Cerebellar output neurons can impair non-motor behaviors by altering development of extracerebellar connectivity

Abstract

The capacity of the brain to compensate for insults during development depends on the type of cell loss, whereas the consequences of genetic mutations in the same neurons are difficult to predict. We reveal powerful compensation from outside the mouse cerebellum when the excitatory cerebellar output neurons are ablated embryonically and demonstrate that the main requirement for these neurons is for motor coordination and not basic learning and social behaviors. In contrast, loss of the homeobox transcription factors Engrailed1/2 (EN1/2) in the cerebellar excitatory lineage leads to additional deficits in adult learning and spatial working memory, despite half of the excitatory output neurons being intact. Diffusion MRI indicates increased thalamo-cortico-striatal connectivity in En1/2 mutants, showing that the remaining excitatory neurons lacking En1/2 exert adverse effects on extracerebellar circuits regulating motor learning and select non-motor behaviors. Thus, an absence of cerebellar output neurons is less disruptive than having cerebellar genetic mutations.

Fig. 1. Acute chemogenetic inhibition of the adult MedP eCN impairs reversal learning and not motor behaviors.

a Schematics (adapted from https://atlas.brain-map.org/) of a lateral sagittal mouse cerebellum. (left), vertical lines (i and ii) indicating anterior and posterior coronal schematics (right). b, c Coronal images of tdTomato and MEIS2 immunostaining in anterior (b) and posterior (c) CN of SepW1-Cre; Ai75D. CN was subdivided into five subregions (Paxinos and Franklin, 2007). MedA Anterior medial, MedP Posterior medial, IntA Anterior interposed, IntP Posterior interposed, Lat Lateral. Scale bars = 500 µm. d Quantification of tdTomato+ cells on every second coronal section of SepW1-Cre; Ai75D mice in the lateral CN (Lat) and subregions of the intermediate (Int) and medial (Med) CN (n = 4 per genotype). e Image of tdTomato (magenta) and MEIS2 (green) co-expressing eCN in SepW1-Cre; Ai75D mice. Scale bars = 50 µm. f Quantification of tdTomato+ cells that co-express MEIS2 and the reverse in SepW1-Cre; Ai75D mice (n = 4 mice). g Schematic (adapted from https://atlas.brain-map.org/) of viral injection to express mCherry (control) or hM4Di-mCherry (MedP-hM4Di) in adult MedP eCN. Dashed line indicates the region shown in (i). h Images of viral mCherry (red) in MedP eCN in control (top) and MedP-hM4Di (bottom) mice. Scale bars = 250 µm. i Images of MedP eCN mCherry+ axon terminals (black) in four thalamic nuclei of control mice. Fluorescent images were inverted using the look-up table in Fiji. MD mediodorsal, CL centrolateral, VM ventromedial, PF parafascicular. Scale bars = 250 µm. j Experimental timeline of surgery, CNO injection, and behavioral tests. k (left) Representative images of footprints from control and MedP-hM4Di mice. (right) quantification of stride, stance, and sway (n = 11 per group). l Total distance traveled during basal locomotion (n = 11 per group). m Latency to fall during the accelerating rotarod test (MedP-hM4Di: n = 11, control: n = 10). n Forelimb grip strength normalized to body weight (MedP-hM4Di: n = 11, control: n = 10). o Total number of correct trials during the water Y-maze test (MedP-hM4Di: n = 10, control: n = 9). p Percentage spontaneous alternations in the Y-maze (MedP-hM4Di: n = 10, control: n = 9). Chance level performance is 50% (dotted line). ns, not significant: P ≥ 0.05. Data are presented as mean values ± SEM. Detailed statistics are provided in Supplementary Table 6. Source data are provided as a Source Data file.

Fig. 2. Generation of mice lacking the MedP eCN by conditional knockout of En1/2.

a Quantification of eCN number (large (100–600 µm²) NeuN+ cells) along the medial-lateral axis in adult SepW1-En1/2 CKOs (n = 5) and littermate controls (n = 6). Med medial, Int interposed, Lat lateral. b Coronal images of MEIS2 labeling in the CN. Scale bars = 500 µm. c Images of H&E labeled sagittal sections of vermis (Ver), paravermis (pVer), and hemisphere (Hemi). Scale bars = 1 mm. d–l Quantification of three cerebellar (CB) areas (d), molecular layer (ML) area as a percent of total CB area (e), internal granule cell layer (IGL) area as a percent of total CB area (f), Purkinje cell (PC) density (g), PV+ ML interneuron (MLI) density (h), granule cell (GC) density in the IGL (i), estimated ratio of the number of PCs to PV+ MLIs (j), estimated ratio of the number of PCs to GCs (k), estimated ratio of the number of PV+ MLIs to GCs (l) in SepW1-En1/2 CKOs (n = 5) and littermate controls (n = 6). ns, not significant: P ≥ 0.05. Data are presented as mean values ± SEM. Detailed statistics provided in Supplementary Table 6. Source data are provided as a Source Data file.

Fig. 3. Mice lacking MedP eCN have only minor motor behavioral deficits.

a (left) Representative images of footprints from one SepW1-En1/2 CKO and littermate control. (right) Quantification of stride, stance, and sway (SepW1-En1/2 CKOs: n = 27, littermate controls: n = 22). b Total distance traveled during basal locomotion (SepW1-En1/2 CKOs: n = 24, littermate controls: n = 27). c Number of missteps per trial during balance beam assay (SepW1-En1/2 CKOs: n = 9, littermate controls: n = 15). d Latency to fall during the accelerating rotarod test (SepW1-En1/2 CKOs: n = 23, littermate controls: n = 27). e Forelimb grip strength normalized to body weight (SepW1-En1/2 CKOs: n = 23, littermate controls: n = 27). f Total number of correct trials during the water Y-maze test (SepW1-En1/2 CKOs: n = 18, littermate controls: n = 23). g Percentage of spontaneous alternations in the Y-maze (SepW1-En1/2 CKOs: n = 23, littermate controls: n = 26). Chance level performance is 50% (dotted line). h Latency to right onto four paws at P7 (SepW1-En1/2 CKOs: n = 17, littermate controls: n = 13). i Latency to turn upward on a negative slope at P7 and P11 (SepW1-En1/2 CKOs: n = 17, littermate controls: n = 13). ns, not significant: P ≥ 0.05. Data are presented as mean values ± SEM. ns, not significant: P ≥ 0.05. Data are presented as mean values ± SEM. Detailed statistics provided in Supplementary Table 6. Source data are provided as a Source Data file.

Fig. 4. Generation of mice in which all embryonic eCN are ablated using Diphtheria toxin.

a Intersectional approach to pharmacogenetically ablate the embryonic eCN. A doxycycline (Dox)-controlled and recombinase-activated gene overexpression allele (DRAGON) for attenuated diphtheria toxin fragment A (DTA) (Igs7^DRAGON-DTA) combined with an Atoh1-tTA transgene and heterozygous En1^Cre knock-in allele results in embryonic killing of eCN when Dox is administered starting at E13.5 via expression of DTA. The genotypes of littermate controls are Atoh1-tTA/+ or En1^Cre/+ along with the lgs7^DRAGON-DTA/+ allele. b Images of E17.5 medial and lateral cerebellar sagittal sections stained for MEIS2 (white). Scale bars = 250 µm. c Quantification of MEIS2+ cells on every 10th sagittal section (n = 3 per genotype). d Mediolateral distribution of NeuN+ large cells (100–600 µm²)(eCN-DTA mice: n = 6, littermate controls: n = 5). Med medial, Int interposed, Lat lateral. e RNA in situ analysis of coronal sections for Slc17a6 (n = 3 per genotype). Images are single-channel inverted using a look-up table in Fiji. Scale bars = 500 µm; Scale bars for i–vi = 50 µm. f Images of triple RNA in situ of coronal sections in the medial CN showing some neurons co-express Slc32a1, Slc17a6, and Slc6a5. Arrowhead and asterisk indicate neurons expressing only Slc32a1 or Slc17a6 in controls, respectively (n = 3 per genotype). Scale bars = 50 µm. g Images of H&E labeled vermis (Ver), paravermis (pVer), and hemisphere (Hemi) sagittal sections. Scale bars = 1 mm. h–p Quantification of three cerebellar (CB) areas (h), molecular layer (ML) area as a percent of total CB area (i), internal granule cell layer (IGL) area as a percent of total CB area (j), Purkinje cell (PC) density (k), PV+ ML interneuron (MLI) density in (l), granule cell (GC) density in the IGL (m), estimated ratio of the number of PCs to PV+ MLIs (n), estimated ratio of the number of PCs to GCs (o), estimated ratio of the number of PV+ MLIs to GCs (p) in eCN-DTA mice (n = 5) and littermate controls (n = 6). ns, not significant: P ≥ 0.05. Data are presented as mean values ± SEM. Detailed statistics provided in Supplementary Table 6. Source data are provided as a Source Data file.

Fig. 5. Loss of all eCN impairs motor coordination and WYM performance, but not motor learning and additional non-motor behaviors.

a Latency to right onto four paws at P7 (eCN-DTA mice: n = 15, littermate controls: n = 43). b Latency to turn upward on a negative slope at P7 and P11 (eCN-DTA mice: n = 15, littermate controls: n = 43). c (left) Representative images of footprints from one eCN-DTA and littermate control. (right) Quantification of stride, stance, and sway (eCN-DTA mice: n = 16, littermate controls: n = 15). d Total distance traveled during basal locomotion (eCN-DTA mice: n = 16, littermate controls: n = 15). e Number of missteps per trial during balance beam assay (eCN-DTA mice: n = 7, littermate controls: n = 17). f Latency to fall in the accelerating rotarod test (eCN-DTA mice: n = 15; littermate controls: n = 15). g Forelimb grip strength normalized to body weight (eCN-DTA mice: n = 16, littermate controls: n = 15). h Total number of correct trials during the water Y-maze test (eCN-DTA mice: n = 15, littermate controls: n = 12). i Percentage of spontaneous alternations in the Y-maze (eCN-DTA mice: n = 15, littermate controls: n = 14). Chance level performance is 50% (dotted line). j Percentage of spontaneous alternations in the plus-maze (eCN-DTA mice: n = 16, littermate control: n = 15). Chance level performance is 22.2% (dotted line). k Social preference (percent time nose spent within novel mouse contact zone) during the three-chamber social approach test (eCN-DTA mice: n = 16, littermate controls: n = 15). l Percentage of time spent in the open arms of an elevated plus-maze (eCN-DTA mice: n = 16, littermate controls: n = 14). m Total time spent self-grooming (eCN-DTA mice: n = 16, littermate controls: n = 15). ns not significant: P ≥ 0.05. Data are presented as mean values ± SEM. Detailed statistics provided in Supplementary Table 6. Source data are provided as a Source Data file.

Fig. 6. Loss of En1/2 in all eCN impairs motor coordination and learning, cognitive flexibility, and spatial working memory.

a Latency to right onto four paws at P7 (Atoh1-En1/2 CKOs: n = 19, littermate controls: n = 22). b Latency to turn upward on a negative slope at P7 and P11 (Atoh1-En1/2 CKOs: n = 19, littermate controls: n = 22). c (left) Representative images of footprints from one Atoh1-En1/2 CKO and littermate control. (right) Quantification of stride, stance, and sway (Atoh1-En1/2 CKOs: n = 28, littermate controls: n = 30). d Total distance traveled during basal locomotion (Atoh1-En1/2 CKOs: n = 33, littermate controls: n = 35). e Number of missteps per trial during balance beam assay (Atoh1-En1/2 CKOs: n = 10, littermate controls: n = 10). f Latency to fall in the accelerating rotarod test (Atoh1-En1/2 CKOs: n = 32, littermate controls: n = 30). g Forelimb grip strength normalized to body weight (Atoh1-En1/2 CKOs: n = 32, littermate controls: n = 30). h Total number of correct trials during the water Y-maze test (Atoh1-En1/2 CKOs: n = 31, littermate controls: n = 35). i Percentage of spontaneous alternations in the Y-maze (n = 35 per genotype). Chance level performance is 50% (dotted line). j Percentage of spontaneous alternations in the plus-maze (n = 22 per genotype). Chance level performance is 22.2% (dotted line). k Social preference (percent time nose spent within novel mouse contact zone) during the three-chamber social approach test (Atoh1-En1/2 CKOs: n = 38, littermate controls: n = 40). l Percentage of time spent in the open arms of an elevated plus-maze (Atoh1-En1/2 CKOs: n = 45, littermate control: n = 46). m Total time spent self-grooming (n = 34 per genotype). ns not significant: P ≥ 0.05. Data are presented as mean values ± SEM. Detailed statistics provided in Supplementary Table 6. Source data are provided as a Source Data file.

Fig. 7. Atoh1-En1/2 CKOs have reduced cerebellothalamic projections, but no ectopic cerebellothalamic projections.

a Schematic (https://atlas.brain-map.org/) of anterograde tracing of MedA CN. b Images of coronal sections showing injection site and mCherry+ axon terminals (brown) in various thalamic regions of an Atoh1-En1/2 CKO and control (n = 3 per genotype). Scale bar: injection site = 1 mm and mCherry images = 250 µm. c Summary of mCherry+ axon terminals observed in thalamic nuclei of Atoh1-En1/2 CKOs (blue) versus controls (dark gray) on three representative coronal planes (adapted from https://atlas.brain-map.org/). Blue indicates reduced density. d Schematic (https://atlas.brain-map.org/) of anterograde tracing of IntA CN in adult Atoh1-En1/2 CKOs and controls. e, Images showing injection site and biotinylated dextran amine (BDA)+ axon terminals (brown) in thalamic regions of an Atoh1-En1/2 CKO and control (n = 3 per genotype). Scale bar: injection site = 1 mm and BDA images = 250 µm. f Summary of BDA+ axon terminals observed in thalamic nuclei of Atoh1-En1/2 CKOs (blue) and controls (dark gray) on three representative coronal planes (adapted from https://atlas.brain-map.org/). Blue indicates reduced density. g Schematic (https://atlas.brain-map.org/) of retrograde tracing in adult Atoh1-En1/2 CKOs and controls. h, i Images of the injection site of Fluoro-Ruby (red) and Hoechst (blue) in centrolateral thalamus (CL, h Atoh1-En1/2 CKO n = 7 and control n = 5) and parafascicular thalamus (PF, i Atoh1-En1/2 CKO: n = 4 and control: n = 5). Scale bars = 1 mm. j Quantification of Fluoro-Ruby+ cells in CN subregions retrogradely labeled from CL. k Quantification of Fluoro-Ruby+ cells in CN subregions that are retrogradely labeled from PF injection. AD Anterodorsal nucleus, AM Anteromedial nucleus, AV Anteroventral nucleus of thalamus, CL Central lateral nucleus, CM Central medial nucleus, IAD Interanterodorsal nucleus, IAM Interanteromedial nucleus, IMD Intermediodorsal nucleus, LD Lateral dorsal nucleus of thalamus, LGv Ventral part of the lateral geniculate complex, LP Lateral posterior nucleus, MD Mediodorsal nucleus of thalamus, PCN Paracentral nucleus, PF Parafascicular nucleus, PO Posterior complex, PT Parataenial nucleus, PVT Paraventricular nucleus, RE Nucleus of reuniens, RH Rhomboid nucleus, RT Reticular nucleus, SMT Submedial nucleus, SPFm Subparafascicular nucleus, magnocellular part, VAL Ventral anterior-lateral complex, VM Ventral medial nucleus, LGd Dorsal part of the lateral geniculate complex, VPM Ventral posteromedial nucleus, VPL Ventral posterolateral nucleus, SPA Subparafascicular area, VPMpc Ventral posteromedial nucleus, parvicellular part, VPLpc Ventral posterolateral nucleus, parvicellular part. ns not significant: P ≥ 0.05. Data are presented as mean values ± SEM. Detailed statistics provided in Supplementary Table 6. Source data are provided as a Source Data file.

Fig. 8. Diffusion MRI shows Atoh1-En1/2 CKOs have connectivity changes outside the cerebellum that are distinct from eCN-DTA mice.

a, b Quantification of cerebellar (CB) volume (a) and regional volumes normalized to forebrain plus midbrain combined volume (b) in Atoh1-En1/2 CKOs (n = 12) compared to littermate controls (n = 13). c Schematic representation of global connectivity in Atoh1-En1/2 CKOs compared to littermate controls. Black lines indicate no significant difference, red lines indicate reduced connectivity in Atoh1-En1/2 CKOs and blue lines indicate increased connectivity in Atoh1-En1/2 CKOs compared to littermate controls. d, e Quantification of global efficiency (Geff, d) and small-worldness (SW, e) in Atoh1-En1/2 CKOs (n = 12) compared to littermate controls (n = 13). f–j Representative images of ILM-SS (f), ILM-MO (g), ILM-DS (h), SS-DS (i), MO-DS (j) tractographies in the right hemisphere of one Atoh1-En1/2 CKO and littermate control. Target regions are outlined in dotted lines. The color of streamlines indicates their orientations, as indicated by the colored arrows on the right. AP anteroposterior, ML mediolateral, DV dorsoventral. k–o Quantification of average (left plus right hemispheres) of ILM-SS tractography (k), ILM-MO tractography (I), ILM-DS tractography (m), SS-DS tractography (n), and MO-DS tractography (o) in Atoh1-En1/2 CKOs compared to littermate controls (Atoh1-En1/2 CKOs: n = 12, littermate controls: n = 13). p, q Quantification of cerebellar (CB) volume (p) and regional volumes normalized to forebrain plus midbrain combined volume (q) in eCN-DTA mice compared to littermate controls (n = 5 per genotype). r Schematic representation of global connectivity for eCN-DTA mice compared to littermate controls. Black lines indicate no significant difference and red lines indicate reduced connectivity in eCN-DTA mice compared to littermate controls. s, t Quantification of global efficiency (Geff, d) and small-worldness (SW, e) in eCN-DTA mice compared to littermate controls (n = 5 per genotype). u–y, Quantification of average (left and right hemispheres) ILM-SS tractography (u), ILM-SS tractography (v), ILM-SS tractography (w), ILM-SS tractography (x), and ILM-SS tractography (y) in eCN-DTA mice compared to littermate controls (n = 5 per genotype). CTX cerebral cortex, OLF olfactory bulb, HPF hippocampal formation, AMY amygdala, STR striatum, PAL pallidum, TH thalamus, HY hypothalamus, MB midbrain, HB hindbrain, CB cerebellum, ILM intralaminar nuclei, SS primary somatosensory cortex, MO primary motor cortex, DS dorsal striatum. ns, not significant: P ≥ 0.05. Data are presented as mean values ± SD for a, j, and mean value ± SEM for d, e, k–o, s–y. Detailed statistics provided in Supplementary Table 6. Source data are provided as a Source Data file.

Fig. 9. Cerebellar size, behavioral phenotypes, and diffusion MRI changes after adult or developmental manipulation of eCN.

Red indicates an impairment/alteration in the behavior of mutants compared to littermate controls, gray indicates no change, and white indicates that the behavior was not tested. Med medial, Int interposed, Lat lateral, Ver vermis, pVer paravermis, Hemi hemispheres, PC Purkinje cells, GC granule cells, MLI molecular layer interneurons, Geff global efficiency, SW small-worldness.