blue) (top panels) for each frequency. Polar plots of all complex-spike activity, with the amplitudes depicted by the radius and the phase indicated by the angle, reveal that 48 out of 61 Purkinje cell recordings in Ptf1a::cre;Robo3lox/lox mice show reversed modulation. Each dot represents a single cell. (C) PSTHs and polar plots of simple-spike modulations of the Purkinje cells presented in (B). Note that the majority of the Purkinje cell recordings in Ptf1a::cre;Robo3lox/lox mice have reversed and deeper simple-spike modulation compared to the controls. See also Table S2 for details.

Neuron 78, 700–713, May 22, 2013 ª2013 Elsevier Inc. 705

Neuron Climbing Fiber Dominance in Reciprocity

A

B

Vertical Axis stimulation

●

Control Ptf1a::cre;Robo3lox/lox

● ● ●

Control

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

Ptf1a::cre;Robo3lox/lox

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

VA Purkinje Cells

●

270° 180° 90°

*

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

*

Horizontal Axis stimulation 270° 180° 90°

● ● ● ● ● ● ● ● ● ●

*

*

●

HA Purkinje Cells

● ● ●

● ● ● ●

Non-reciprocal (SS-CS< 90°) Reciprocal ( SS-CS > 90°)

Figure 4. Virtually All VA and HA Floccular Purkinje Cells Show Intact Reciprocity (A) Each line represents one floccular Purkinje cell responding to either vertical axis stimulation (VA Purkinje cells; top) or horizontal axis stimulation (HA Purkinje cells; bottom); Ptf1a::cre;Robo3lox/lox cells are indicated in red, control cells in blue. The angle of each individual line relative to the midline depicts absolute difference in the phase of the complex-spike and simple-spike modulation. Lines on the right show cells that show reciprocity between complex spikes and simple spikes (i.e., 90 to 270 phase difference with 180 being the value for perfect reciprocity); a deflection upward indicates an angle bigger than 180 , whereas a downward deflection indicates a smaller phase difference. Line on the left shows in-phase activity of complex spikes and simple spikes (i.e., 0 –90 and 270 –360 phase difference). Arrows on the right indicate cells in Ptf1a::cre;Robo3lox/lox mice that receive a presumptive normal contralateral climbing fiber input (according to the direction-selectivity of their responses). (B) Schematic illustration of the optokinetic stimulation around the vertical axis and horizontal axis in controls (left panels) and mutants (right panels). The asterisks show the location of the recording side in the flocculus. Note that in Ptf1a::cre;Robo3lox/lox mice the optokinetic stimulation will convey ‘‘reversed’’ signals from the opposite eye due to the rerouted climbing fibers; since the optimal axes for optokinetic modulation in the flocculus correspond to the axes that run through the semicircular canals (Schonewille et al., 2006b), we depict the semicircular canals as a reference frame. Whereas the eye movements in controls accurately follow the target, those in the Ptf1a::cre;Robo3lox/lox mice are unstable and often move in the opposite direction due to a swap in the internal reference frame of the optokinetic system. As can be deduced from the scheme, this swap has impact on both vertical axis and horizontal axis movements.

and Jo¨rntell, 2001), which acts synergistically with depression at the parallel fiber to Purkinje cell input (Gao et al., 2012). Thus, to find out whether the phase of the molecular layer interneurons in the VA zones was shifted together with that of the climbing fibers we recorded their activity during vertical axis optokinetic stimulation at 0.4 Hz. We first identified VA Purkinje cells as described above, and subsequently looked for molecular layer interneurons 706 Neuron 78, 700–713, May 22, 2013 ª2013 Elsevier Inc.

within a radius of 200 mm. The identity of these molecular layer interneurons was determined by their characteristic combination of distribution of interspike intervals, firing frequency and firing regularity (Barmack and Yakhnitsa, 2008; Ruigrok et al., 2011). In addition, we verified these criteria in a separate set of wildtype animals (n = 6) using in vivo whole-cell recordings combined with intracellular labeling of neurobiotin and subsequent

Neuron Climbing Fiber Dominance in Reciprocity

(A) Purkinje cells. Compared to controls (blue) the average simple-spike firing frequency in the mutants (red) during the trough of the modulation was significantly lower, suggesting a change in the inhibition of Purkinje cells. Error bars denote SEM; see Table S2 for details. (B) Interneurons. Raw trace of a vertical-axis (VA) floccular molecular layer interneuron in a Ptf1a::cre;Robo3lox/lox mouse (red) and control (blue) during optokinetic stimulation at 0.4 Hz. (C) Interneurons. Corresponding PSTHs of same cells. (D) Interneurons. Logarithmic scatter plot represents the phase and amplitude of modulation of each individual molecular layer interneuron for both controls and mutants, blue and red squares, respectively. The average phase of the modulating molecular layer interneurons (depicted as hollow rhombus) is also reversed in the mutants compared to the controls.

term plastic process such as potentiation of the parallel fiber to molecular layer interneuron synapse (Ekerot and Jo¨rntell, 2001; Jo¨rntell et al., 2010; Wulff et al., 2009), one expects that the phase-shift in simple-spike activities in Ptf1a::cre;Robo3lox/lox mice is at least partly maintained when the climbing fiber modulation is suddenly minimized. To investigate this prediction we investigated the activities of 37 VA Purkinje cells in Ptf1a::cre; Robo3lox/lox mice during vestibular stimulation. In part of these cells (n = 22) we compared the activity during vestibular stimulation in the dark to that during vestibular stimulation in the light when the drum was stationary (also known as VVOR stimulation; i.e., mostly evoking climbing fiber activity when the head moved to the ipsilateral side), and in the other part (n = 15) we compared the activity during vestibular stimulation in the dark to that during vestibular stimulation in the light, while the drum was rotating inphase with the table at the same amplitude (also known as VOR gain-down stimulation; i.e., mostly evoking climbing fiber activity when the head moved to the contralateral side). On average the absence of the stationary or moving optokinetic image, which evoked most of the climbing fiber activity when the light was on, did not modify the phase by more than 12 degrees in both the first and the second group (Figures 6A and 6B and Table S3). Again, this held true for both the majority of cells that received an ipsilateral climbing fiber projection (n = 13 in the first group; n = 11 in the second group) and the minority of cells that received a presumptive contralateral projection (n = 9 in the first group; n = 4 in the second group). Thus, temporarily removing the visual climbing fiber modulation had a relatively mild impact on the phase of the simple-spike modulation (12 ) compared to the average phase difference of simple-spike activity in the mutants and controls in the light (across all frequencies 145 ; Table S2) suggesting that the phase reversal of the simple spikes in the Ptf1a::cre;Robo3lox/lox mice may at least partly be due to a more long-term effect evoked by the climbing fibers.

histological analysis (Figure S6). We recorded the activity of 20 molecular layer interneurons in Ptf1a::cre;Robo3lox/lox mice, 14 of which were modulating significantly in response to optokinetic stimulation about the vertical axis (see methods for details). In addition, we recorded the activity of 27 molecular layer interneurons in controls, 15 of which were modulating. The phase of the activities of the modulating molecular layer interneurons in Ptf1a::cre;Robo3lox/lox mice (i.e., n = 14) was indeed reversed when compared to controls (i.e., n = 15) (Figures 5B–5D). Their phase with respect to the optokinetic stimulus differed on average 107 from that of controls (p < 0.0001; Student’s t test), implying that the trough and peak of the activity of the modulating molecular layer interneurons approximately coincided with the peak of the simple spikes and complex spikes, respectively, even though the phase of the molecular layer interneurons, taken across all cells, was more broadly tuned than that of the Purkinje cell activity (Figure 5D). Similar to that of Purkinje cell complex-spike activity, the phase locking coefficient of molecular layer interneurons did not differ among mutants and controls (Figure S4C). If the phase-shift in molecular layer interneuron activity is at least partly controlled by a climbing fiber-dependent, long-

Compensatory Eye Movements Are Severely Impaired in Ptf1a::cre;Robo3lox/lox Mice Ptf1a::cre;Robo3lox/lox mice are ataxic and show prominent deficits during locomotion (Figure S7A and Movie S1). Since the climbing fiber input to the flocculus is predominantly driven by visual signals (Graf et al., 1988; Schonewille et al., 2006b), one can expect that the compensatory eye movements driven by vision, i.e., the OKR and vestibulo-ocular reflex in the light (VVOR), are most prominently affected in Ptf1a::cre;Robo3lox/lox mice. The data are in line with this prediction. Ptf1a::cre; Robo3lox/lox mice showed significant deficits in the gain of both their OKR and VVOR (p = 0.001 and p < 0.001, respectively; repeated-measures ANOVA) as well as in the phase of both reflexes (p = 0.020 and p = 0.027; repeated-measures ANOVA) (Figures 7A and 7B). Moreover, the variability of OKR values and that of VVOR values were also significantly higher than those of controls (SD, for gain and phase values of OKR and VVOR, all p < 0.001; Student’s t test). If the direction of most of the visual climbing fiber signals is reversed, one expects that vision does not enhance the VOR. This prediction is again upheld. In contrast to controls, both gain and phase values during VOR in the light (i.e., VVOR) were not significantly different from those during VOR in the dark in Ptf1a::cre;Robo3lox/lox mice (p = 0.64; and

60

B Interneurons

Ptf1a::cre;Robo3lox/lox VA **

**

HA **

*

**

50

1 mV

40 30 20 10 0

0.1

0.2

0.4

0.4

0.8

Stimulus Frequency (Hz)

D Interneurons Amplitude of Modulation (Hz)

Interneuron Spike Optokinetic Stimulus (º) Rate (Hz)

C Interneurons 2 0 -2

5 2.5 0 8 4 0 0.5

1

500 ms

Raw Traces Interneuron

SS Trough Modulation Amplitude (Hz)

Control

Optokinetic Stimulus

A Purkinje Cell

1.5

Time (s)

2

64 32 16 8 4 2 1 0

60

120 180

240 300

360

Phase of modulation peak (º)

Figure 5. Inhibition May Contribute to Phase Shift in Simple-Spike Modulation in Ptf1a::cre;Robo3lox/lox Mice

Neuron 78, 700–713, May 22, 2013 ª2013 Elsevier Inc. 707

Neuron Climbing Fiber Dominance in Reciprocity

A

Vestibular Stationary Paradigm

B

5 2.5 0 -2.5 -5

Vestibular Stimulus (°)

5 2.5 0 -2.5 -5 100 50 0

Simple Spike 100 50 FF LIGHT 0 (Hz)

Simple Spike FF DARK (Hz)

80 40 0

60 30 0 0

0.2

0.4

0.6

0.8 1 Time (s)

1.2

1.4

1.6

90

90

60

60

Phase shift (º)

Phase shift (º)

Vestibular Suppression Paradigm

30

0

20 40 60 80 100 ∆ Amplitude SS Light-Dark (Hz)

120

0

0.2

0.4

0.6

0.8 1 Time (s)

1.2

0

20 40 60 ∆ Amplitude SS Light-Dark (Hz)

1.4

30

-10

p = 0.20 for gain and phase, respectively; repeated-measures ANOVA) (compare Figures 7B and 7C). One can go even one step further and predict that Ptf1a::cre;Robo3lox/lox mice should perform worse than when there is no output of the flocculus at all, because the modulation of their Purkinje cell activity is opposite to what is needed. Indeed, if one compares for example the visual contribution to the VOR (i.e., the difference between VVOR and VOR) in Ptf1a::cre;Robo3lox/lox mice to that recorded previously in lurcher mice (Van Alphen et al., 2002), which lack Purkinje cell output, the Ptf1a::cre;Robo3lox/lox mice perform significantly worse (p < 0.01 for both gain and phase values; Student’s t test) (Figure S7B, top panels). Moreover, even the variability of the visual contributions in the Ptf1a::cre;Robo3lox/lox mice was significantly greater than that of the lurchers (p < 0.01; Student’s t test) (Figure S7B, bottom panel). We conclude that motor performance in Ptf1a::cre;Robo3lox/lox mice is robustly affected and that the severity and modality of their motor deficits during compensatory eye movements are in line with the changes observed in their Purkinje cell activity. DISCUSSION Rerouting of the Climbing Fiber Projection in Ptf1a::cre;Robo3lox/lox Mutants The climbing fiber projection to the flocculus of the vestibulocerebellum in Ptf1a::cre;Robo3lox/lox mutants is, like that to other parts of their cerebellum (Morara et al., 2001; Renier et al., 2010), predominantly, but not exclusively, ipsilateral. In contrast, the laterality of the mossy fiber projections to the flocculus in these mutants is unaffected. This is due to the fact that Robo3 is expressed by precerebellar neurons only for about 24–36 hr 708 Neuron 78, 700–713, May 22, 2013 ª2013 Elsevier Inc.

80

1.6

Figure 6. Phase Shift in Simple-Spike Modulation Persists in the Absence of Complex-Spike Modulation (A) (From top to bottom) Vestibular stimulation was provided at 0.6 Hz and 5 amplitude, while the optokinetic stimulus was stationary when present. PSTH of simple-spike activity of a VA Purkinje cell in a Ptf1a::cre;Robo3lox/lox mouse during vestibular stimulation in the light (VVOR) and dark (VOR) before and after turning off the light. Phase shift and change in the amplitude of modulation of simple-spike activity comparing light and dark conditions for each individual cell (black dashed lines and dots). Red lines on y axis and x axis indicate average phase shift and average amplitude difference among light and dark conditions across 22 VA Purkinje cells. (B) Same as for (A), but now while the optokinetic stimulus was moving in-phase with the vestibular stimulus when the light was on (also called vestibular gain-down or suppression paradigm). Here 15 cells are depicted. The simple-spike phase in the dark was not significantly different from that in the light in either paradigm (p = 0.82 for VVOR and p = 0.14 for gain-down; paired-samples Student’s t test), whereas the depth of modulation was significantly higher during stimulation in the light (p < 0.001 for VVOR and p = 0.009 for gaindown; paired-samples Student’s t test). See Table S3 for details.

during their migration to the floor plate (between E12 and E16 depending on the nucleus) and that during this period Ptf1a is only expressed by a large majority of the inferior olivary neurons, but not by mossy fiber projection neurons (Renier et al., 2010; Yamada et al., 2007). Likewise, in the cerebellum Robo3 is transiently expressed around E13 by some cerebellar nuclei neurons but is absent from E15 on and therefore does not play a role at later stages of development such as target recognition or synaptogenesis. Hence, the Ptf1a::cre;Robo3lox/lox mutant turns out to be an excellent model to study the role of climbing fibers in controlling reciprocity of complex spikes and simple spikes, because it allows us to disentangle the role of the climbing fibers from that of the mossy fibers and because it presents a predominant, but not complete, change in laterality providing a perfect control within subjects. Climbing Fiber Input Dominates Reciprocity of Complex-Spike and Simple-Spike Firing Our electrophysiological recordings reveal a dominant role of the climbing fiber input in phase control of simple-spike firing. The reciprocity was maintained in virtually all VA and HA cells in the Ptf1a::cre;Robo3lox/lox mice, even in the minority of Purkinje cells that had a preserved contralateral climbing fiber input (of all 117 cells recorded in this study 116 showed reciprocity; Figure 4A). Thus, even if the laterality of the mossy fiber projection would have been completely reversed in the Ptf1a::cre;Robo3lox/lox mice, it still would have been difficult to explain the nearly perfect maintenance of the reciprocity of complex-spike and simple-spike firing without the dominant shaping impact of the climbing fibers. Our data are in line with the hypothesis that the climbing fibers control various forms of

Neuron Climbing Fiber Dominance in Reciprocity

A

1.2

OKR

1.0

0.6 0.4

VVOR

-60

-120

0.2

B

0 Phase (º)

Gain

0.8

Figure 7. Compensatory Eye Movements Are Impaired in Ptf1a::cre;Robo3lox/lox Mice

60

Control Ptf1a::cre;Robo3lox/lox

0.0

-180

1.2

180

1.0 120 Phase (º)

Gain

0.8 0.6 0.4

0

0.2

C

VOR

60

0.0

-60

1.2

180

1.0 120 Phase (º)

Gain

0.8 0.6 0.4

60 0

0.2 0.0

-60 0

0.2

0.4 0.6 0.8 Frequency (Hz)

1

0

plasticity in the molecular layer of the cerebellar cortex in a synergistic fashion (Gao et al., 2012) possibly filtering the widespread and diverse information provided by the granular layer (Dean et al., 2010). Indeed, it appears possible that the phase information carried by a particular subset of parallel fibers is synergistically enhanced by various forms of climbing fiber mediated plasticity and that these signals are used to improve motor coordination (De Zeeuw et al., 2011). In this respect, it is interesting to note that severe disorganization of the pontine nucleus, the main source of mossy fibers (Ruigrok, 2003; Voogd and Barmack, 2006), does not lead to ataxia in human HGPPS patients carrying mutations in ROBO3, presumably as there is no conflict of laterality in the olivary information flow (Bosley et al., 2005; Jen et al., 2004). Thus, in line with distributed synergistic plasticity, which is controlled by the climbing fiber system (Gao et al., 2012), proper laterality of climbing fiber inputs and their feedback to the olive may compensate rather well for disorganization in the mossy fiber system, but not the other way around. Disruption in the Olivocerebellar System Leads to Severe Motor Deficits In principle, the changes in phase of the simple-spike modulation that result from the shift in climbing fiber modulation may be sufficient to explain the behavioral changes in the Ptf1a::cre;Robo3lox/lox mice. Since the climbing fiber input to the flocculus is predominantly driven by vision (Graf et al., 1988; Schonewille et al., 2006b), it is not surprising that the differences in behavior among the mutants and controls were most prominent during OKR and VVOR (or visual contribu-

0.2

0.4 0.6 0.8 Frequency (Hz)

1

(A) The optokinetic reflex (OKR) is impaired in Ptf1a::cre;Robo3lox/lox mice (n = 10, red) in that gain (amplitude) is lower than that in controls (n = 4, blue) (p = 0.001; repeated-measures ANOVA) and that phase (timing) is lagging that in controls (p = 0.020; repeated-measures ANOVA). Note that error bars represent SD to indicate the range of responses. (B) Similarly, average gain and phase of the visually-enhanced vestibulo-ocular reflex (VVOR) are affected in the mutants (p < 0.001 and p = 0.027, respectively; repeated-measures ANOVA). (C) The phase of the VOR in Ptf1a::cre;Robo3lox/lox mice was also significantly affected (p = 0.005; repeated-measures ANOVA), and there was a trend toward lower gain (p = 0.067; repeatedmeasures ANOVA). Note that the variability of the gain and phase values of the paradigms that include vision (i.e., OKR and VVOR) was significantly greater in Ptf1a::cre;Robo3lox/lox mice than in controls (for gain and phase values, all p < 0.001; Student’s t test); note that, unlike those of the mutants, many of the SDs of the phase values of controls are in fact so small that they are hidden behind the symbols for the averages. In contrast, during VOR the differences in variability were less pronounced (p = 0.10; and p = 0.003; Student’s t test). Error bars indicate SDs.

tion to the VOR; i.e., VVOR minus VOR). Accordingly, vision did apparently not allow any enhancement of the VOR in Ptf1a::cre;Robo3lox/lox mice in that both gain and phase values during VVOR were not different from those during VOR and that they were actually worse than those of animals with no Purkinje cell output at all. This suggests that the floccular target neurons in Ptf1a::cre;Robo3lox/lox mice are counteracted by their Purkinje cell activity, which is in line with the notion that the olivocerebellar system operates as a phase lead-lag compensator to control eye movement behavior (De Zeeuw et al., 1995). Although we focused here on eye movement control because of the reliable tractability of the behavior and electrophysiological recordings, the overall and severe ataxic behavior of the Ptf1a::cre;Robo3lox/lox mice indicates that similar disturbances in complex-spike and simple-spike activity may also take place in other parts of their cerebellar sensorimotor system. For example, Movie S1 illustrates that the direction of their limb movements is often opposite to what is optimal. Pushing and pulling, retraction and protraction, and clockwise and counterclockwise rotation are frequently reversed. The finding that complex-spike activity of primate Purkinje cells in nonvestibular cerebellar regions that are involved in limb movement control also show direction-selectivity and reciprocal firing with simple-spike activity during physiological operations (Ebner et al., 2002) further highlights the possibility that control of reciprocity by the climbing fiber system is critical for cerebellar function. Likewise, the whisker system of rodents shows similar intriguing patterns of reciprocity that can deviate following various manipulations and mutations of the olivo-cerebellar Neuron 78, 700–713, May 22, 2013 ª2013 Elsevier Inc. 709

Neuron Climbing Fiber Dominance in Reciprocity

system (Bosman et al., 2010; Bryant et al., 2010; C.I.D.Z., unpublished observations). Potential Mechanisms Mediating Climbing Fiber Dominance in Reciprocity Several of our findings suggest that the underlying mechanisms by which the climbing fibers form and preserve the reciprocity may include some form of long-term effect involving inhibitory molecular layer interneurons (Jo¨rntell et al., 2010; Jo¨rntell and Ekerot, 2003; Szapiro and Barbour, 2007). First, the trough of the simple-spike modulation in the Ptf1a::cre; Robo3lox/lox mutants was significantly lower than that in controls at all frequencies tested; since Purkinje cells are intrinsically active (Ha¨usser et al., 2004; Raman and Bean, 1999), it presumably takes an active form of inhibition to reach such low level of simple-spike activity. Second, the phase of the modulating molecular layer interneurons in the Ptf1a::cre; Robo3lox/lox mice was shifted in the proper direction to exert the phase shift in simple-spike activity. Third, the phase shift in simple-spike activity was largely maintained when the modulation of the visual climbing fibers was stopped; this preservation of simple-spike modulation points toward, at least in part, the contribution of a form of a long-term plastic process that depends on climbing fiber activation. And finally, blockage of the output of molecular layer interneurons indeed impairs phase reversal learning of the vestibulo-ocular reflex (Wulff et al., 2009). Whether any form of plasticity involving molecular layer interneurons is actually involved in the formation and/or maintenance of reciprocity remains to be demonstrated. At least two forms of plasticity, both of which are climbing fiber dependent, are potential candidates; these include induction of potentiation at the parallel fiber to molecular layer interneuron synapse (Jo¨rntell et al., 2010; Jo¨rntell and Ekerot, 2003) and rebound potentiation at the molecular layer interneuron to Purkinje cell synapse (Kano, 1996; Kano et al., 1992). In addition, other phenomena such as a direct control of the climbing fibers onto the molecular layer interneurons via glutamate spillover (Mathews et al., 2012; Szapiro and Barbour, 2007) or tonic GABA release from Bergmann glia cells (Lee et al., 2010), which may be activated by climbing fibers releasing corticotropin-releasing factor (Tian et al., 2008), could in principle also contribute to inhibition and reciprocal firing of Purkinje cells. Likewise, induction of long-term potentiation or long-term depression at the parallel fiber to Purkinje cell synapse (Ito, 2001; Schonewille et al., 2010), which are associated with the absence and presence of climbing fiber activity, respectively (Coesmans et al., 2004), may also contribute to the mechanisms involved in the reciprocity of Purkinje cell firing. All these forms of cerebellar cortical plasticity appear to operate in synergy in that potentiation at the parallel fiber to molecular layer interneuron synapse and at the molecular layer interneuron to Purkinje cell synapse occur concomitantly with depression at the parallel fiber to Purkinje cell synapse when the climbing fibers are active, and vice versa these synapses are depressed and potentiated, respectively, when climbing fibers are silent (Gao et al., 2012; Ito, 2001; Jo¨rntell et al., 2010; Kano, 1996; Schonewille et al., 2010). Thus, climbing fibers appear to have 710 Neuron 78, 700–713, May 22, 2013 ª2013 Elsevier Inc.

multiple mediators in place when they are bound to depress Purkinje cells simple-spike activity and thereby to shape the reciprocity. EXPERIMENTAL PROCEDURES Mouse Lines All experiments involving animals were conducted in accordance with The Dutch Ethical Committee for animal experiments. The Rig-1/Robo3::GFP knockout and the Ptf1a::cre;Robo3lox/lox mouse lines used in this study have been described previously (Renier et al., 2010; Sabatier et al., 2004). Robo3 conditional knockout mice (Robo3lox/lox, Institut Clinique de la Souris, Illkirch, France; http://www-mci.u-strasbg.fr) were crossed with Ptf1a::Cre mice (C.V. Wright, Vanderbilt University) and Ptf1a::cre;Robo3lox/lox (mutant), Robo3+/
, Robo3lox/lox or Ptf1a::cre;Robo3lox/+ were used for experiments. To identify the neurons expressing cre recombinase in Ptf1a:cre animals, Ptf1a::cre mice were crossed to Tau-lox-Stop-lox-mGFP-IRES-nls-lacZ mice (TaumGFP) (Hippenmeyer et al., 2005). Extracellular Electrophysiology of Purkinje Cells In Vivo Mice (15–40 weeks old) were prepared for recordings as described previously (Wulff et al., 2009). In short, to create a recording chamber a craniotomy was made in the left occipital bone under general anesthesia with isoflurane/O2. After 3 days of recovery the animals were placed onto a turntable surrounded by a random-dotted cylindrical screen and glass electrodes were lowered into the flocculus of the cerebellum. Cells were recorded during optokinetic and/or vestibular stimulation. Signals were filtered, amplified and stored for off-line analysis. Single-unit activity was confirmed by a pause in simple-spike firing following each complex-spike (De Zeeuw et al., 1995). Tuning curves of complex spikes were made to determine the optimal axis of stimulation (Schonewille et al., 2006b) and cells were categorized according to their optimal responses to stimulation around the vertical axis (VA) or horizontal axis (HA) Cells were recorded during optokinetic stimulation at frequencies ranging from 0.1 to 0.8 Hz with a peak velocity of 8 /s and/or during vestibular stimulation at 0.6 Hz and an amplitude of 5 degrees. Extracellular Electrophysiology of Molecular Layer Interneurons In Vivo VA molecular layer interneurons in the flocculus were identified by their waveform, firing frequency and pattern (Ruigrok et al., 2011), immediate proximity (<200 mm) of a VA Purkinje cell, and by their location in the molecular layer, which can be recognized by the direction of the complex-spike potential (negative) (De Zeeuw et al., 1995; Graf et al., 1988). They were recorded during optokinetic stimulation at 0.4 Hz. The main characteristics of our extracellularly recorded molecular layer interneurons in awake mice (CV log 0.2 ± 0.01; CV2 0.62 ± 0.03; 5th percentile of ISIs 0.015 ± 0.001; median ISI 0.09 ± 0.03) were in line with those obtained by Ruigrok et al. (2011) obtained in anesthetized rats. Using their decision tree 100% of the cells fulfilled the first two criteria for molecular layer interneurons, while 64% of the interneurons also followed the third criterion (based on the logarithmic CV and percentile of interspike intervals) and were thus categorized as stellate or basket cells. The remaining 36% were part of the so-called border interneuron group due to relatively high values for the 5% percentile and relatively low values for the logarithmic CV. Indeed, the firing frequency in awake mice (21.02 ± 12.66 Hz) was somewhat higher than in Ruigrok’s experiments in anesthetized rats (9 Hz) and in our whole-cell experiments in anesthetized mice (7.1 ± 10.2 Hz), presumably due to the impact of ketamine and xylazine (see also Supplemental Experimental Procedures; Figure S6). Tracing and Immunocytochemistry For the retrograde tracing experiments the flocuclus was identified with the use of extracellular electrophysiological recordings after which the recording electrode was consecutively replaced with a pipette containing cholera toxin B-subunit (CTb) and a pipette containing gold-lectin conjugate of bovine serum albumin with wheat germ agglutinate and 10 nm gold sol (Aurion, Wageningen, NL) (for technical details see Ruigrok et al., 1995 and Ruigrok

Neuron Climbing Fiber Dominance in Reciprocity

and Apps, 2007). Four days postinjection the animals were anesthetized with pentobarbital and perfused with paraformaldehyde and glutaraldehyde, and the brains were embedded in gelatin and sectioned transversally on a freezing microtome (40 mm). Sections were processed for anti-CTb immunocytochemistry and silver intensification of gold-lectin labeling and subsequently mounted on chrome-gelatinized glass slides, air-dried, counterstained with thionin, and coverslipped. For the anterograde tracing experiments the medial vestibular nucleus was identified with the use of extracellular electrophysiological recordings after which the recording electrode was exchanged for a pipette (15–18 mm tip) containing Alexa-594 DEXTRAN 10.000 dissolved in 0.5 M NaCl (D22919, Invitrogen). Five days after pressure injection the animals were anesthetized with pentobarbital and perfused with 4% PFA. Sections were washed with PBS, counterstained with DAPI, mounted on glass and covered with VectaShield (Vector Laboratories, Inc.). Immunocytochemistry of VGLUT1, VGLUT2, and ChAT was done on sections of animals perfused with 4% PFA. Sections were immunostained with rabbit anti-VGLUT1 (1:2,000, Synaptic Systems), guinea pig anti-VGLUT2 (1:2,000, Chemicon), and goat anti-ChAT (1:200, Chemicon) as primaries and with donkey anti-rabbit-CY5, donkey anti-guinea pig Cy3 and donkey anti-goat FITC as secondaries (all antibodies 1:400, Jackson) in PBS containing 2% NS and 0.4% Triton. Sections were counterstained with DAPI, mounted on glass and covered with VectaShield (Vector Laboratories, Inc.). For quantification of neurons in the olivary subdivisions serial cross-sections of the brainstem were harvested and immunostained for Foxp2 (Abcam). Cross-section areas of the subnuclei were quantified in the middle of each subnucleus and Foxp2+ neurons were counted in two sections from each animal. Eye Movement Recordings Mice (12–30 weeks old) were prepared under general anesthesia with isoflurane/O2. A construct with two nuts or a magnet in a holder with a screw hole was attached to the skull as described above. After R3 days of recovery, mice were placed in a restrainer with the head construct fixed to a metal bar. The restrainer was fixed onto the turntable surrounded by the screen described above. Prior to the behavioral measurements the animals received one habituation session (1 hr in the restrainer) to get used to the experimental settings. The optokinetic reflex (OKR) and vestibulo-ocular reflexes (VOR in dark, VVOR in light) were elicited by rotating the screen and turntable, respectively, at different frequencies (AC servo-motors, Harmonic Drive AG, MA, USA). The position of table and drum were recorded by potentiometers and the signal was digitized (CED Limited, UK) and stored for off-line analysis. Eye movements were recorded at 240 Hz using an infrared CCD camera fixed to the turntable (ISCAN Inc., MA, USA) described (Stahl et al., 2000). Eye movements were calibrated, a sine was fitted through averaged eye and stimulus velocity signals and the ratio (gain) and the temporal shift (phase) were calculated. All calibrations and subsequent eye movement computations were performed as described previously (Goossens et al., 2004). Data Analysis Off-line analysis of the eye movements and spiking activity was performed using SpikeTrain (Neurasmus BV, Rotterdam, NL) and custom-made Matlab (MathWorks) routines (Batschelet, 1981; Goossens et al., 2004; Stahl et al., 2000). Regularity of inter simple spike intervals (ISI) was determined for the entire recording period by ISI SD / ISI mean (CV) and between adjacent intervals (CV2) as the mean of 2 * jISIn – ISIn+1j/(ISIn + ISIn+1) (Schonewille et al., 2006a). The mean spiking frequencies during peak and trough were calculated from 10 bins (bin size 3.6 degree) preceding and 10 bins following the peak and trough of the fitted sine wave in the peristimulus phase histograms. The phase locking coefficients (r) were calculated with respect to the stimulus and significance was tested with the Rayleigh statistic (z value, p < 0.01) (Batschelet, 1981). Cells that did not show significance for the phase locking coefficient were excluded from further analysis. For in vitro off-line analysis we used Patch Master (HEKA Electronics). Histological material presented in Figures 1B, Figures 2A and 2E and Figure S2A was analyzed with a Leica DMR light microscope equipped with a DC 300 digital camera. Labeled neurons in the inferior olive and brainstem were plotted serially from one out of four series of sections with the use of an Olympus microscope equipped with a Lucivid miniature

monitor and Neurolucida TM software (Microbrightfield, Colchester, VT, USA). The reconstructions of drawings and photo panels were performed in Adobe Photoshop CS3 Version 10.0, and brightness, contrast and color balance were corrected. Quantification of the immunocytochemical data for Figure S1B and Table S1 were performed as previously described (Renier et al., 2010). Images of immunocytochemical stainings presented in Figures 1D, 2C, 2D, and Figure S2B were collected on Zeiss 700 confocal microscope and processed in ImageJ where brightness contrast and color balance were adjusted. Density measures of VGLUT1 and VGLUT2 were performed in ImageJ (images where transformed using z stack maximum projection, thresholding and finally area measurements with ROIs drawn around the flocculus). Density measures for ChAT were also performed in ImageJ but given the sparse labeling the ChAT positive mossy fibers were manually counted using a ImageJ ‘‘Cell Counter’’ plugin. Statistical Analyses Unless stated otherwise, single comparisons were performed with two-tailed paired or un-paired Student’s t tests, as appropriate. Electrophysiological recordings and eye movement recordings were statistically analyzed using ANOVA as indicated. Differences were considered significant when p < 0.05. Unless stated otherwise, all data are presented as mean ± standard error of the mean (SEM).

SUPPLEMENTAL INFORMATION Supplemental Information includes seven figures, three tables, one movie, and Supplemental Experimental Procedures and can be found with this article online at http://dx.doi.org/10.1016/j.neuron.2013.03.018. ACKNOWLEDGMENTS We thank E. Haasdijk, R. de Avila Freire, E. Goedknegt, and M. Rutteman for their technical assistance; C.V. Wright (Vanderbilt Univerisity) for providing the Ptf1a::Cre mice; and we thank the Dutch Organization for Medical Sciences (ZonMw; C.I.D.Z., F.E.H.), Life Sciences (ALW; M.S., C.I.D.Z., F.E.H.), Senter (Neuro-Basic; C.I.D.Z.), Erasmus University Fellowship (M.S., F.E.H.), the ERC, CEREBNET, and C7 programs of the European Community (C.I.D.Z.), the ‘‘fondation pour le recherche me´dicale’’ - programme Equipe FRM (A.C.), the Association Franc¸aise contre les Myopathies - AFM, ASSSUB06-00123 (A.C.), the Labex lifesenses (A.C.), and the Agence Nationale de la Recherche ANR-2011 BSV40091(A.C.) for their financial support. A.B. is currently a postdoctoral research associate at the Department of Molecular Biology and Princeton Neuroscience Institute, Princeton University, Princeton, NJ, USA. C.I.D.Z. conceived the project. A.B. and M.S. obtained and analyzed the in vivo electrophysiological, behavioral, and histological data. K.V. designed MatLab software for spiking and eye movement analysis and analyzed in vivo electrophysiological data. L.W. performed in vivo patchclamp recordings, Z.G., E.G., and F.E.H. performed the in vitro experiments in slice preparation. N.R. and A.C engineered and analyzed the Ptf1a and Robo mutants. A.B., M.S., A.C., and C.I.D.Z. designed the experiments, discussed the results, and prepared the manuscript. Accepted: March 14, 2013 Published: May 2, 2013 REFERENCES Barmack, N.H., and Yakhnitsa, V. (2008). Functions of interneurons in mouse cerebellum. J. Neurosci. 28, 1140–1152. Batschelet, E. (1981). Circular Statistics in Biology (New York: Academic Press). Bosley, T.M., Salih, M.A., Jen, J.C., Lin, D.D., Oystreck, D., Abu-Amero, K.K., MacDonald, D.B., al Zayed, Z., al Dhalaan, H., Kansu, T., et al. (2005). Neurologic features of horizontal gaze palsy and progressive scoliosis with mutations in ROBO3. Neurology 64, 1196–1203.

Neuron 78, 700–713, May 22, 2013 ª2013 Elsevier Inc. 711

Neuron Climbing Fiber Dominance in Reciprocity

Bosman, L.W.J., Koekkoek, S.K.E., Shapiro, J., Rijken, B.F.M., Zandstra, F., van der Ende, B., Owens, C.B., Potters, J.-W., de Gruijl, J.R., Ruigrok, T.J.H., and De Zeeuw, C.I. (2010). Encoding of whisker input by cerebellar Purkinje cells. J. Physiol. 588, 3757–3783. Bryant, J.L., Boughter, J.D., Gong, S., LeDoux, M.S., and Heck, D.H. (2010). Cerebellar cortical output encodes temporal aspects of rhythmic licking movements and is necessary for normal licking frequency. Eur. J. Neurosci. 32, 41–52. Coesmans, M., Weber, J.T., De Zeeuw, C.I., and Hansel, C. (2004). Bidirectional parallel fiber plasticity in the cerebellum under climbing fiber control. Neuron 44, 691–700. De Zeeuw, C.I., Wylie, D.R., Stahl, J.S., and Simpson, J.I. (1995). Phase relations of Purkinje cells in the rabbit flocculus during compensatory eye movements. J. Neurophysiol. 74, 2051–2064. De Zeeuw, C.I., Hoebeek, F.E., Bosman, L.W., Schonewille, M., Witter, L., and Koekkoek, S.K. (2011). Spatiotemporal firing patterns in the cerebellum. Nat. Rev. Neurosci. 12, 327–344. Dean, P., Porrill, J., Ekerot, C.F., and Jo¨rntell, H. (2010). The cerebellar microcircuit as an adaptive filter: experimental and computational evidence. Nat. Rev. Neurosci. 11, 30–43. Ebner, T.J., Johnson, M.T., Roitman, A., and Fu, Q. (2002). What do complex spikes signal about limb movements? Ann. N Y Acad. Sci. 978, 205–218. Ekerot, C.F., and Jo¨rntell, H. (2001). Parallel fibre receptive fields of Purkinje cells and interneurons are climbing fibre-specific. Eur. J. Neurosci. 13, 1303–1310. Fujita, H., and Sugihara, I. (2012). FoxP2 expression in the cerebellum and inferior olive: development of the transverse stripe-shaped expression pattern in the mouse cerebellar cortex. J. Comp. Neurol. 520, 656–677. Gao, Z., van Beugen, B.J., and De Zeeuw, C.I. (2012). Distributed synergistic plasticity and cerebellar learning. Nat. Rev. Neurosci. 13, 619–635. Gebre, S.A., Reeber, S.L., and Sillitoe, R.V. (2012). Parasagittal compartmentation of cerebellar mossy fibers as revealed by the patterned expression of vesicular glutamate transporters VGLUT1 and VGLUT2. Brain Struct. Funct. 217, 165–180.

Kano, M. (1996). Long-lasting potentiation of GABAergic inhibitory synaptic transmission in cerebellar Purkinje cells: Its properties and possible mechanisms. Behav. Brain Sci. 19, 503–527. Kano, M., Rexhausen, U., Dreessen, J., and Konnerth, A. (1992). Synaptic excitation produces a long-lasting rebound potentiation of inhibitory synaptic signals in cerebellar Purkinje cells. Nature 356, 601–604. Kawaguchi, Y., Cooper, B., Gannon, M., Ray, M., MacDonald, R.J., and Wright, C.V. (2002). The role of the transcriptional regulator Ptf1a in converting intestinal to pancreatic progenitors. Nat. Genet. 32, 128–134. Lee, S., Yoon, B.E., Berglund, K., Oh, S.J., Park, H., Shin, H.S., Augustine, G.J., and Lee, C.J. (2010). Channel-mediated tonic GABA release from glia. Science 330, 790–796. Leonard, C.S., and Simpson, J.I. (1986). Simple Spike Modulation of Floccular Purkinje Cells during the Reversible Blockade of Their Climbing Fiber Afferents (Oxford: Pergamon). Marillat, V., Sabatier, C., Failli, V., Matsunaga, E., Sotelo, C., Tessier-Lavigne, M., and Che´dotal, A. (2004). The slit receptor Rig-1/Robo3 controls midline crossing by hindbrain precerebellar neurons and axons. Neuron 43, 69–79. Mathews, P.J., Lee, K.H., Peng, Z., Houser, C.R., and Otis, T.S. (2012). Effects of climbing fiber driven inhibition on Purkinje neuron spiking. J. Neurosci. 32, 17988–17997. Medina, J.F., and Lisberger, S.G. (2008). Links from complex spikes to local plasticity and motor learning in the cerebellum of awake-behaving monkeys. Nat. Neurosci. 11, 1185–1192. Mittmann, W., Koch, U., and Ha¨usser, M. (2005). Feed-forward inhibition shapes the spike output of cerebellar Purkinje cells. J. Physiol. 563, 369–378. Montarolo, P.G., Palestini, M., and Strata, P. (1982). The inhibitory effect of the olivocerebellar input on the cerebellar Purkinje cells in the rat. J. Physiol. 332, 187–202. Morara, S., van der Want, J.J., de Weerd, H., Provini, L., and Rosina, A. (2001). Ultrastructural analysis of climbing fiber-Purkinje cell synaptogenesis in the rat cerebellum. Neuroscience 108, 655–671. Peltier, A.C., and Bishop, G.A. (1999). The site of origin of calcitonin generelated peptide-like immunoreactive afferents to the inferior olivary complex of the mouse. Neurosci. Res. 34, 177–186.

Goossens, H.H., Hoebeek, F.E., Van Alphen, A.M., Van Der Steen, J., Stahl, J.S., De Zeeuw, C.I., and Frens, M.A. (2004). Simple spike and complex spike activity of floccular Purkinje cells during the optokinetic reflex in mice lacking cerebellar long-term depression. Eur. J. Neurosci. 19, 687–697.

Person, A.L., and Raman, I.M. (2011). Purkinje neuron synchrony elicits time-locked spiking in the cerebellar nuclei. Nature 481, 502–505.

Graf, W., Simpson, J.I., and Leonard, C.S. (1988). Spatial organization of visual messages of the rabbit’s cerebellar flocculus. II. Complex and simple spike responses of Purkinje cells. J. Neurophysiol. 60, 2091–2121.

Raman, I.M., and Bean, B.P. (1999). Ionic currents underlying spontaneous action potentials in isolated cerebellar Purkinje neurons. J. Neurosci. 19, 1663–1674.

Ha¨usser, M., and Clark, B.A. (1997). Tonic synaptic inhibition modulates neuronal output pattern and spatiotemporal synaptic integration. Neuron 19, 665–678.

Renier, N., Schonewille, M., Giraudet, F., Badura, A., Tessier-Lavigne, M., Avan, P., De Zeeuw, C.I., and Che´dotal, A. (2010). Genetic dissection of the function of hindbrain axonal commissures. PLoS Biol. 8, e1000325.

Ha¨usser, M., Raman, I.M., Otis, T., Smith, S.L., Nelson, A., du Lac, S., Loewenstein, Y., Mahon, S., Pennartz, C., Cohen, I., and Yarom, Y. (2004). The beat goes on: spontaneous firing in mammalian neuronal microcircuits. J. Neurosci. 24, 9215–9219.

Rizzolatti, G., and Wolpert, D.M. (2005). Motor systems. Curr. Opin. Neurobiol. 15, 623–625.

Hippenmeyer, S., Vrieseling, E., Sigrist, M., Portmann, T., Laengle, C., Ladle, D.R., and Arber, S. (2005). A developmental switch in the response of DRG neurons to ETS transcription factor signaling. PLoS Biol. 3, e159. Ito, M. (2001). Cerebellar long-term depression: characterization, signal transduction, and functional roles. Physiol. Rev. 81, 1143–1195. Jen, J.C., Chan, W.M., Bosley, T.M., Wan, J., Carr, J.R., Ru¨b, U., Shattuck, D., Salamon, G., Kudo, L.C., Ou, J., et al. (2004). Mutations in a human ROBO gene disrupt hindbrain axon pathway crossing and morphogenesis. Science 304, 1509–1513. Jo¨rntell, H., and Ekerot, C.F. (2003). Receptive field plasticity profoundly alters the cutaneous parallel fiber synaptic input to cerebellar interneurons in vivo. J. Neurosci. 23, 9620–9631. Jo¨rntell, H., Bengtsson, F., Schonewille, M., and De Zeeuw, C.I. (2010). Cerebellar molecular layer interneurons - computational properties and roles in learning. Trends Neurosci. 33, 524–532.

712 Neuron 78, 700–713, May 22, 2013 ª2013 Elsevier Inc.

Ruigrok, T.J. (2003). Collateralization of climbing and mossy fibers projecting to the nodulus and flocculus of the rat cerebellum. J. Comp. Neurol. 466, 278–298. Ruigrok, T.J., and Apps, R. (2007). A light microscope-based double retrograde tracer strategy to chart central neuronal connections. Nat. Protoc. 2, 1869–1878. Ruigrok, T.J., Teune, T.M., van der Burg, J., and Sabel-Goedknegt, H. (1995). A retrograde double-labeling technique for light microscopy. A combination of axonal transport of cholera toxin B-subunit and a gold-lectin conjugate. J. Neurosci. Methods 61, 127–138. Ruigrok, T.J., Hensbroek, R.A., and Simpson, J.I. (2011). Spontaneous activity signatures of morphologically identified interneurons in the vestibulocerebellum. J. Neurosci. 31, 712–724. Sabatier, C., Plump, A.S., Le Ma, Brose, K., Tamada, A., Murakami, F., Lee, E.Y., and Tessier-Lavigne, M. (2004). The divergent Robo family protein rig-1/Robo3 is a negative regulator of slit responsiveness required for midline crossing by commissural axons. Cell 117, 157–169.

Neuron Climbing Fiber Dominance in Reciprocity

Schmahmann, J.D. (2012). Cerebellar cognitive affective syndrome and the neuropsychiatry of the cerebellum. In Handbook of the Cerebellum and Cerebellar Disorders, Volume 3, M. Manto, D. Gruol, J.D. Schmahmann, N. Koibuchi, and F. Rossi, eds. (New York: Springer), pp. 1717–1752. Schonewille, M., Khosrovani, S., Winkelman, B.H., Hoebeek, F.E., De Jeu, M.T., Larsen, I.M., Van der Burg, J., Schmolesky, M.T., Frens, M.A., and De Zeeuw, C.I. (2006a). Purkinje cells in awake behaving animals operate at the upstate membrane potential. Nat. Neurosci. 9, 459–461, author reply 461. Schonewille, M., Luo, C., Ruigrok, T.J., Voogd, J., Schmolesky, M.T., Rutteman, M., Hoebeek, F.E., De Jeu, M.T., and De Zeeuw, C.I. (2006b). Zonal organization of the mouse flocculus: physiology, input, and output. J. Comp. Neurol. 497, 670–682. Schonewille, M., Belmeguenai, A., Koekkoek, S.K., Houtman, S.H., Boele, H.J., van Beugen, B.J., Gao, Z., Badura, A., Ohtsuki, G., Amerika, W.E., et al. (2010). Purkinje cell-specific knockout of the protein phosphatase PP2B impairs potentiation and cerebellar motor learning. Neuron 67, 618–628. Simpson, J.I., Wylie, D.R., and De Zeeuw, C.I. (1996). On climbing fiber signals and their consequence(s). Behav. Brain Sci. 19, 380–394. Stahl, J.S., van Alphen, A.M., and De Zeeuw, C.I. (2000). A comparison of video and magnetic search coil recordings of mouse eye movements. J. Neurosci. Methods 99, 101–110. Szapiro, G., and Barbour, B. (2007). Multiple climbing fibers signal to molecular layer interneurons exclusively via glutamate spillover. Nat. Neurosci. 10, 735–742. Tamada, A., Kumada, T., Zhu, Y., Matsumoto, T., Hatanaka, Y., Muguruma, K., Chen, Z., Tanabe, Y., Torigoe, M., Yamauchi, K., et al. (2008). Crucial roles of

Robo proteins in midline crossing of cerebellofugal axons and lack of their up-regulation after midline crossing. Neural Dev. 3, 29. Tian, J.B., King, J.S., and Bishop, G.A. (2008). Stimulation of the inferior olivary complex alters the distribution of the type 1 corticotropin releasing factor receptor in the adult rat cerebellar cortex. Neuroscience 153, 308–317. Timmann, D. (2012). [Contribution of the cerebellum to cognition]. Fortschr. Neurol. Psychiatr. 80, 44–52. Van Alphen, A.M., Schepers, T., Luo, C., and De Zeeuw, C.I. (2002). Motor performance and motor learning in Lurcher mice. Ann. N Y Acad. Sci. 978, 413–424. van Vreeswijk, C., and Sompolinsky, H. (1996). Chaos in neuronal networks with balanced excitatory and inhibitory activity. Science 274, 1724–1726. Voogd, J., and Barmack, N.H. (2006). Oculomotor cerebellum. Prog. Brain Res. 151, 231–268. Welsh, J.P., Lang, E.J., Suglhara, I., and Llina´s, R. (1995). Dynamic organization of motor control within the olivocerebellar system. Nature 374, 453–457. Wulff, P., Schonewille, M., Renzi, M., Viltono, L., Sassoe`-Pognetto, M., Badura, A., Gao, Z., Hoebeek, F.E., van Dorp, S., Wisden, W., et al. (2009). Synaptic inhibition of Purkinje cells mediates consolidation of vestibulocerebellar motor learning. Nat. Neurosci. 12, 1042–1049. Yakhnitsa, V., and Barmack, N.H. (2006). Antiphasic Purkinje cell responses in mouse uvula-nodulus are sensitive to static roll-tilt and topographically organized. Neuroscience 143, 615–626. Yamada, M., Terao, M., Terashima, T., Fujiyama, T., Kawaguchi, Y., Nabeshima, Y., and Hoshino, M. (2007). Origin of climbing fiber neurons and their developmental dependence on Ptf1a. J. Neurosci. 27, 10924–10934.

Neuron 78, 700–713, May 22, 2013 ª2013 Elsevier Inc. 713