Loss of YY1 in Purkinje cells triggers cerebellar ataxia via Wnt10b-mediated impairment of dendrite development

Summary

The cerebellum relies on Purkinje cells (PCs) to coordinate motor functions and maintain neural circuitry. Impaired PC dendritogenesis is linked to neurological disorders, yet intrinsic molecular mechanisms remain unclear. We demonstrate that postnatal deletion of Yin Yang 1 (Yy1) in PCs disrupts dendrite and spine formation, impairs synaptic plasticity, and alters protein kinase C gamma (PKCγ) expression, leading to cerebellar ataxia and progressive atrophy. Mechanistically, YY1 transcriptionally activates wingless-type MMTV integration site family, member 10B (Wnt10b), which signals through Frizzled-4 (Fzd4) in PCs. Wnt10b downregulation mimics the dendritic defects seen in Yy1 mutants, indicating its crucial role in dendrite growth and synaptic environment regulation. These changes result in defective synapse innervation. Our findings reveal that YY1 governs PC dendritogenesis and synaptogenesis via Wnt10b, while also modulating PKCγ and β-catenin signaling. The YY1-Wnt10b axis emerges as a critical pathway underlying YY1-associated cerebellar dysfunction.

Graphical abstract

Highlights

• •YY1 loss in PCs causes cerebellar ataxia and atrophy
• •YY1 is crucial for PC dendrite and spine development
• •YY1 activates Wnt10b, forming a key signaling axis in postnatal PCs
• •YY1-Wnt10b-Fzd4 axis regulates postnatal PC development and function

Cell biology; Molecular biology; Neuroscience

Introduction

The cerebellum serves as a crucial neural center that coordinates motor functions and contributes to non-motor processes, including social behavior, language processing, and cognitive functions.¹ Purkinje cells (PCs), the primary output neurons of the cerebellar cortex, function as essential signal integration centers by processing information from two distinct cerebellar circuits: mossy fiber inputs via granule cell (GC) parallel fiber (PF) and climbing fiber (CF) inputs from the lower medulla.² In addition to these excitatory inputs, PCs receive strong inhibitory modulation from molecular layer (ML) interneurons, which are themselves innervated by PFs and regulate PC simple spike activity.^3,4 Through this integrated circuitry, PCs transmit inhibitory output to the cerebellar and vestibular nuclei. Consequently, impaired PC dendritogenesis and synaptogenesis can lead to cerebellar circuit dysfunction, contributing to various neurodegenerative diseases, especially cerebellar ataxia.^5,6,7

PC development in mice occurs predominantly during the first three postnatal weeks, a critical period for dendrite growth and synapse formation. During postnatal dendritic development, an apical cone emerges from the apical pole of the PC soma, developing into a primary dendrite that undergoes rapid elongation and expansion to form an extensive branched dendrite tree.⁸ The proper development of PC dendrites and their synaptic function is crucial, as their disruption can lead to motor incoordination and various neurological disorders. Elucidating the cellular and molecular mechanisms of PC dendritogenesis and synaptogenesis is thus critical for understanding the molecular basis of cerebellar diseases.

The mechanism governing postnatal PC dendrite growth is particularly complex in vivo and remains poorly understood, especially regarding intrinsic versus extrinsic determinants.⁹ Different extrinsic factors, including neuronal interactions, neurotrophic factors, and chemotropic signals, collectively shape the PC dendritic architecture,^8,9,10 while fewer intrinsic determinants have been identified. Our previous work identified Lhx1/5 as key transcription factors regulating PC dendritic development through control of Espin (Espn) expression which in turn modulates F-actin localization in PC dendrites and spines,¹¹ demonstrating how a transcription factor can orchestrate complex dendritic organization.

Yin Yang 1 (YY1), a ubiquitously expressed transcription factor, exhibits diverse developmental roles. Studies in mice have revealed that Yy1 performs region- and stage-specific functions essential for embryonic cortical development,¹² astrocyte maturation,¹³ and neuroepithelial cell-cycle progression.¹⁴ YY1 contributes to synaptic vesicle recycling through regulation of dynamin I^15,16 and promotes late neuronal differentiation via dopamine β-hydroxylase expression in noradrenergic neurons.¹⁷ More recently, YY1 has been shown to cooperate with CCCTC-binding factor (CTCF) to enhance axon regeneration following neuronal injury.¹⁸ In humans, deletion or missense mutations in YY1 cause Gabriele-de Vries syndrome, characterized by developmental delays, facial dysmorphism, and intrauterine growth restriction, with affected individuals frequently presenting neurological manifestations, such as gait abnormalities, hypotonia, dystonia, and cerebellar ataxia,^19,20,21 though the underlying pathogenic mechanisms remain enigmatic.

This study investigated the role and mechanistic basis of YY1 in postnatal PC development, as well as the downstream signaling pathway responsible for these regulatory effects on cerebellar function. Our findings reveal that Yy1 deletion in postnatal PCs leads to multiple cytoskeletal defects, compromising dendritogenesis and synaptic plasticity. Mechanistically, YY1 regulates Wnt10b expression in PCs, which is responsible for dendrite growth and the regulation of synaptic strength through its control of PKCγ and β-catenin expression. These findings establish a YY1-Wnt10b molecular axis controlling PC dendrite development and functional synaptic innervation.

Results

Postnatal mice lacking the transcription factor YY1 in PCs exhibit cerebellar atrophy and early signs of ataxia

YY1 is expressed ubiquitously in the ML, Purkinje cell layer (PCL), and granular layer (GL) of the postnatal mouse cerebellum (Figure 1A). To investigate the functional roles of YY1 in mammalian postnatal PCs, we employed the Cre/LoxP system to conditionally inactivate Yy1 in postnatal PCs by crossing mice with the Yy1-flox (Yy1^fx) allele (Figure S1A) with Pcp2-Cre (L7-Cre) transgenic mice, which specifically express Cre recombinase in PCs starting from postnatal day (P) 2.¹¹ Immunohistochemistry confirmed the removal of YY1 expression exclusively in the PCs of cKO mutants (Pcp2 ^Cre/+; Yy1 ^fx/fx), while it remained in the PCs of the controls (Pcp2 ^Cre/+; Yy1^fx/+ or Yy1^fx/fx) (Figure 1A). The removal process was nearly complete by P8 and fully complete by P21 (Figure 1A), indicating successful inactivation of Yy1 in the mutant PCs.

Figure 1.

Yy1 cKO mice displayed progressive cerebellar atrophy, motor deficits, and defective PC dendritogenesis

(A) Immunostaining analysis of YY1 (red) with Purkinje cell (PC) marker Calbindin (green) confirming successful inactivation of YY1 in postnatal day (P) 8 and P21 cKO PCs. White arrows indicate control PC nuclei with normal YY1 expression. Scale bars: 50 μm.

(B) Representative figures of P28 control and cKO whole-brain top and mid-sagittal. Scale bars: 1 mm.

(C) Hematoxylin and eosin (H&E) staining showing cerebellar cytoarchitecture and foliation in P28 control and cKO. Scale bars: 500 μm.

(D and E) Bar graphs showing reduced cerebellar vermis size (cm³) and reduced cerebellar vermis perimeter (cm) in cKO mutants at P28.

(F) Bar graph showing statistical quantification of molecular layer (ML) thickness reduction in P14 and P21 cKO mutants.

(G) Bar graphs showing statistical quantification of mid-sagittal vermis surface area (left) and dorsal surface area (right) in P21, 1-month-old (1M), and 2-month-old (2M) mice.

(H) Area-fill plot showing a significant relative cerebellum size reduction over time in cKO mutants compared with controls, with mid-sagittal (tan) and dorsal area (blue) components shown separately.

(I) Representative footprint patterns of control and Yy1 cKO 1M littermates. Front paws were painted red, and hind paws were painted dark blue. The blue left brackets and orange right brackets show representative lengths of hind paw stride and front paw stride, respectively.

(J) Bar graph showing front and hind paw stride length (cm) with left (L) and right (R) in control without Cre, control with Cre, and Yy1 cKO mice, demonstrating significantly reduced stride length in cKO mutants.

(K and L) Scatterplots showing increased time required to complete crossing an elevated balance beam (traveling time, seconds) and increased number of missteps along the balance beam in cKO mutants compared with controls.

(M and N) Scatterplots showing reduced latency time before falling from both a stationary rotarod at 4 rpm, an accelerating rotarod (4–40 rpm) in 5 min (left) and 2 min (right) in cKO mutants.

Filled circles represent male mice; open circles represent female mice.

(O) Representative immunostaining figures showing PC marker InsP3R-I (red) in P28 control and cKO cerebellar sections. The upper panels show reduced dendritic arborization in cKO PCs (scale bars: 100 μm), while the lower panels highlight distal dendritic spine structures indicated by arrowheads, showing diminished spine density in cKO PCs (scale bars: 10 μm).

(P) High-magnification images of Golgi-impregnated P28 PCs showing the morphology and arrangement of dendritic spines in controls and cKO mutants. Upper part of the image shows the overall PC dendritic tree (scale bars: 50 μm); lower part of the image shows high-magnification spine morphology (scale bars: 1 μm).

(Q) Bar graph showing shorter overall PC dendrite length (μm) in cKO mutants.

(R) Bar graph showing the average number of spines per 10 μm on control and cKO PC dendrites, indicating lower spine density in cKO PCs.

(S) Bar graph showing the average number per 10 μm of each spine type (thin, stubby, and mushroom) in control and cKO PCs, demonstrating a selective reduction in mushroom-type spines in cKO mutants.

For all scatter and bar plots, bars indicate the mean values, and each point represents a mouse cerebellum. rpm, revolutions per minute; ML, molecular layer; PCL, Purkinje cell layer. For each measurement, at least three mice were analyzed per group; t test; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. Data are represented as mean ± SEM.

The P28 cerebellum of the cKO mutants exhibited normal foliation and cytoarchitecture when compared with the controls (Figure 1B). However, a significant decrease in the size of the mutant cerebellum was observed in the vermis (Figures 1C–1E) with comparable cerebellum length, internal granular layer (IGL) thickness, and PC density to controls (Figures S1B–S1D). We then examined cerebellum sizes at various developmental stages and over time. The thickness of the ML, occupied by the PC dendritic tree, was slightly reduced in P14 cKO and further reduced by 43% in P21 cKO mice (Figure 1F). Moreover, P21 cKO cerebella showed about a 25% reduction in mid-sagittal area and 32% reduction in dorsal area, while 2-month-old (M) cKO cerebella exhibited a 75% reduction in mid-sagittal and 40% reduction in dorsal area compared with control groups (Figures 1G and 1H). Impaired postnatal cerebellum development driven by Yy1 ablation in PCs further promoted cerebellar atrophy, with cerebellum size reduction becoming more significant over time. This implies that Yy1 is a key regulator of PC maturation, and Yy1 ablation in PCs not only impacts cerebellar development but also contributes to cerebellar degeneration.

The cKO mutants displayed noticeable ataxic locomotion beginning on P28 (Videos S1). To quantitatively analyze their motor abilities, we subjected P28 mice to classic motor tasks, including gait analysis, elevated balance beam, and rotarod tests. In gait analysis, cKO mice had shorter steps, altered stride distance, stride width, and reversed front-hind paw overlap (Figures 1I, 1J, and S1E–S1G). In balance beam analysis, cKO mice had longer times crossing the beam (Figure 1K) and a significant increase in the number of missed steps (Figure 1L). Although cKO mice were able to complete the test, they struggled to climb along the beam, slipping at nearly every step. In the rotarod test, cKO mice performed significantly worse than the controls on both the stationary and accelerating rod (Figures 1M and 1N). Comparable motor performance between male and female cKO mice suggests that the defects are not gender dependent. These results indicate that Yy1 is required in PCs for motor coordination, with Yy1 cKO mice showing ataxic gait characteristic of cerebellar dysfunction.

Yy1 is required for PC dendrite development and spine maturation

To examine cKO PCs in detail, we visualized dendrites and dendritic spines by labeling PCs with InsP3R-I (Figure 1O) and performed Golgi-impregnation staining on P28 mouse cerebella (Figure 1P). InsP3R-I staining revealed reduced dendritic arborization in cKO PCs, highlighting that PCs lacking YY1 exhibit visibly diminished branching compared with controls. Higher magnifications further showed marked loss of distal dendritic spine density,²² indicating impaired spine formation and dendrite development. Dendritic length and branching order were systematically quantified in individual PCs. Mutant PCs exhibited significantly shorter proximal dendritic segments at the 1^st and 2^nd branch orders, while higher-order branches (≥3^rd order) remained comparable to controls. Notably, mutant PCs failed to generate extensively branched dendrites beyond the 26^th order (Figure S2A). Also, mutant PCs had a significant reduction in PC length (measured from the cell body to the dendritic tree) (Figure 1Q), underscoring that Yy1 deletion constrains dendritic arbor complexity and limits the elaboration of the PC dendritic tree in mice. We then examined the spine density and morphology in the hypoplastic dendritic trees of cKO PCs. Dendritic spines are categorized as thin, stubby, or mushroom (refined into mature spines).²³ High-magnification images of Golgi-impregnated P28 control and cKO cerebella were obtained and spines in the secondary branches of apical dendrites were measured for quantification. Strikingly, a significant reduction in overall dendritic spine density was found in cKO PCs (Figure 1R). At this stage, mature spine density was markedly reduced in Yy1 cKO dendrites (Figure 1S), as also suggested by fewer InsP3R-I positive spines observed in cKO PCs (Figure 1O). These findings indicate that PCs exhibit less mature spine formation following Yy1 ablation.

Cytoskeletal proteins, F-actin and MAP2, are essential for establishing spine and dendrite formation and cell body architecture.^11,24,25 Phalloidin immunostaining revealed F-actin aggregation in cKO PC primary dendrites and somatic periphery, with comparable distribution in dendritic shafts between genotypes. MAP2 was similarly mislocalized to the soma and primary dendrites in cKO PCs. Additionally, actin regulatory proteins ESPN, IRSp53, and ARPC3 were expressed at reduced levels in P14 cKO PCs. These findings indicate that Yy1 inactivation in PCs leads to mislocalization of multiple cytoskeletal proteins within the somatic and dendritic regions, thereby contributing to impaired dendritogenesis and spine formation (Figures S2B–S2G).

YY1 regulates PC-specific Wnt10b expression and Fzd4 receptor signaling

To elucidate the molecular mechanisms underlying ataxic pathogenesis, we aimed to identify YY1 downstream genes influencing the growth of PC dendrites and spines during postnatal development. We compared transcriptomes of P14 cerebella from controls and Yy1 cKO mutants via RNA sequencing (RNA-seq). RNA-seq quality control metrics confirmed adequate sample separation and reproducibility (Figures S3A and S3B). Differential expression analysis identified 231 genes being upregulated and 723 genes being downregulated (Data S1).

To predict YY1 transcription targets, we performed iRegulon analysis to identify differentially expressed genes (DEGs) that potentially contain the YY1 motif.²⁶ We then cross-referenced these genes with the Allen Brain Atlas in situ hybridization (ISH) database to identify DEGs exclusive to P14 PCs (Figure 2A). This analysis identified 13 DEGs with 2 significantly upregulated genes and 6 significantly downregulated genes (Figure 2B). Transcript levels of these potential YY1 downstream genes were validated by quantitative reverse-transcription PCR (RT-qPCR), confirming the significant upregulation of Pappalysin-1 (Pappa, encoding a metalloprotease) and downregulation of Wnt10b in cKO cerebella (Figure 2C).

Figure 2.

YY1 transcriptionally activates Wnt10b, a transiently and specifically expressed in postnatal PCs and signals through Fzd4

(A) Schematic diagram of the transcriptomic filtering pipeline used to identify YY1 downstream targets in P14 PCs. RNA-seq identified 954 differentially expressed genes (DEGs) in Yy1 cKO cerebella, of which 133 contain a YY1-binding motif as predicted by iRegulon, and 13 of these are specifically expressed in postnatal PCs as confirmed by the Allen Developing Mouse Brain Atlas.

(B) Scatterplot (volcano-style) showing the fold change (Log₂) versus significance (−Log₁₀) of the 13 PC-specific YY1-target DEGs identified by the pipeline in (A). Wnt10b (blue) is among the most significantly downregulated targets, while Pappa (red) is the most significantly upregulated.

(C) Bar graph showing RT-qPCR validation of the relative mRNA expression levels of all 13 PC-specific DEGs in P14 control and Yy1 cKO cerebella. Wnt10b is significantly downregulated in cKO, confirming it as a key YY1 transcriptional target.

(D) In situ hybridization (ISH) showing Wnt10b mRNA expression in control and Yy1 cKO cerebellar sections at P12, P17, and P21. Control PCs display strong Wnt10b expression, whereas cKO PCs show markedly reduced signals. Scale bars: 200 μm (low magnification); 50 μm (insets, pink box).

(E) Wnt10b immunohistochemistry with Hematoxylin counterstaining in P14 control and cKO cerebellar sections. The Wnt10b chromogenic signal (indicated by white arrows) is reduced in cKO PCs. Scale bars: 50 μm.

(F) Schematic illustration showing that Wnt10b expression is transient and PC-specific during postnatal cerebellar development, with peak expression at P17 (immature stage) declining by P21 (mature stage).

(G) Left: schematic illustration of the Wnt10b luciferase promoter constructs, showing the conserved YY1-binding motif (atccattttctc, red) within the 451 bp promoter region (pGL3-Wnt10b-Pro) and the deletion construct lacking the YY1 motif (pGL3-ΔWnt10b-Pro, 439 bp). Right: luciferase reporter assay results in SH-SY5Y cells co-transfected with each construct and either YY1 overexpression vector (+YY1) or negative control (+NC), showing that YY1 significantly activates the Wnt10b promoter in a motif-dependent manner. (H) Heatmap showing RT-qPCR results of six Frizzled (Fzd) receptor family members (Fzd1, Fzd2, Fzd3, Fzd4, Fzd6, and Fzd7) in P14 control and Yy1 cKO cerebella. Fzd1 and Fzd4 are significantly downregulated in cKO, while Fzd7 is significantly upregulated.

(I) Immunostaining analysis of Fzd4 (red) and Hoechst (blue) showed low expression levels following YY1 deletion in PCs. Scale bars: 50 μm.

(J) Top: schematic of the Fzd4 promoter showing three chromatin immunoprecipitation (ChIP)-PCR amplicon regions (a: 85 bp, b: 217 bp, c: 121 bp). Bottom: ChIP-PCR results showing in vivo binding of YY1 to b and c regions of the Fzd4 promoter in P14 mouse cerebella. Input (5%), YY1 IP, and mock (IgG) lanes are shown. The ChIP experiment was repeated three times with n = 3 mouse cerebella pooled per analysis.

(K) Representative results of Fzd4 western blot analysis of immunoprecipitated protein using anti-Wnt10b antibody or anti-IgG (mock) in SH-SY5Y cells, demonstrating that Wnt10b physically binds to Fzd4 in vitro (55 kDa band). Input lane confirms protein loading. ML, molecular layer; PCL, Purkinje cell layer; GL, granular layer. For each measurement, at least three mice were analyzed per group; t test; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. Data are represented as mean ± SEM.

Previous studies established Wnt ligands as essential regulators of early cerebellar development²⁷ and implicated them in spinocerebellar ataxia (SCA) pathogenesis.²⁸ We therefore focused on the gene Wnt10b, which encodes a member of the Wnt ligand gene family that activates the highly conserved canonical Wnt/β-catenin signaling cascade.²⁹ Wnt10b mRNA was highly enriched in P14 PCs, as confirmed by Allen Brain Atlas ISH data and our ISH (Figures 2D, S3C, and S3D). Consistently, Yy1 cKO PCs showed marked reduction in Wnt10b protein levels (Figures 2E, S3E, and S3F). Notably, by P21 Wnt10b transcript and protein shifted to the GL from PCL (Figures 2A and S3E). This transient, PC-specific expression pattern (Figure 2F) identifies Wnt10b as a potential candidate mediator of postnatal PC development during a critical developmental window.

To determine whether YY1 directly regulates Wnt10b, dual-luciferase promoter assay was performed by co-transfecting the Yy1 overexpression vector with a luciferase reporter plasmid harboring the Wnt10b promoter fragment containing 5′atccattttctc3′, a conserved putative potential YY1-binding site, predicted by UCSC Genome Browser (GRCm39) (Figure S3G) into SH-SY5Y cells. The luciferase activity showed a 0.5-fold increase for the construct containing the complete Wnt10b promoter fragment without YY1 overexpression (+NC), whereas luciferase activity significantly increased by 1.7-fold with YY1 overexpression (+YY1), indicating that YY1 induced the luciferase activity (Figure 2G). Furthermore, the luciferase activities of the mutated Wnt10b promoter fragment with the YY1-binding site deletion, both with and without YY1 overexpression, were comparable to two negative controls (empty luciferase construct and no construct), further indicating that YY1 likely binds to the 5′atccattttctc3′ site in the Wnt10b promoter fragment to induce luciferase activity (Figure 2G). To demonstrate the in vivo binding of YY1 to the Wnt10b promoter, we further performed chromatin immunoprecipitation (ChIP) with P14 wild-type cerebellum chromatin extracts. The Wnt10b promoter fragment was successfully immunoprecipitated in chromatin reacted with anti-YY1 (Figure S3H).

We next investigated Wnt10b binding partners, focusing on six frizzled (Fzd) receptors (Fzd1, 2, 3, 4, 6, and 7) expressed in P14 PCs.³⁰ RT-qPCR results revealed that Fzd1 and Fzd4 were downregulated, while Fzd7 was upregulated in Yy1 cKO mutants (Figure 2H). Since Fzd1 and Fzd4 are exclusively expressed in PCs (Figure S3I), as reported by the Allen Brain Atlas ISH database, we focused on investigating Fzd1 and Fzd4 further. Consistent with the RT-qPCR results, immunofluorescent analysis of Fzd1 and Fzd4 (Figures 2I and S3J) showed reduced protein levels in cKO PCs. The UCSC Genome Browser (GRCm39) suggested Fzd1 has two YY1-binding motifs, whereas Fzd4 has three YY1-binding motifs. ChIP-PCR confirmed that YY1 binds to both Fzd1 and Fzd4 gene regions (Figures 2J and S3K). Co-immunoprecipitation (coIP) using the SH-SY5Y cell system with Wnt10b overexpression, confirmed by semi-qPCR, revealed that Wnt10b interacts with Fzd4 but not with Fzd1 (Figure 2K, S3L, and S3M). These findings suggest that Wnt10b binds Fzd4 through an autocrine mechanism during postnatal PC dendrite development, and that Fzd4 downregulation may critically impair Wnt10b signaling pathway during postnatal PC dendritogenesis.

Wnt10b positively regulates PC dendrite growth

Next, we investigated the functional role of Wnt10b in late-stage postnatal development of PCs using cerebellar organotypic cultures (Figures 3A and 3F). siRNA against Wnt10b or non-targeting (NT) controls was delivered into P12 cerebellar slices, a stage characterized by robust Wnt10b expression and the onset of active dendritic expansion in PCs. Both NT and Wnt10b siRNA cultures displayed transparent edges, indicating healthy slice conditions; however, Wnt10b knockdown (KD) slices exhibited smaller mid-sagittal vermis areas compared with NT controls (Figure S4A). Following siRNA transfection, Wnt10b mRNA levels were reduced by ∼60% after three days and protein expression was repressed (Figures 3B, S4B, and S4C). Quantification of dendritic morphology revealed that the longest dendrite length was markedly decreased in Wnt10b KD PCs (Figures 3C and S4D). Consistently, ML thickness was reduced by 14.28% in Wnt10b KD slices compared with NT controls, similar to the 11.25% reduction observed in Yy1 cKO cerebella (Figures 3D and S4D). Immunostaining for PKCγ and InsP3R-I further demonstrated diminished dendritic arborization and reduced PKCγ expression in KD PCs (Figure 3E). These findings indicate that Wnt10b is required for dendritic growth and ML expansion in postnatal PCs.

Figure 3.

Wnt10b is required for PC dendrite development

(A) Experimental scheme of Wnt10b siRNA knockdown (Wnt10b KD) approach in organotypic culture.

(B) Representative images of Wnt10b immunofluorescent analysis with PC labeling by Calbindin in Wnt10b KD approach demonstrating Wnt10b siRNA successfully knocked down Wnt10b expression shown by the Wnt10b immunoactivity reduction in Wnt10b siRNA treated samples compared with NT siRNA control samples. Scale bar: 200 μm.

(C) A scatterplot showing the longest PC dendrite length was reduced in KD groups. Values in the brackets indicate the number of PCs analyzed.

(D) A scatterplot showing the molecular layer thickness in lobule V was reduced following Wnt10b KD ex vivo and genetic ablation of Yy1 in vivo.

(E) Immunostaining of two PC markers, InsP3R-I (green) and PKCγ (red), in control and Wnt10 KD cerebellar sections suggests that PC dendrite growth was hindered with PKCγ expression reduction followed by Wnt10b downregulation. Scale bars: 100 μm.

(F) Experimental scheme of Wnt10b recombinant protein (recWnt10b) treatment in organotypic culture.

(G) A scatterplot showing the length of the longest dendrites of PCs with 0 mg/mL or 200 mg/mL recWnt10 delivery in control and cKO cerebellar section culture systems. The results showed that PC dendrite growth was promoted in both control and cKO mutants by recWnt10b treatment. Values in the brackets indicate the number of PCs analyzed.

(H and I) Immunostaining analysis of PKCγ showed that recWnt10b treatment enhances PKCγ expression in control PCs (H) but not in cKO mutant PCs (I). Nevertheless, recWnt10b promotes the towering growth of dendritic trees in both control and mutant PCs. Scale bars: 50 μm. Data are mean ± SEM; t test; ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

To test the sufficiency of Wnt10b, recombinant Wnt10b (recWnt10b) protein was delivered into P13 cerebellar slices from control and Yy1 cKO mice (Figures 3F and S4E). Quantification showed that recWnt10b treatment significantly increased dendrite length in both control and cKO PCs, with a 21.95% increase in controls and a 15.79% increase in cKO cerebellar group (Figure 3G). Immunostaining and western blot analysis revealed that recWnt10b enhanced PKCγ expression in control PCs, whereas PKCγ levels remained unchanged in cKO PCs despite increased dendritic growth (Figures 3H, 3I, and S4F). These findings suggest that Wnt10b promotes PC dendritogenesis and PKCγ expression, but the rescue effect on PKCγ is dependent on intact YY1 function, highlighting the YY1-Wnt10b axis in regulating PC dendrite development.

Wnt10b-knockdown PCs reveal proteome alterations at synaptic innervation

We then performed a proteomic study to evaluate the function and structure of cerebellar proteins following Wnt10b knockdown. By profiling proteins in NT and Wnt10b siRNA-transfected cerebellar samples via liquid chromatography-tandem mass spectrometry (LC-MS/MS), 4,273 proteins were identified, with 2,192 (51.3%) proteins quantified in all 18 samples (3 technical replicates of 3 NT control and 3 Wnt10b KD samples). Principal component analysis (PCA) and hierarchical clustering confirmed sample separation, and 142 differentially expressed proteins (DEPs) were identified (Figures S5A–S5C and Data S2). Analysis of DEPs identified 142 proteins that varied significantly between Wnt10b KD and NT controls. Of these, 56 proteins were upregulated, while 86 were downregulated in case samples relative to controls (Figure S5C and Data S2). Gene Ontology (GO) enrichment analysis using the Panther 18.0 classification system revealed an overrepresentation of certain biological process terms among the DEPs (Figure 4A). Our results indicated that the downregulation of Wnt10b in cerebellar slices substantially altered the synapse protein composition.

Figure 4.

Alteration of synapse protein proteome in Wnt10b KD cerebellar sections

(A) Gene Ontology enrichment analysis using the Panther 18.0 classification system revealed overrepresentation of certain biological process terms among the differentially expressed proteins (DEPs) in Wnt10b KD. It suggests Wnt10 downregulation in cerebellar sections altered proteome in pre- and post-synapses.

(B) Immunostaining of Calbindin (red) and VGluT1 (green) showed differences in PF-PC innervations between the P16 controls and the cKO mutants. Insets (1–4) show higher magnification of the PC middle (1–2) and proximal (3–4) dendritic tree regions in Yy1 cKO and its control. It suggests both middle and distal dendrites in cKO PCs had obvious VGluT1 immunopunta reduction compared with controls. Immunostaining of Calbindin (red) and VGluT1 (green) between controls and Wnt10b KD. VGluT1 distribution height along PC dendrites had shorter length in Wnt10b KD compared with controls. Scale bars: 25 μm.

(C) Immunostaining analysis of PKCγ showing less PKCγ in both Yy1 cKO and Wnt10b KD PC cytosol and dendrite. Scale bars: 50 μm.

We found that F-actin, which is essential for the PC spine structure, displayed significantly fewer immunopuncta in the mid-region of dendrites of both Yy1 cKO and Wnt10b KD PCs when compared with the corresponding control groups (Figure S5D). These findings suggest both the Yy1 cKO and Wnt10b KD groups exhibit a spine reduction, as indicated by the decreased F-actin in their dendrites. Given the impaired dendrite development and spine formation observed in Wnt10b KD PCs and to validate the GO analysis results, we further investigated PC synapse innervation in PF and CF.³¹ We examined VGluT1 immunoreactivity, a presynaptic marker for PF terminals. As expected, VGluT1 was abundantly present throughout the ML in control cerebella, with lower intensity observed in the middle and distal parts of the ML in cKO compared with control. This observation was also evident in Wnt10b KD groups compared with NT groups (Figure 4B). Additionally, there was notably reduced immunoreactivity of VGluT2, a presynaptic marker for CF terminals, along cKO but slightly reduced along Wnt10b KD PC dendrites compared with corresponding controls (Figure S5E).

Furthermore, we examined four PC-specific synaptic plasticity markers³² including PKCγ, PKCα, Synapsin-1, and GluRδ2. The PKCγ expression pattern was suppressed in Wnt10b KD, similar to what we observed in Yy1 cKO PCs (Figures 3E and 4C). Unexpectedly, PKCα, another PC-exclusive isoform of PKC, which was downregulated in Yy1 cKO PCs, was not downregulated by Wnt10b KD, while Synapsin-1 and GluRδ2 protein levels were similarly to Yy1 cKO (Figures S5F–S5I). Together, our results suggest that Wnt10b, regulated by YY1, is required for synaptic plasticity maintenance and late-phase PC dendrite development, with PKCγ being one of the signal transductions targets.

Yy1 loss and Wnt10b downregulation reduce β-catenin accumulation in PCs

Given the well-defined role of Wnt/β-catenin pathway in cerebellar circuit formation^33,34 and its involvement in long-term cerebellar health in SCA,²⁸ as well as previous studies demonstrating that Wnt10b specifically activates Wnt/β-catenin signaling, we sought to investigate whether Wnt/β-catenin signaling is involved in Yy1 cKO. We compared β-catenin subcellular distribution in control and cKO PCs. Immunofluorescence staining of β-catenin showed that β-catenin expression in cKO mutant PCs was markedly reduced, and this reduction was also observed in Wnt10b KD PCs (Figure 5A). Consistently, there was little to no observable immunoreactivity of active β-catenin (ABC) in the nuclei of cKO and Wnt10b KD PCs compared with the corresponding control groups (Figure 5B). The significant reduction in total β-catenin and ABC suggests that the β-catenin-Tcf/Lef transcriptional program in Wnt/β-catenin pathway may be suppressed following Yy1 ablation or Wnt10b KD.

Figure 5.

Wnt/β-catenin signaling is disrupted in both Yy1 cKO and Wnt10b KD, whereas CaMKIIβ, a Wnt/Ca2⁺ signaling transducer, reduced only in Yy1 cKO

(A) Immunostaining of β-catenin (green) in cerebellar sections from control, Yy1 cKO (in vivo), NT siRNA, and Wnt10b siRNA (ex vivo). Grayscale show magnified regions highlighting reduced β-catenin expression in cKO and KD PCs. White asterisks indicated the PC nucleus. White dotted line outlined the PC perimeters. Scale bars: 50 μm.

(B) Immunostaining of active β-catenin (green) and InsP3R-I (red) in the same conditions, showing reduced nuclear localization of active β-catenin in Yy1 cKO and Wnt10b KD PCs. Scale bars: 50 μm.

(C) RT-qPCR analysis of Wnt/β-catenin target genes Axin2, Ccnd1, and Myc. Left graph compares control and Yy1 cKO cerebella; right graph compares NT siRNA and Wnt10b KD cerebellar slice cultures. Ccnd1 and Myc had significantly downregulated following Yy1 ablation in PCs whereas Axin2 and Myc significantly downregulated following Wnt10b downregulation.

(D) Immunostaining of Calbindin (red) marked PCs and CaMKII (green) isoform α in the cerebellum sections from P28 controls and the Yy1 cKO suggested that less CaMKIIβ expressed in Yy1-ablated PC dendrites. Scale bars: 50 μm. Immunostaining of CaMKIIα in Wnt10b KD cerebellar samples shown that no obvious CaMKIIα. Scale bars: 25 μm.

(E) Immunostaining of Calbindin (red) marked PCs and CaMKII isoform β (green) in the cerebellum sections from P28 controls and the Yy1 cKO suggested that less CaMKIIβ expressed in PCL and GL. Scale bars: 50 μm. Immunostaining of CaMKIIβ in Wnt10b KD cerebellar samples shown that no obvious CaMKIIβ downregulation following Wnt10b KD. Scale bars: 25 μm.

ML, molecular layer; PCL, Purkinje cell layer; GL, granular layer. For each measurement, at least three mice were analyzed per group; t test; ∗p < 0.05, and ∗∗p < 0.01. Data are represented as mean ± SEM.

Quite unexpectedly, RT-qPCR analysis of Wnt/β-catenin pathway downstream genes (Axin2, Ccnd1, and Myc) revealed that Ccnd1 and Myc were slightly downregulated in Yy1 cKO cerebella. However, the reduction in these two genes was not as pronounced as the decrease in β-catenin levels observed in Yy1 cKO PCs (Figure 5C). Moreover, reductions in Axin2 and Myc were found in Wnt10b KD samples (Figure 5C). These findings show that β-catenin was suppressed in both Yy1 cKO and Wnt10b KD, but changes in canonical target genes do not fully recapitulate the signaling status, suggesting that additional YY1-regulated pathways may contribute to the pathogenesis.

We next investigated potential non-canonical contributions by examining whether Wnt/Ca²⁺ signaling, another potential neuronal Wnt signaling pathway,³⁵ participates in Yy1 cKO and Wnt10b KD. Previous studies have reported that the Wnt/Ca²⁺ signaling pathway can regulate dendritic spine size by modulating the actin cytoskeleton through the activation of Ca²⁺/calmodulin-dependent protein kinase II (CaMKII) and PKC.³⁶ CaMKIIα, which contributes to PC-PF synapse formation³⁷ and is exclusively expressed in PCs, had lower immunoreactivity in cKO PC dendrites and higher immunoreactivity in cKO PC cytoplasm compared with controls (Figure 5D). No detectable reduction of CaMKIIα was observed following Wnt10b KD (Figure 5D). Additionally, immunofluorescent staining of CaMKIIβ, an F-actin-binding partner,^38,39 in 1M cerebellar sections revealed that CaMKIIβ localized mainly in control PC cytosols and minimally in PC distal dendrites and the periphery of GCs (Figure 5E). Interestingly, CaMKIIβ immunoreactivity was nearly absent in cKO PCs and GCs (Figure 5E) and no detectable reduction of CaMKIIβ was observed following Wnt10b KD (Figure 5E). These findings support the idea that Wnt/Ca²⁺ signaling is suppressed following YY1 inactivation in PCs, and the dysregulation of Wnt/Ca²⁺ signaling potentially contributes to the ataxia phenotype in Yy1 cKO. Nevertheless, it appears that Wnt10b is likely independent of this signaling pathway but specifically regulates PKCγ in postnatal PCs. This pattern may reflect crosstalk between canonical Wnt/β-catenin signaling and YY1-regulated non-canonical pathways, such that β-catenin suppression does not uniformly translate into canonical target gene changes in Yy1 cKO versus Wnt10b KD.

Discussion

Our data suggest that the YY1-Wnt10b signaling contributes to the PC dendrite growth and synaptic plasticity regulation for cerebellar functions through autocrine signaling in the postnatal developmental stage (Figure 6).

Figure 6.

Schematic diagram of the proposed pathogenetic mechanism in Yy1 cKO model

YY1-downstream Wnt10b and Fzd4, expressed in postnatal PCs, play crucial roles in regulating PC dendritogenesis and synaptic strength. Wnt10b binds to Fzd4 to initiate signal transduction leading to β-catenin nuclear translocation and PKCγ activation in PCs through an autocrine mechanism. Disruption of the YY1-Wnt10b axis can potentially impair the development of the PC dendritic tree and spine formation, resulting in an ataxic symptom in mice.

YY1-related human disease provides an important context for interpreting our findings. In patients, YY1 haploinsufficiency and pathogenic variants cause Gabriele-de Vries syndrome, a neurodevelopmental disorder frequently accompanied by cerebellar signs, such as gait abnormalities and ataxia. Our data demonstrate that YY1 in PCs regulates dendritic growth and synapse maturation via the Wnt10b axis, engaging both canonical Wnt/β-catenin signaling and Ca²⁺-dependent pathways. These results provide a mechanistic framework linking YY1 dysfunction to cerebellar pathology: reduced YY1 activity may downregulate Wnt10b, attenuate canonical signaling, and disrupt Ca²⁺-mediated processes essential for PC maturation and circuit integrity. Moreover, YY1 likely regulates a broader transcriptomic network in PCs beyond Wnt10b, integrating Wnt signaling with growth-factor pathways to drive maturation. In addition to controlling Wnt10b, YY1 may regulate transcripts such as Pappa which modulates insulin-like growth factor (IGF) signaling by liberating IGFs from binding proteins and can promote dendritic growth and synaptogenesis^40,41 as well as Fzd4.⁴² Elevated Pappa and/or suppressed Fzd4 in YY1-deficient PCs could dampen IGF- and Wnt-signaling inputs, respectively, contributing to Yy1 cKO ataxic model pathogenesis. These possibilities align with the notion that YY1 coordinates multiple transcriptional programs to sculpt PC maturation causing early onset of motor coordination defect. Future work should test direct YY1 occupancy at Pappa and Fzd4 loci (e.g., via ChIP-seq/qPCR), quantify Pappa-driven IGF signaling and Fzd4-mediated Wnt signaling in YY1-deficient PCs, and evaluate whether simultaneous modulation of Pappa or Fzd4 with Wnt10b enhances synaptic maturation and motor function.

The role of Wnt10b in the CNS was reported in 2017, showing that Wnt10b can activate the mTOR pathway to stimulate injured CNS neurons for regeneration via autocrine signaling.⁴³ However, evidence for mTOR signaling in PC development and maintenance is limited. Transgenic mice with direct hyperactivation of mTORC1 in PCs showed impaired synapse elimination of CFs and motor incoordination, but their social behaviors were unaffected. Additionally, mTORC1 hyperactivation led to significant apoptosis of PCs.⁴⁴ Different SCAs are associated with varying levels of mTOR signaling, and correcting mTOR signaling in these SCAs could restore PC function.⁴⁵ Thus, Wnt10b downregulation may influence mTOR signaling in postnatal PCs, contributing to impaired dendritogenesis and spine formation. Although WNT10B/mTOR signaling was not examined here, our data provide valuable information on how Wnt10b is involved in cerebellar PCs development. Our results show that Wnt10b has a specific transient expression pattern in postnatal PCs, providing convincing evidence that Wnt10b participates in PC development within P12-16. Both Wnt10b loss-of-function and gain-of-function approaches in cerebellar organotypic cultures demonstrated that Wnt10b is necessary to drive PC dendritogenesis and support the maintenance of synaptic plasticity in a β-catenin- and PKCγ-associated manner.

In our proteomic datasets of Wnt10b KD cerebellar samples, most DEPs engage in translation at both the pre-synapse and post-synapse. This suggests that the downregulation of Wnt10b could potentially dysregulate local protein synthesis in the synaptic compartment, leading to defects in synaptic plasticity. In fact, the downregulation of synaptic plasticity markers was observed following Wnt10b KD in cerebellar slices. Consistently, the presence of additional Wnt10b in normal cerebellar slices increased the expression of synaptic plasticity markers in PCs. Thus, our results suggest that Wnt10b positively regulates postnatal PC dendrite development and contributes to synapse formation. Undoubtedly, we have shown that Wnt10b promote PKCγ expression, predominantly expressed in PCs, in postnatal developing PCs. Previous research has reported that PKCγ modulates the strength of synapse innervation of PCs,^46,47 implicating its involvement in the pathogenesis of several CNS diseases, especially ataxias,⁴⁸ and regulating various aspects of PC physiology and morphology.⁴⁹ Additionally, PKCγ mutations have been identified in SCA14 pathogenesis with Long-term depression (LTD) disruption.⁵⁰ These mutations lead to abnormal PKCγ function, affecting the normal functioning of PCs and contributing to the development of ataxic symptoms.⁵¹ Therefore, combining previous studies and our work, Wnt10b may regulate PC dendritogenesis and synapse formation through PKCγ activation during the third postnatal week in mice, which could serve as a potential molecular mechanism of pathogenesis in Yy1 cKO ataxia.

Our data show reduced nuclear translocation of β-catenin in PCs under both Yy1 cKO and Wnt10b KD conditions, consistent with attenuated canonical Wnt/β-catenin signaling downstream of the YY1-Wnt10b axis. RT-qPCR analysis of canonical targets revealed a partial, condition-specific response: Myc was downregulated in both Yy1 cKO and Wnt10b KD; Axin2 and Myc were reduced with Wnt10b KD but not robustly in Yy1 cKO, and Ccnd1 remained largely unchanged. This divergence suggests that canonical signaling is impaired but does not fully explain the phenotypes, pointing to contributions from non-canonical Wnt pathways and PC-specific regulatory mechanisms. In line with prior work, the role of β-catenin in cerebellar development and disease appears spatiotemporal dependent and integrated with other signaling inputs.^28,52,53 This condition-specific divergence in canonical target gene responses most likely reflects the spatiotemporal complexity of Wnt/β-catenin signaling in the postnatal cerebellum, where target gene activation is highly context dependent and integrated with cell-type-specific co-regulators. Importantly, the discordance between β-catenin protein levels and target gene transcription in the Yy1 cKO paradigm may also reflect the dilution of PC-specific transcriptional changes by the bulk cerebellar lysate used for RT-qPCR, as noted in the “limitations of the study” section. Taken together, these findings indicate that while canonical Wnt/β-catenin signaling is attenuated, it does not fully account for the observed phenotypes, and that the YY1-Wnt10b axis operates within a broader regulatory landscape that includes non-canonical Wnt pathways and additional YY1-dependent transcriptional programs.

In parallel, our findings implicate Wnt/Ca²⁺ signaling as an additional non-canonical contributor. YY1 inactivation disrupted Ca²⁺-dependent signaling in PCs, evidenced by marked reductions in CaMKIIβ in Yy1 cKO PCs and GCs (a result not recapitulated by Wnt10b KD) and by altered distribution of CaMKIIα between dendritic and cytoplasmic compartments in Yy1 cKO PCs. These changes point to disturbed Ca²⁺-dependent actin dynamics and synaptic regulation that may underlie dendritic and synaptic defects. Wnt10b, by contrast, appears to influence specific non-canonical outputs such as PKCγ independently of CaMKII changes, suggesting parallel but interacting routes from YY1 to PC maturation.

Since the downregulation of Wnt10b significantly contributes to the dendrite defects in Yy1 cKO PCs, we speculate that the addition of extra Wnt10b protein could restore dendrite growth in PCs of Yy1 cKO and drive normal synaptic plasticity strength. Considering that Wnt10b is a ligand with therapeutic potential, we decided to directly apply recombinant Wnt10b protein to cerebellar slice cultures. recWnt10b promotes dendritic growth in Yy1 cKO PCs, reaching levels comparable with control PCs. However, recWnt10b does not restore PKCγ, GluRδ2, or Synapsin-1 in the mutant PCs, even though these markers rise in control slices exposed to recWnt10b. This pattern indicates that Wnt10b is a major downstream effector of YY1 for dendrite expansion but is not sufficient to fully reconstitute synaptic maturation, suggesting additional YY1-dependent programs and/or signaling competence are required. Two non-mutually exclusive explanations may account for the incomplete rescue. First, YY1 loss is likely to deregulate multiple synapse-related genes, as suggested by our transcriptomic study (e.g., Grin2b, Grm5, Plcd1, Plch1, and Prkar2b),^54,55,56 which encode components of glutamate receptor signaling and PKA-dependent plasticity pathways. Restoring Wnt10b ligand availability is therefore insufficient to re-establish the full complement of synaptic proteins required for functional maturation, as these parallel transcriptional programs remain disrupted. Second, the downregulation of Fzd4 in Yy1 cKO PCs, itself a direct YY1 transcriptional target, may impose an intrinsic receptor-level constraint that limits the cellular response to exogenous recombinant Wnt10b, providing a mechanistic explanation for why dendritic growth is partially rescued while synaptic markers are not. Fzd4 has a mutation phenotype that includes cerebellar hypoplasia and PC loss documented in Mouse Genome Informatics (MGI), providing a possible explanation for why exogenous recWnt10b addition to mutant cerebellar sections could only partially rescue the phenotype. Indeed, further examination is essential to validate this hypothesis.

Taken together, our findings suggest a pathogenic mechanism of an ataxic-like phenotype related to the YY1-Wnt10b axis in cerebellar PC dendrite development and synaptic plasticity regulation (Figure 6). These findings may provide strategies for addressing YY1 and PKCγ-associated neurodevelopmental and neurodegenerative disorders, with potential therapeutic and pharmacological implications.

Limitations of the study

While our findings provide robust evidence for the YY1-Wnt10b axis in PC development, several limitations should be acknowledged. First, the transcriptomic and RT-qPCR analysis in the Yy1 cKO paradigm were performed using whole cerebellar lysates, which include GCs, interneurons, glia, and other cell types and can mask PC-specific transcriptional changes. In contrast, the Wnt10b KD experiments were carried out in cerebellar slices under a defined perturbation, which may produce a more readily detectable shift in canonical target-gene expression at the tissue level. Therefore, while our immunostaining results support an attenuation of β-catenin signaling in PCs, our current transcript measurements do not fully resolve the magnitude and timing of canonical Wnt transcriptional output specifically within PCs. Second, the direct interaction between Wnt10b and the Fzd4 receptor was validated through coIP in a heterologous cell system (SH-SY5Y). Although these results are suggestive, future studies utilizing in situ proximity ligation assays (PLAs) or PC-specific Fzd4 conditional knockout models would be beneficial to reinforce the spatial specificity of this ligand-receptor interaction in vivo or Fzd4 knockdown ex vivo. Third, there are practical constraints in our Wnt10b gain-of-function experiments. Recombinant ligand addition in organotypic slices has practical constraints including diffusion gradients, treatment duration, and reduced physiological activity-dependent refinement that may be sufficient to drive morphological outgrowth yet insufficient to re-establish synaptic proteostasis. These limitations motivate future in vivo experiments using temporally controlled, PC-specific Wnt10b restoration. Finally, the behavioral characterization in this study was restricted to motor assays aligned with the functional output of the cerebellar vermis, which is the primary region examined for morphological and molecular analyses. The non-motor and affective functions of the cerebellum, which are associated with neocerebellar regions such as Crus I and Crus II, were not assessed. Future studies should evaluate anxiety-related and cognitive behaviors in Yy1 cKO mice to provide a more complete phenotypic profile and to determine whether YY1 deficiency in PCs contributes to non-motor neurodevelopmental symptoms, as observed in patients with Gabriele-de Vries syndrome.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Kin Ming Kwan (kmkwan@cuhk.edu.hk).

Materials availability

The plasmids to generate the Wnt10b in situ probe, Wnt10b cDNA and Wnt10b promoter sequence for luciferase assay are available from the lead contact upon request. This study did not generate new unique reagents.

Data and code availability

RNA-seq data have been deposited at GEO and are publicly available as of the date of publication. Accession numbers are listed in the key resources table: GEO: GSE336211. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD079944. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

We thank Prof. Richard R. Behringer (University of Texas M.D. Anderson Cancer Center, Houston, TX, USA) for helpful advice and comments on the manuscript. We thank Prof. Huating Wang (Department of Orthopedics and Traumatology, CUHK, Hong Kong) for providing Yy1 conditional mutant mouse line. We thank Ms. Sze Nee Lim and Mr. Freddie Kwok for their technical support, Dr. Ka Kui Tong for assisting with the ISH experiments, and Dr. Nga Chu Lui (Kathy), Dr. Keng Ioi Vong (Harry) and Dr. Tsz Ching Ma (Charlotte) for suggestions on the manuscript. The work described in this article was supported by the grants from the Research Grant Council of the Hong Kong Special Administration Region (project nos. GRF 14112224, 14114921, CRF C4002-21E, and AoE/M-402/25-N) and a Chinese University of Hong Kong Direct Grant for Research.

Author contributions

Y.L.L. and K.M.K. conceptualized the project, designed the experiments, analyzed the data and wrote the initial manuscript draft. X.D. performed behavioral analysis and Golgi staining analysis. T.H.L. and T.F.C. performed the transcriptomic study and T.C.N.L. performed the LC-MS/MS study. All authors read and approved the final manuscript.

Declaration of interests

All authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse monoclonal anti- Active-β-Catenin	Millipore	Cat#05-665 RRID: AB_309887
Rabbit polyclonal anti-ARPC3	Millipore	Cat#ABN176 RRID: AB_10806335
Mouse monoclonal anti-Calbindin-D-28K	Sigma-Aldrich	Cat#C9848 RRID: AB_10000347
Mouse monoclonal anti-CaMKII alpha	Thermo Fisher Scientific	Cat#Ma1-048 RRID: AB_325403
Mouse monoclonal anti-CaMKII beta	Invitrogen	Cat#13-9800 RRID: AB_2533045
Mouse monoclonal anti-Espin	BD Transduction	Cat#611656 RRID: AB_399174
Alexa Fluor® 647 Phalloidin	Invitrogen	Cat#A30107 RRID: AB_2782963
Goat polyclonal anti-FZD1	Invitrogen	Cat#PA5-47072 RRID: AB_2610121
Rat polyclonal anti-FZD4	BiCell Scientific	Cat#50604
Mouse monoclonal anti-GluRδ2	BD Transduction	Cat#611006 RRID: AB_398881
Rabbit polyclonal anti-InsP3R, Type1	Sigma-Aldrich	Cat#ABS55
Rabbit polyclonal IRSp53	Abcam	Cat#ab126057
Mouse monoclonal anti-MAP2	Sigma-Aldrich	Cat#MAB3418 RRID: AB_94856
Rabbit monoclonal anti-PKCα	Cell Signaling	Cat#59754 RRID: AB_10693963
Rabbit polyclonal anti-PKC gamma	Abcam	Cat#ab71558
Mouse monoclonal anti-PKC gamma	Invitrogen	Cat#13-3800 RRID: AB_2533015
Rabbit polyclonal anti-Synapsin1	Sigma-Aldrich	Cat#S193
Rabbit polyclonal anti-VGLUT1	SYSY	Cat#135-303 RRID: AB_887878
Guinea pig polyclonal anti-VGluT2	Millipore	Cat#AB2251 RRID: AB_1587626
Rabbit polyclonal anti-Wnt10b	Invitrogen	Cat#PA5-116125 RRID: AB_2900759
Rabbit monoclonal anti-YY1	Abcam	Cat#ab109237
Rabbit polyclonal anti-β-Actin	Cell Signaling	Cat#4967
Mouse monoclonal anti-β-Catenin	BD Transduction	Cat#610153
Donkey anti-Goat IgG (H + L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 568	Invitrogen	Cat#A-11057; RRID: AB_2534104
Donkey Anti-Goat IgG Antibody, HRP conjugate	Sigma-Aldrich	Cat#AP180P
Goat anti-Mouse IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 488	Invitrogen	Cat#A-11029 RRID: AB_2534088
Goat anti-Mouse IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 647	Invitrogen	Cat#A-21236 RRID: AB_2535805
Goat Anti-Mouse IgG Antibody, HRP conjugate	Sigma-Aldrich	Cat#AP181P
Goat Anti-Rat IgG H&L (Alexa Fluor® 647)	Abcam	Cat#ab150159 RRID: AB_2687948
Goat Anti-Rat IgG Antibody, HRP conjugate	Sigma-Aldrich	Cat#AP136P
Goat anti-Rabbit IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 488	Invitrogen	Cat#A-11034 RRID: AB_2576217
Goat anti-Rabbit IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 647	Invitrogen	Cat#A-21245 RRID: AB_2535813
Goat Anti-Rabbit IgG Antibody, HRP conjugate	Sigma-Aldrich	Cat#12-348
Bacterial and virus strains
DH5α Competent Cells	Thermo Fisher Scientific	Cat#18265017
Chemicals, peptides, and recombinant proteins
Anti-Digoxigenin-AP	Roche	Cat#11093274910
ProbeQuant™ G-50 Micro Columns	Cytiva	Cat#28-9034-08
Normal Rabbit IgG	Sigma-Aldrich	Cat#12-370
Hoechst 33342	Sigma-Aldrich	Cat#B2261
Dynabeads M-280 Sheep Anti-Rabbit IgG	Thermo Fisher Scientific	Cat# 11204D
Protein A/G PLUS-Agarose	Santa Cruz Biotechology	Cat# sc-2003
TRIzol Reagent	Invitrogen	Cat# 15596-026
Recombinant Mouse Wnt-10b Protein	R&Dsystems	Cat# 2110-WN
INTERFERin	PolyPlus	Cat#101000028
Lipofectamine™ 2000 Transfection Reagent	Invitrogen	Cat# 11668027
Critical commercial assays
SYBR™ Green Universal Master Mix	Applied Biosystems	Cat# 4309155
WESTSAVE Up	AbFrontier	Cat#LF-QC0102
Rapid GolgiStain Kit	FD NeuroTechnologies	Cat# PK401
Dual Luciferase Reporter System	Promega	Cat #E1960
Transcriptor First Strand cDNA Synthesis Kit	Roche	Cat# 04379012001
Experimental models: Cell line
SH-SY5Y	ATCC	CRL-2266
Experimental models: Organisms/strains
Mouse: Pcp2 Cre/+	The Jackson Laboratory	JAX:010536
Mouse: Yy1 fx/fx	The Jackson Laboratory	JAX:014649
Mouse: C57BL/6J	CUHK LASEC	N/A
Oligonucleotides
ON-TARGETplus SMARTpool Wnt10b siRNA	Dharmacon	L-043676-01-0005
ON-TARGETplus Non-targeting Control Pool	Dharmacon	D-001810-02-05
Primers for genotyping see Table S1	This paper	N/A
Primers for chipPCR see Table S1	This paper	N/A
Primers for in situ hybridization see Table S1	This paper	N/A
Primers for luciferase promoter assay see Table S1	This paper	N/A
Primers for realtime-qPCR see Table S1	This paper	N/A
Recombinant DNA
Myc-DDK-tagged Mouse YY1 transcription factor plasmid	Origene	Cat# MR20653
pRL-SV40	Promega	N/A
pGL3-Wnt10b promoter	This paper	N/A
pDsRed1-N1-Wnt10b	Ths paper	N/A
pBluescript-Wnt10b	This paper	N/A
Deposited data
RNA-seq data	This paper	GEO: GSE336211
LC-MS/MS proteomics data	This paper	PRIDE: PXD079944
Software and algorithms
Fiji/ImageJ	NIH	RRID: SCR_002285
GraphPad Prism 8	GraphPad Software	RRID: SCR_002798
Leica Application Suite X	Leica Microsystems	RRID: SCR_013673
Proteome Discoverer Version 2.4	Thermo Fisher Scientific	N/A
STAR version 2.7.1a	Dobin et al.57	N/A
DESeq2 version 1.26.0	Love et al.58	N/A
Trimmomatic version 0.39	Bolger et al.59	N/A

Experimental model and study participant details

Experimental mice

The generation and genotyping methods of Pcp2-Cre/+ (also known as L7-Cre/+) and Yy1 ^flox/flox C57BL/6 J mice were described previously.^60,61 For genotyping, genomic DNA of postnatal mice was obtained by toes and conduct the Jackson laboratory’s genotyping protocols No. 014649 for Yy1, and No. 007684 for Cre. Yy1 conditional knockout (cKO) mice were generated and maintained on C57BL/6J background by crossing Pcp2-Cre/+; Yy1^fx/+ females with Yy1^fx/fx males. After confirmation of the heterozygous conditional inactivation of Yy1 no causing any morphological defect and behavioral alteration, no-Cre and Pcp2-Cre/+; Yy1^fx/+ littermate mice were used as the control. Control and mutant pairs from at least 3 littermates with both sexes were examined for each of the following assays. All mice were kept in the Animal House of the Chinese University of Hong Kong with a 10:14 hour light/dark cycle. All animal procedures were conducted with the approval of the Animal Experimentation Ethics Committee of The Chinese University of Hong Kong (Ref No.: 23-051-GRF and 24-074-GRF).

Cell line

SH-SY5Y cells were maintained in DMEM/F12 with 10% fetal bovine serum (FBS), 1 × GlutaMAX and 1 × penicillin–streptomycin (Life Technologies) at 37°C humidified 5% CO₂ incubator.

Organotypic culture of cerebellar slice sections

Cerebella from P12 or P13 C57BL/6J mice were dissected and freshly sectioned into 350 μm thick slices in ice-cold HBSS with 0.6% glucose using Leica VT1000S vibratome. The cerebellar slices were then washed twice with cerebellar slice culture medium (Neurobasal-A medium, 1 × B-27 supplement, 1 × GlutaMAX, and 1 × penicillin–streptomycin, Life Technologies) and placed on 30 mm cell culture inserts (0.4 μm pore size, Millipore). The culture inserts were then placed into the 6-well plate with 1 mL culture medium per well and maintained in a 37°C humidified 5% CO₂ incubator for 16 hours before recombinant Wnt10b protein treatment or siRNA transfection.

Method details

Region selection and sampling

Only age-matched controls and cKO mutants were used for group-to-group comparisons in all analyses. The cerebellar vermis was selected as the region of interest for all analyses. The vermis is a core component of the spinocerebellum and is intimately involved in motor control, posture, and locomotion, with somatic sensory input from the trunk and proximal body via spinocerebellar pathways.^62,63 To maintain methodological consistency and to focus on motor-related PC changes, analyses were restricted to vermis. Cerebellar lobule V-VI was used to capture images of PCs unless specifically mentioned.

Wnt10b recombinant protein treatment ex vivo

P13 cerebellar slices were treated with recombinant Mouse Wnt-10b Protein (R&D Systems Inc., 2110-WN) one day after plating. Recombinant mouse protein was added directly into the medium and cerebellar slices were harvested after two additional days for analysis.

siRNA transfection ex vivo

P12 cerebellum slices were transfected by different sets of siRNAs on the second day of culture. The old culture medium was replaced by 1 mL of fresh culture medium in each well of the 6-well culture plates before transfection. ON-TARGETplus SMARTpool siRNA for Wnt10b (Dharmacon, L-043676-01-0005) was used to knock down Wnt10b, or mouse non-target (NT) siRNA (Dharmacon, D-001810-02-05) was used as a control. 100 nM of siRNA was dissolved in 100 μL of 5% glucose solution mixed with 4 μL of INTERFERin transfection reagent (Polyplus transfection) and incubated for 20 min at room temperature. The mixture was then slowly applied to the cerebellar slices three times at 30 min intervals. The final concentration of siRNA for transfection would be 10 nM. The transfected cerebellum slices were maintained for 3 days.

Plasmid transfection in vitro

Transfection was performed using Lipofectamine™ 2000 Transfection Reagent (Thermofisher) following the manufacturer’s instructions when cells reached 70% confluence. The culture medium was then changed 4 hours post-transfection.

Image analysis

Images of PCs were analyzed with NeuronJ, Fiji ImageJ and Imaris Filament Tracer to measure the dendritic length and the number of branch points. Cell density, spine density and fluorescence intensity were quantified using Fiji ImageJ. Only the central focal planes of the PC proximal dendrites from the z-stacks of confocal images were used for the assessment of F-actin localization.

Motor behavior examinations

For all the motor behavior tests, mice littermates were divided into male and female groups. All experiments were performed using at least three different litters. Data from male groups are typically used to avoid the disturbance generated by differential menstrual periods of females, unless specifically labeled otherwise.

For the ledge test, mice were acclimated in a box with bedding and food for 10 min. Then the mice were lifted on the ledge of the box. Normally, the mice will walk along the ledge and attempt to get down back inside the box and land with their front paws. “In” was taken when mice got down back inside the box, and “Out” was taken when mice dropped outside the box. The values were used for statistical analysis.

For the elevated balance beam, the elevated balance beam test used a 2 cm x 70 cm narrow beam set 50 cm above the bench. Mice were trained for 3 days and 1 day before testing. On the day of the test, time to complete the 70 cm distance and number of missteps were recorded over five trials, with 3-minute rest periods between trials.

For the rotarod test, mice were trained 3 days and 1 day before the measurement on the rotarod (Panlab) at 0 rpm for 1 min, then at 4 rpm for 1 min. On the day of measurement, mice were placed on the rotarod with the fixed speed at 4 rpm, and with an increasing speed from 4 to 40 rpm in 5 min. The walls and rods were wiped with 95% Ethanol before use and before each test. Five measurements were recorded for each mouse with 5-minute rest periods, and mean values were used for statistical analysis.

For footprint analysis, mice's forepaws were painted red and hind paws blue, then allowed to run through a dark tunnel (10 cm width x 10 cm height x 85 cm length) lined with white paper. The stride length, forelimb-hindlimb overlap, and left-right stride widths were measured from five consecutive steps. Mean values for each parameter were used for statistical analysis.

Dual luciferase reporter assay

Putative YY1 binding sites in the Wnt10b promoter were predicted referenced by UCSC Genome Browser. To clone the Wnt10b promoter region, specific primer pairs targeting GRCm39 chr15:98,672,350-98,672,801 (-451 from transcription initiation site ATG) and KAPA HiFi HotStart DNA Polymerase (KAPA Biosystems) were used. The PCR products were cut with HindIII and XhoI and then inserted into pGL3-basic to obtain -451 construct. Deletion of potential YY1 motif in Wnt10b promoter sequence was performed using Q5® Site-Directed Mutagenesis Kit according to the manufacturer’s instruction (NEB). The successful deletion of the potential YY1 binding site in the Wnt10b promoter sequence was confirmed by full-length sequencing. The Dual luciferase reporter assay was performed according to the manufacturer’s instructions with pRL-SV40 as the internal control. Tagged mouse Yy1 overexpression plasmid (Origene), pCMV6-entry empty vector was used as the negative control for co-transfection. SH-SY5Y cells were seeded at 2.5 x 10⁴ cells/well with 200 μL medium in 96 well plates. After 16 hours, the medium was refreshed, and plasmids were transfected with Lipofectamine™ 2000 transfection Reagent (Thermofisher) following the manufacturer’s instructions. The culture medium was then changed 4 hours post-transfection. Luminescence signals were measured by the CLARIOStar monochromator-based microplate reader 48 hours after transfection.

Western Blot analysis

Dissected cerebella or cerebellar sections were lysed in SDS lysis buffer (1% SDS, 10% glycerol, 5% β-mercaptoethanol and 125 mM Tris-HCl, pH 6.8) with proteinase inhibitor cocktail (cOmplete Mini, Roche). The tissues were homogenized and centrifuged at 4°C 12,000 g for 10 min. The supernatant was collected, and protein concentrations were measured by standard Bradford assay. Protein samples were then denatured by boiling with 6X loading buffer at 100°C for 10 min. The denatured protein samples were then separated by 12% SDS-polyacrylamide gel electrophoresis. The separated proteins were blotted to polyvinylidene difluoride membrane by Bio-Rad Trans-Blot Turbo Transfer System. The membranes were blocked with 5% non-fat dry milk in Tris-buffered saline with 0.1% Tween-20 (TBST) for 1 h at room temperature. The blocked membrane was incubated overnight at 4°C with primary antibodies: rabbit anti-Wnt10b (1:1000, Abcam), rabbit anti-β-actin (1:1000, Cell Signaling), or mouse anti-β-actin (1:1000, Cell Signaling). After washing with TBST, the membranes were incubated with respective secondary antibodies conjugated to horseradish peroxidase (HRP) (1:2000, GE Healthcare) for 1 h at room temperature. The signals were detected using Western Blotting Substrate WESTSAVE Up (Abfrontier) by Chemidoc imaging system. For each experiment, at least three control and mutant littermate pairs from different litters were tested.

Golgi staining

Golgi staining was performed using the FD Rapid GolgiStain Kit (FD NeuroTechnologies) according to the manufacturer’s user manual. Briefly, mouse whole cerebella were quickly dissected and cut at the parasagittal plane of the hemisphere to facilitate gluing to the sample stage during the latter sectioning steps. Unwanted tissue attached to the cerebella such as choroid plexus was removed by microdissection in double distilled H2O. The cerebella were rinsed in double distilled H2O to remove blood and then impregnated in premixed solutions A and B supernatant at room temperature for two weeks in the dark, with the impregnation buffer replaced on the second day. The tissue was then transferred to solution C and kept at 4 °C for three to four days with solution C replaced on the second day of immersion. The cerebella were sectioned in solution C at 100 μm thickness on ice using the vibrating microtome (Leica). Sections were mounted on the 3% gelatine-coated sections and air-dried at room temperature in the dark for up to one week. The dried sections were stained in a mixture containing solution D and E for 10 min and washed with distilled water. Dehydration was accomplished by immersing the slides in ethanol series and clearing them in xylene. Slides were mounted with DPX and observed using light microscopy or confocal microscope SP8. For dendrite quantification, light microscopy images of PCs with completely elaborated dendritic trees were analyzed. Four to ten PCs from control-cKO littermate pairs were quantified. Measurements were made using LAS X and ImageJ software. For spine quantification, confocal z-stack images were used, and spines were categorized based on the following hierarchical criteria: (1) Branched = more than one spine; (2) Filopodium = length > 2 μm; (3) Mushroom = spine head width > 0.5 μm; (4) Thin = length: width ratio > 1; (5) Stubby = length: width ratio ≤ 1. For each time point, 2-5 dendritic segments from 3 control-cKO littermate pairs of at least 2 different litters were quantified.

RNA extraction, reverse transcription, and real time quantitative PCR (qPCR)

Total cerebellum RNA from postnatal controls and cKO mutants was extracted by TRIzol reagent (Life Technologies). The RNA was then reverse-transcribed to complementary DNA (cDNA) using M-MLV (Moloney murine leukemia virus) Reverse Transcriptase (Life Technologies) with oligo(dT)20 primers. For qPCR, diluted cDNA samples were amplified by SYBR™ Universal Master Mix (Applied Biosystems) using specific primers with PCR profiles of 95°C (5 min) followed by 40 cycles of 95°C (15 s), 62°C (20 s) and 72°C (20 s). Fluorescence was measured by Bio-Rad CFX96 Real-Time PCR Detection System at the end of each cycle. Each sample was assayed in technical duplicate and gene expression was normalized to β-actin expression.

Section in situ hybridization

Cerebella were dissected from the control and cKO mice and were fixed in 4% paraformaldehyde (PFA) in PBS at 4°C for 16 h. For the preparation of paraffin sections for in situ hybridization, fixed cerebella were dehydrated, paraffin-embedded, and sectioned into 7 μm thick sections with a microtome. After prehybridization treatment, the sections were hybridized with the digoxigenin-labeled riboprobes against Wnt10b, synthesized using a transcriptional kit (Roche) from pBluescript-Wnt10b with T3 RNA polymerase (Promega), at 65°C overnight following by purifying labeled probes with ProbeQuant™ G-50 Micro Columns (Cytiva). After several rounds of washing with different dilutions of SSC, the sections were blocked at room temperature for 1 h. The sections were then incubated with 1:1,000 anti-digoxigenin alkaline phosphatase-conjugate (Roche) at 4°C overnight. The signals were visualized by staining 250 μg/mL NBT (nitroblue tetrazolium) and 125 μg/mL BCIP (5-bromo-4-chloro-3-indolyl-phosphate) at 37°C overnight. Images of sections were captured by an Olympus BX43 microscope equipped with a DP72 camera.

Histological analysis

Dissected cerebella from adult mice were fixed in 4% PFA in PBS at 4°C for 16 h. For the paraffin sections used in hematoxylin and eosin staining or immunohistochemistry, fixed cerebella were dehydrated, paraffin-embedded, and sectioned into 7 μm thick sections with a microtome. For cryosections, the fixed cerebella were treated with 30% sucrose/PBS for cryoprotection at 4°C for 2 nights, subsequently frozen in optimal cutting temperature (OCT) compound using liquid nitrogen, and nitrogen and sectioned into 10 μm thick slices using a cryostat microtome. For immunohistochemistry, antigen retrieval was performed by microwave treatment in 10 mM sodium citrate buffer (pH 6.0) if needed. The sections were blocked with blocking solution (0.2% bovine serum albumin, 10% lamb serum, and 0.1% Triton X-100 in PBS) at room temperature for 1 h. The blocked sections were then incubated with primary antibodies overnight at 4°C. After washing, the sections were incubated with respective secondary antibodies at room temperature for 1 h. Secondary antibodies were conjugated by Alexa Fluor 488/568/647 (Life Technologies, 1:1000). F-actin staining was performed with 1:1000 Alexa Fluor 647 phalloidin (Life Technologies) at 4°C overnight if needed. For immunohistochemistry, HRP signals were visualized by incubating the sections with diaminobenzidine (Dako) and counterstained with hematoxylin. For organotypic culture slice fixation and immunostaining, the culture slices were fixed with 4% PFA/PBS for 30 minutes at room temperature after washing. They were then incubated at 4°C for 3 nights with a blocking solution containing 0.2% (w/v) bovine serum albumin (BSA), 10% (v/v) horse serum, and 0.1% (v/v) Triton X-100 in PBS. Primary antibodies were diluted in the same blocking solution and incubated at 4°C for two nights. Secondary antibodies were also diluted in the blocking solution and incubated at 4°C overnight. Nuclear staining was achieved with Hoechst 33342 (1:2000) (Sigma). Images of hematoxylin and eosin and immunohistochemistry sections were captured by a microscope equipped with an Olympus DP72 camera. Fluorescence images were acquired by Leica TCS SP8 microscope.

Chromatin immunoprecipitation

Cerebella from P14 controls were finely minced into small pieces in PBS. The samples were incubated with 1:10 volume of crosslinking buffer (11% PFA, 0.1 M NaCl, 1 mM EDTA, 0.5 mM EGTA and 50 mM HEPES, pH 7.6) for 10 min under agitation at room temperature. Crosslinking was quenched by incubating 125 mM glycine for 5 min, followed by two PBS washes. Sequential tissue lysis was performed using two buffers: Lysis Buffer 1 (140 mM NaCl, 1 mM EDTA, 10% glycerol, 0.5% NP-40, 0.25% Triton X-100, 50 mM HEPES, pH 7.6) and Lysis Buffer 2 (200 mM NaCl, 1 mM EDTA, 0.5 mM EGTA, 10 mM Tris, pH 7.6), each supplemented with cOmplete Mini protease inhibitors (Roche). Between buffer treatments, samples were centrifuged at 2,000 g for 10 minutes at 4°C. After lysis, the pellets were resuspended with Sonication Buffer (1 mM EDTA, 0.5 mM EGTA, 1% Triton-X100, 1% sodium deoxycholate and 10 mM Tris-HCl, pH 7.6, with cOmplete Mini protease inhibitors) and sonicated on ice using a Branson Sonifier 150 (50 rounds of 20 pulses, 1-second pulse at 50% power, with 20-second intervals). The chromatin solution was centrifuged at 10,000 g for 10 min at 4°C and chromatin containing supernatant was then incubated either with 10 μg rabbit anti-YY1 antibody (Abcam ab109237) or 10 μg rabbit IgG (Millipore 12-370) served as negative control (Mock) at 4°C overnight. After antibody incubation, the immunocomplexes were captured using Dynabeads M-280 Sheep Anti-Rabbit IgG (Life Technologies) at 4°C for 1–2 h. The beads were collected and washed five times with RIPA buffer (1 mM EDTA, 0.7% deoxycholic acid, 1% NP-40, 0.5 M LiCl and 50 mM HEPES, pH 7.6) and once with TE buffer (10 mM Tris-HCl, pH 8.0, and 0.1 mM EDTA). Chromatin was eluted with Elution Buffer (50 mM Tris, pH 8.0, 10 mM EDTA and 1% SDS) at 65°C for 10 min. The eluted chromatin was reverse-crosslinked with Elution Buffer at 65°C overnight and then treated with Proteinase K at 55°C for 1–2 h. The chromatin was extracted and purified by phenol/chloroform. The purified DNA was subjected to PCR to evaluate YY1 binding at the promoter regions of Wnt10b, Fzd1, and Fzd4 genes.

Pull-down assay and co-immunoprecipitation

The control and mutant mouse cerebella were dissected in cold non-denaturing lysis buffer (20 mM Tris HCl, pH 8 with 137 mM NaCl, 1% Nonidet P-40 (NP-40) and 2 mM EDTA). Samples homogenize in ice-cold lysis buffer supplemented with cOmplete Mini protease inhibitors (Roche). Tissue homogenization was performed on ice using an electric homogenizer, followed by additional mechanical disruption through a 27-gauge needle (10 passes) using a 1 mL syringe. The homogenate then centrifuged at 12,000 g at 4°C for 10 minutes. Supernatants were collected and topped up to 1 mL with ice-cold lysis buffer. For preclearing, the lysate was incubated with 0.25 μg Rabbit IgG (Millipore 12-370) and 20 μL of Protein A/G PLUS-Agarose (sc-2003) for 30 minutes at 4°C, followed by centrifugation at 1,000 g for 30 seconds at 4°C. Supernatant was transferred into a new tube and then incubated with either 2 μg rabbit anti-YY1 antibody (Abcam ab109237) or 10 μg rabbit IgG (Millipore 12-370) served as negative control (Mock) at 4°C for 2 hours. The immunocomplexes were captured with 20 μL Protein A/G PLUS-Agarose at 4°C on a rocker platform overnight. The agarose beads were collected by centrifugation at 1,000 g for 30 seconds at 4°C and washed five times with ice-cold RIPA buffer, with centrifugation steps (1,000 g, 30 seconds, 4°C) between washes. The pellet was resuspended in 25 μL of RIPA, followed by Western blot analysis.

Protein sample preparation

After transfection, cerebellar slices were lysed in urea lysis buffer (6 M urea, 2 M thiourea, Tris pH 8.0) supplemented with Protease and Phosphatase Inhibitor Mini Tablets (Pierce, Thermo Fisher Scientific). Protein concentration was determined using the 660nm Protein Assay Kit (Pierce, Thermo Fisher Scientific). The urea concentration of the samples was diluted to 1 M with 25 mM ammonium bicarbonate (ABC, pH 8.5; Sigma Aldrich). Proteins were reduced using 5 mM dithiothreitol (Sigma Aldrich) for 1 h, then alkylated with 5 mM iodoacetamide (Sigma Aldrich) for 30 min at room temperature in dark. Protein digestion was performed overnight at 37°C using sequencing grade trypsin (Promega) at a 1:10 trypsin:protein ratio. The tryptic peptides were purified using C18 spin columns (Thermo Fisher Scientific) and resuspended in 25 mM ABC, at pH 8.5. Digested peptides were fractionated into 8 fractions using High pH Reversed-Phase Fractionation Kit (Thermo Fisher Scientific). The fractionated samples were vacuum-dried and stored at -80°C until further analysis.

LC-MS/MS analysis

Peptide fractions were separated using a Dionex UltiMate 3000 RSLC nano system equipped with a 25 cm-long, 75 um-i.d. C18 column. The columns were heated to 50 °C for all experiments. Mobile phases were 0.1% formic acid in water (buffer A) and acetonitrile (buffer B). Peptide fractions were loaded onto the trapping column for 5 min at 100% A at 5 μL/min, then eluted through the analytical column at 300 nL/min with the gradient begins with 0% solvent B, gradually increases through defined intervals to reach 80% by 35 minutes, holds at that concentration until 40 minutes, and then returns to 0% by 45 minutes, completing the gradient cycle. Eluates were analyzed by an Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Fisher Scientific). The MS scan range was set to 375–1500 m/z, with a Orbitrap resolution of 60,000, an automatic gain control (AGC) target of 4 × 10⁵, and a maximum injection time of 50 ms. Fragmentation was performed in higher-energy collisional dissociation (HCD) mode with 30% collision energy, and precursor isolation windows were set to 1.6 m/z. The Orbitrap resolution of MS/MS scan was set to 15 000 with the AGC target and maximum injection time were 5 × 10⁴ and 250 ms, respectively. The Raw files were analyzed using Proteome Discoverer (version 2.4) and searched with SEQUEST using the UniProt protein databases for Mus musculus (UP000000589; downloaded December 1, 2023). The searching parameters were as follows: oxidation of methionine (+15.9949 Da) and carbamidomethylation of cysteine (+57.0215 Da) was set as dynamic modification; precursor-ion mass tolerance, 10 ppm; fragments-ion mass tolerance, 0.02 Da. Protein level false discovery rate was estimated by Percolator at an experimental q-value (exp. q-value) threshold of 0.05. Label-free quantification was conducted utilizing the precursor ion quantification module of Proteome Discoverer. Differential protein abundances between groups were tested with t-test, with significance determined by a Benjamini-Hochberg adjusted p-value of 0.05.

Whole cerebellum RNA extraction and transcriptome sequencing

The cerebella of P14 controls and mutants were dissected and rinsed in ice-cold DEPC- treated PBS. RNA extraction was performed using TRIzol-chloroform, the same as described in RNA extraction part. RNA concentration and purity were measured using Nanodrop 2000 (Thermo Scientific) and Qubit (Invitrogen). RNA samples from two Pcp2^Cre/+; Yy1^fx/+ (heterozygous control) and three Pcp2^Cre/+; Yy1^fx/fx (cKO) were used. 50 μL of each RNA sample was sent to Groken Biosciences for transcriptome sequencing using Illumina sequencing platforms with reading lengths of paired-end 150 bp.

RNA-seq data analysis

Raw sequencing files were checked using FastQC version 0.11.8 to confirm absence of adapter. Quality trimming was performed using Trimmomatic version 0.39.⁵⁹ Trimmed reads were aligned to mouse reference genome GRCm38 using STAR version 2.7.1a.⁵⁷ Gene expression quantification was performed with featureCounts function in Rsubread version 2.0.0⁶⁴ using GENCODE vM23 annotation. Normalization and differential expression analysis were performed using DESeq2 version 1.26.0.⁵⁸

Quantification and statistical analysis

For quantifications of density, random areas were used for measurement. All the measurements were done blinded to the mouse genotypes, although the phenotypes could be clearly recognized at certain time points. Statistical analysis was performed using the software GraphPad Prism 9.0 (GraphPad Software Inc., San Diego, CA, USA). All data were expressed as means ± SEM. Student's t-test was used for group-to-group comparisons. Probability values p <0.05 were considered as statistically significant.

Supplemental information

Data S1. (Excel datasheet_RNA-seq_study, related to Figure 2)

Differential expression analysis results of comparison between P14 Yy1 control and Yy1 cKO cerebella revealed a total of 231 upregulated genes and 723 downregulated differentially expressed genes (DEGs). iRegulon analysis suggested that 133 DEGs contain the YY1 binding motif, and 12 of these DEGs showed PC-specific expression in the P14 wild-type cerebellum, as indicated by the Mouse Brain Atlas ISH database.

Data S2. (Excel datasheet_LCMSMS_study, related to Figure 4)

Proteomic study results of Wnt10b KD ex vivo cerebellar sample. A total of 4,273 proteins were detected by LC-MS/MS. Comparison between control and Wnt10b KD cerebella sections identified 142 differentially expressed proteins (DEPs). Gene Ontology (GO) analysis revealed that these DEPs are enriched in translation at both presynaptic and postsynaptic sites.

Video S1. (P28 control and cKO motor performances, related to Figure 1)

P28 control mice walk smoothly along the ledge of the cage and land easily using their paws; while Yy1 cKO mice walk along the ledge of the cage with severe body shaking and fail to land properly with their paw in the ledge test.