Oligodendroglial TANGO2 Regulates Lipid Metabolism to Control Motor Coordination

ABSTRACT

TANGO2 deficiency disorder (TDD) is a rare genetic disease caused by mutations in the TANGO2 gene, characterized by prominent neurological symptoms. However, the pathological mechanisms underlying TANGO2 loss‐of‐function in neurologic symptoms remain unknown. Here, we generated constitutive and cell‐type‐specific Tango2 knockout mouse models to examine TANGO2's role in the central nervous system (CNS). Behavioral analyses revealed that both constitutive and oligodendrocyte‐specific deletion of Tango2 recapitulate the motor deficits associated with individuals with TDD. Morphological quantifications further showed that Tango2 deletion led to robust cerebellar myelin loss and an increase in synapse number in the cerebellar cortex. In addition, transcriptional analysis and lipidomic profiling demonstrated that Tango2 deletion downregulated key processes involved in phospholipid metabolism. Significantly, vitamin B5 supplementation alleviated motor deficits and cerebellar myelin defects in Tango2 knockout mice. Overall, our findings establish that TANGO2 is essential for maintaining normal motor behaviors by regulating lipid metabolism in oligodendroglia.

TANGO2 deficiency leads to motor deficits and defective cerebellar myelination in mice, with impaired phospholipid metabolism in oligodendrocytes as an underlying mechanism. Vitamin B5 supplementation restores myelination and improves motor performance. These findings identify TDD as a myelin disorder and highlight lipid metabolism as a potential therapeutic target.

1. Introduction

TANGO2 deficiency disorder (TDD) is a rare genetic disease caused by biallelic variants in the TANGO2 (Transport and Golgi Organization 2) gene. In addition to experiencing metabolic crisis, rhabdomyolysis, and cardiac arrhythmias, most TDD individuals exhibit a range of neurological abnormalities, including neurodevelopmental delay, gross and fine motor dysfunction, intellectual disability, dysarthria, epilepsy, and paroxysmal neurologic episodes [1, 2]. Neuroimaging assessments have revealed that approximately 60% of TDD individuals exhibit diffuse ventricular enlargement, reduced cerebral volume (global or cerebellar), and white matter abnormalities [1, 2, 3, 4]. Postmortem examinations of TDD brain samples have further identified gliosis and ectopic neurons within the cerebral white matter and cerebellum [5]. However, it remains unknown whether these neurological abnormalities are a direct consequence of TANGO2 deficiency within the central nervous system (CNS).

The role of TANGO2 in cellular function remains elusive, as do the precise mechanisms by which TANGO2 mutations or deletions lead to TDD. Overexpression studies suggest that TANGO2 localizes to the Golgi apparatus, cytoplasm, and mitochondria [5, 6, 7, 8, 9]. Furthermore, TANGO2 deficiency has been shown to impair oxidative phosphorylation, reducing ATP production and increasing oxidative stress [4, 10, 11, 12]. In several models—including HepG2 cells, patient‐derived fibroblasts, and zebrafish—TANGO2 loss‐of‐function has been implicated in disrupting lipid metabolism [7, 13, 14]. Notably, treatment with vitamin B5, a precursor of coenzyme A, has been reported to alleviate both motor and non‐motor deficits in individuals with TDD [15, 16]. Approximately half of individuals with TDD report diminished white matter [1, 2]. White matter consists of the myelin sheath and the axons it ensheathes. As a multilayered lipid membrane produced by oligodendrocytes, myelin contains a high proportion of lipids, placing exceptionally high demands on lipid biosynthesis within oligodendrocytes to support its formation and maintenance [17, 18]. Therefore, it is likely that TANGO2 deficiency may adversely affect myelin formation or maintenance in the CNS.

In this study, we generated constitutive and cell‐type‐specific Tango2 knockout mouse models. We found that both constitutive and oligodendrocyte‐specific knockout of Tango2 result in motor deficits and cerebellar myelin loss. Single‐nucleus RNA sequencing and lipidomic analyses revealed that phospholipid disruption underlies these myelin defects. Finally, we demonstrated that vitamin B5 treatment improves motor performance and cerebellar myelination in Tango2 knockout mice, thereby paving the way for potential treatments targeting the myelination pathway in TDD‐associated neurological impairments.

2. Materials and Methods

2.1. Mice

All mice were of the C57BL/6J background and were housed in the barrier facility at Peking University Shenzhen Graduate School. The Institutional Animal Care and Use Committee approved all experimental protocols. To generate Tango2 constitutive knockout (KO) mice, exon 3–exon 9 was deleted via the CRISPR/Cas9 technique by insertion of guide RNA 5′‐Taagattgattacattgggcagg‐3′ at intron 2 and guide RNA 5′‐Aactccggggactctgcagaggg‐3′ at the 3′ flanking region of the TANGO2 gene in the mouse genome. The wildtype (WT) and KO mice used for the experiment were all derived from the heterozygous Tango2 KO mice. The Tango2 conditional knockout mice (Tango2 cKO) were generated by inserting two loxP sites at intron 2 and intron 3 to flox exon 3 via homologous recombination and Flp‐mediated recombination. The Tango2 cKO mouse line was crossed with NG2‐Cre^ERT (JAX# 008538)/Aldh1l1‐Cre^ERT2 (JAX#029655)/Vglut2‐Cre (JAX#028863)/Gad2‐Cre (JAX#010802) mice to obtain cell‐specific Tango2 knockout lines. The NG2‐Cre^ERT line was a generous gift from Dr. Feng Mei. Unless otherwise specified, all mice used in this study were aged 2–3 months. Male and female mice were analyzed separately in behavioral tests, whereas male mice were used in all other experiments except where indicated, as no significant sex differences were observed in key behavioral or molecular phenotypes.

To induce Cre recombination, Tamoxifen (Sigma‐Aldrich, T5648) was dissolved in corn oil at 10 mg/mL. For the NG2‐Cre^ERT; Tango2 cKO mice, Tamoxifen was administered by oral gavage to P7 mice at 50 mg/kg and to P56 mice at 100 mg/kg for 4 consecutive days. For the Aldh1l1‐Cre^ERT2; Tango2 cKO mice, Tamoxifen at 80 mg/kg was administered to P7 mice by subcutaneous injection for 3 days and to P56 mice by intraperitoneal injection for 5 days.

2.2. Behavior Tests

The behavioral tests included the Rotarod, Balance beam, Gait analysis, Grip strength, weight‐loaded swimming, Open field test, and Novel object recognition. They were conducted as in our previous study [19]. The Morris water maze was conducted as previously described [20], and the Fear conditioning test was performed using a previously published paradigm [21].

2.2.1. Rotarod Test

An accelerating rotarod test was conducted over four days, with three trials per day, using RotaRod for Mice (Ugo Basile, 47650). The rotarod speed was set to uniformly accelerate from 4 rpm to 40 rpm within 300 s on the first and second days, and from 8 rpm to 80 rpm within 300 s on the third and fourth days. The time taken for the mouse to fall off and the rotarod's terminal speed at that moment were recorded for each trial. The initial coordination (intercept) and motor learning rate (slope) were obtained by linear regression of terminal speed versus 12 trials.

2.2.2. Balance Beam

The balance beam test was conducted using a narrow beam, 1 m in length, positioned 50 cm above the ground, with a dark chamber at the beam's terminus for mice to enter. Four types of narrow beams were provided, with square and round cross‐sections measuring 20 and 10 mm in width/diameter. After 3 consecutive days of training to traverse a 20 mm‐wide beam, mice were tested on four beams of progressively increasing difficulty; the time to cross each beam and the number of hindlimb slips were recorded.

2.2.3. Gait Analysis

Gait parameters were analyzed through footprint patterns. The forepaws and hindpaws of mice were coated with different‐colored pigments, and mice were allowed to traverse a narrow corridor covered with white paper to leave footprints.

2.2.4. Grip Strength and Weight‐Loaded Swimming

Grip strength was measured using a grip strength meter (PUYAN, HPF‐5). Muscle strength in the four limbs was calculated as the mean of three tests. The weight‐loaded swimming test was conducted by attaching a lead wire equal to 10% of the mouse's body weight to its tail and recording the time until exhaustion and sinking in the water, which was used as the endurance time.

2.2.5. Open Field Test

Mice were placed in the center of an open field arena (45 cm × 45 cm × 45 cm) to explore freely and were monitored by a video camera connected to the automated tracking system (Viewer 3.0) for 30 min. The total distance traveled, movement velocity, and time spent in the center area were analyzed.

2.2.6. Morris Water Maze

Briefly, mice were tested in a circular water maze with a diameter of 1 m and a height of 50 cm, and their movements were tracked using a video recorder with an automated tracking system (ANY‐maze 6.33). On day 1, a platform was placed 1 cm above the water surface, and mice were placed into the water from a fixed starting position. Escape latency and swimming speed were recorded. From day 2 to day 6, the platform was submerged. Mice were tested four times per day from random starting positions, and escape latency was recorded. On day 7, the platform was removed, and the number of crossings over the platform location was recorded during the probe trial. The platform was then relocated to the opposite quadrant, and reversal learning and memory were assessed following the same procedure.

2.2.7. Novel Object Recognition

Novel Object Recognition task was conducted in an open field arena (45 cm × 45 cm × 45 cm). The day before the test, mice were allowed to explore the open field freely for 10 min to acclimatize. On the first day, two identical objects were placed into the arena, and mice were given 10 min of free exploration. The next day, one familiar object was replaced with a novel object of similar size but a distinct shape and color, and the mice were allowed to explore for 10 min, with their exploration time for each object tracked and recorded. The discrimination index was calculated as the ratio of novel object exploration time to total exploration time for both objects on day 2.

2.2.8. Fear Conditioning Test

On the first day, mice were placed in the conditioning chamber and presented with white noise as the conditioned stimulus (CS), followed by a foot shock as the unconditioned stimulus (US) during the final 2 s of the sound. The CS‐US pairing was repeated 3 times to facilitate fear association (conditioning test). 24 h later, mice were returned to the same conditioning chamber and allowed to explore freely without CS‐US presentations to measure contextually conditioned fear (context test). The following day, mice were placed in a conditioning chamber with a new context and were presented with the CS to measure conditioned fear to auditory cues (cued test). Freezing behavior was tested using the FreezeFrame system (Coulbourn Instruments, Allentown, PA).

2.3. Metabolic Cages and Lean Mass Test

Metabolic and behavioral phenotyping was measured using the Promethion Core Mouse Metabolic System (Sable Systems). Mice were placed in metabolic cages for 12 h to acclimatize and housed for another 3 days for data analysis. Lean mass and lean ratio of the whole body were detected using InAlyzer DEXA systems (MEDIKORS).

2.4. Vitamin B5 Treatment

Vitamin B5 (D‐Pantothenic acid, P5155, Sigma‐Aldrich) was dissolved in sterile phosphate‐buffered saline (PBS) at a concentration of 25 mg/mL and administered by oral gavage at a dose of 7 mg/day for two consecutive months. Administration was ceased one week before behavioral testing. The dosage was based on the upper intake level for adults recommended by the Institute of Medicine's Food and Nutrition Board.

2.5. Quantitative Real‐Time PCR

Total mRNA was extracted from fresh brain tissues using TRIzol reagent (Thermo Fisher Scientific, 15596018CN) and reverse transcribed into cDNA using TransScript II One‐Step gDNA Removal and cDNA Synthesis SuperMix (TransGen Biotech, AH311‐02). Quantitative Real‐Time PCR (RT‐PCR) was performed using TransStart Top Green qPCR SuperMix (TransGen Biotech, AQ131‐01) with gene‐specific primers (Table S1). Relative mRNA levels were normalized to Gapdh or 18S rRNA and calculated using the 2^−ΔΔCt method.

2.6. Western Blot

Proteins were extracted from fresh brain tissue using RIPA Lysis and Extraction Buffer (Thermo Fisher Scientific, 89900), and concentrations were quantified by BCA Protein Assay Kit (Thermo Fisher Scientific, 23225). The protein mixture with loading buffer (TransGen Biotech, DL101‐02) was boiled for 5 min, and 100 μg of protein was then separated on 4%–20% Bis‐Tris gels (GenScript SurePAGE, M00655). Polyvinylidene fluoride (PVDF) membranes (Millipore, ISEQ00010) were used to transfer proteins, then blocked with 5% BSA (dissolved in Tris‐buffered saline containing 0.1% Tween‐20) for 1 h. Then, membranes were incubated with primary antibodies of target and control (GAPDH or β‐Actin) overnight at 4°C. After washing, anti‐rabbit and anti‐mouse secondary antibodies were incubated with the membrane for 1 h at room temperature. The immunoreactive bands were captured using the Azure 600 Western Blot imaging system and quantified in ImageJ, with protein expression levels normalized to the loading control. Antibodies used for the Western blot are shown in Table S2.

2.7. Immunohistochemistry

Mice were anesthetized with isoflurane (RWD, R510‐22‐10) and sequentially perfused with ice‐cold PBS and 4% paraformaldehyde (PFA). The intact brain region was dissected, postfixed at 4°C overnight, and dehydrated for 48 h in 30% sucrose‐PBS before being embedded in OCT medium (Sakura, 4583). Frozen sections of 30 μm thickness were collected and blocked with 5% BSA (dissolved in PBS containing 0.3% Triton X‐100) for 30 min, then incubated overnight at 4°C with primary antibodies. Sections were washed with PBS, then incubated with DAPI and secondary antibodies for 1 h at room temperature before being mounted onto glass slides. Immunofluorescence images were acquired using a confocal microscope (Nikon A1R), with acquisition parameters held constant across experimental and control groups within the same batch. Quantitative fluorescence analysis was performed using ImageJ. Antibodies used for immunofluorescence are shown in Table S2.

2.8. Hematoxylin and Eosin Staining

Mice were anesthetized and perfused according to the immunohistochemistry protocol. The cerebellum was dissected and post‐fixed in 4% PFA for 24 h, and then embedded in paraffin. Sagittal sections of the cerebellum were cut at a thickness of 6 μm, and staining was performed in accordance with the manufacturer's instructions of Beyotime Hematoxylin and Eosin Staining Kit (C0105S).

2.9. Electron Microscopy

After sequential perfusion with 50 mL of 37°C PBS, 50 mL of 37°C 4% PFA, and 300 mL of ice‐cold 4% PFA, the mouse cerebellum was promptly dissected. A 1 mm³ cube of cerebellar white matter was fixed in 2.5% glutaraldehyde overnight and postfixed with 1% osmium tetroxide for 2 h, dehydrated with a graded acetone series, and embedded in EPON 812. Ultrathin sections of 70 nm were cut using an ultramicrotome (Leica EM UC7) and double‐stained with 2% uranyl acetate‐lead citrate before observation under a transmission electron microscope (Jeol JEM‐1400Flash). ImageJ was used to quantify myelin sheaths and axons.

2.10. Single‐Nucleus RNA Sequencing

BGI Genomics in China performed single‐nucleus RNA sequencing (snRNA‐seq). Briefly, nuclei were isolated following tissue dissociation and subjected to fluorescence‐activated sorting. Libraries were generated using 10× Genomics Chromium platform, followed by reverse transcription, cDNA amplification, and library construction according to the manufacturer's protocol. Sequencing data were processed using Cell Ranger (v6.1.2), and reads were aligned to the mm10 reference genome. Cells with fewer than 200 detected genes, more than 5% mitochondrial gene content, or genes detected in fewer than three cells were excluded from downstream analysis. Data integration and batch correction were performed using canonical correlation analysis (CCA) in Seurat (v5.1.0). Cell clustering was conducted based on k‐nearest neighbor (KNN) and shared nearest neighbor (SNN) graphs using the Louvain algorithm. Dimensionality reduction was performed using t‐distributed stochastic neighbor embedding (t‐SNE). Cell–cell communication analysis was performed using CellChat (v1.6.1), and pathway activity was quantified using AUCell (v1.29.0). Functional enrichment analysis was conducted using clusterProfiler (v4.19.5). Potential transcriptional perturbations were predicted using ClueQuery (v0.3.5).

2.11. Lipidomic Profiling

BGI Genomics in China performed untargeted lipidomics. Brain lipids were extracted using dichloromethane‐methanol (3:1, V/V), then separated and detected via a Waters UPLC I‐Class Plus (Waters, USA) coupled with a Q‐Exactive high‐resolution mass spectrometer (Thermo Fisher Scientific, USA). Mass spectrometry data were first analyzed using LipidSearch v.4.1 (Thermo Fisher Scientific, USA) for peak extraction, lipid identification, and peak alignment, and then imported into metaX for processing, including missing‐value imputation, normalization, and quality‐control analysis. Differential lipids were identified using multivariate (VIP values from the first two principal components of the PLS‐DA model) and univariate (fold‐change and q‐value) analyses, with VIP ≥ 1 and q‐value < 0.05 considered statistically significant.

2.12. Statistics

Statistical analysis and visualization were performed using GraphPad Prism (v 10.3.1) with results shown as mean ± SEM. Comparisons between two groups were performed using unpaired two‐tailed Student's t‐tests. Multiple group comparisons were analyzed using one‐way ANOVA with Tukey's post hoc test, while two‐way ANOVA with Bonferroni's post hoc test was used for repeated measures or grouped data. p‐value < 0.05 was considered statistically significant, and unlabeled comparisons indicate p ≥ 0.05 in all analyses. Each experiment was independently repeated at least three times to ensure reliable results.

3. Results

3.1. Constitutive TANGO2 Deficiency Leads to Motor Deficits

There are nearly 30 distinct variants associated with TANGO2 deficiency in humans, among which deletion of exons 3–9 is the most prevalent pathogenic variant [1, 2]. Therefore, we first generated a constitutive Tango2 knockout (KO) mouse model by deleting exons 3–9 (Figure 1A). RT‐PCR and immunoblotting analyses confirmed the complete loss of Tango2 mRNA and TANGO2 protein in Tango2 KO mice (Figure 1B,C). No changes were observed in gross development, energy metabolism profiles, or serum biochemical markers associated with metabolic crisis under basal conditions in Tango2 KO mice (Figure S1A–C), suggesting that TANGO2 is dispensable for gross development and basal metabolism in mice.

FIGURE 1.

TANGO2 deficiency leads to motor deficits. (A) Schematic diagram of human TANGO2 variant and mouse Tango2 knockout strategy. (B) Tango2 mRNA in WT and Tango2 KO mouse brains. (C) TANGO2 protein in WT and Tango2 KO mouse brain lysates. * Indicates non‐specific bands. (D) Schematic of the accelerating rotarod test. (E) The terminal speed of the rotarod when mice fell off across 12 trials, the calculated initial coordination, and the learning rate for males and females. (F) Schematic of the balance beam test. (G) The latency to cross beams and the number of footslips during traverse for males and females. (H) Illustration of gait analysis and measurements for footprints. (I) Quantitative analysis of gait parameters for males and females. (J) Grip strength for males and females. (K) Endurance time in 10% weight‐loaded swimming test for males and females. Data are means ± SEM. *p < 0.05; ***p < 0.001.

Motor dysfunction is a prominent feature of neurological impairment in individuals with TDD, including motor delays, gait abnormalities, poor coordination, and ataxia [1, 2, 22]. To determine whether Tango2 KO mice exhibit motor deficits, we first employed the accelerated rotarod assay to assess motor performance. Mice of both sexes were tested over four consecutive days, with rotation speed increasing from 4 to 40 rpm on the first two days and from 8 to 80 rpm on the final two days (Figure 1D). The terminal speed at which Tango2 KO mice fell from the rod was significantly lower than that of control littermates. Linear regression analysis of terminal speed versus trial number for each mouse revealed that Tango2 KO mice exhibited significantly impaired initial motor coordination but comparable rates of motor learning (Figure 1E). We next used the balance beam test to evaluate fine‐balance control. Tango2 KO mice of both sexes took longer to cross a 10 mm‐wide beam and had more foot slips (Figure 1F,G), indicating impaired motor control during highly demanding balance tasks. These deficits in motor coordination and balance were not attributable to gait abnormalities or muscle dysfunction, as Tango2 KO mice showed no alterations in gait analysis (Figure 1H,I), grip strength (Figure 1J), time to exhaustion in a 10% body weight‐loaded swimming test (Figure 1K), lean mass (Figure S1D), or locomotor activity in the open field test (Figure S1E), when compared with WT mice. Collectively, these results demonstrate that Tango2 KO mice exhibit pronounced deficits in motor coordination and balance.

Given that a subset of TDD individuals exhibits cognitive impairments [1, 2, 4, 23], we assessed whether Tango2 KO mice display similar deficits. We performed three cognition‐related behavioral assays. First, we used the Morris water maze to evaluate spatial learning and memory (Figure 2A). Tango2 KO mice of both sexes showed a comparable, progressive reduction in escape latency across trials and a similar number of crossings over the original platform location, indicating intact spatial memory consolidation (Figure 2B,C). Second, we used the novel object recognition test to assess recognition memory and found that Tango2 KO mice of both sexes displayed a preference for the novel object similar to that of littermate controls (Figure 2D,E). Third, using the fear conditioning test to evaluate associative fear learning and memory, we found that Tango2 KO mice of both sexes retained normal fear memory capacity (Figure 2F,G). Taken together, these data demonstrate that Tango2 KO mice do not have major cognitive deficits.

FIGURE 2.

TANGO2 deficiency does not affect cognitive behaviors. (A) Time axis diagram of Morris water maze test. (B) Representative swimming trajectories of mice in the hidden probe trial on day 7 and the reversal probe trial on day 13. (C) Morris water maze results for male and female mice. Left to right: Escape latency and swimming speed in the visible platform trial, escape latency in hidden and reversal platform trials, and the platform crossover number in hidden and reversal probe trials. (D) Schematic diagram of the novel object recognition test. (E) The time mice spent interacting with old objects and novel objects on day 2, and the discrimination index for males and females. (F) Schematic diagram of the fear conditioning test. (G) Percentages of freezing time during conditioning test, context test, and cued test for males and females. Data are means ± SEM; **p < 0.01; ***p < 0.001.

3.2. Oligodendroglial TANGO2 Deficiency Impairs Motor Behavior

The selective impairment of motor but not cognitive behaviors in Tango2‐deletion mice suggests two likely mechanisms: (1) TANGO2 is selectively expressed in cells essential for motor behaviors, or (2) the motor‐related cells are more sensitive or vulnerable to Tango2 deletion. To test the first possibility, we quantified TANGO2 protein levels in lysates collected from distinct brain tissues. We found that TANGO2 is expressed at similar levels across the brain regions collected (Figure S2A), indicating that motor deficits are unlikely to result from the unique expression of TANGO2 in specific motor‐related circuits. We then employed whole‐brain rotarod‐induced c‐Fos mapping to narrow down brain regions involved in Tango2‐deletion‐induced motor deficits (Figure 3A). Our analysis revealed that Tango2 deletion significantly increased c‐Fos signals in the cerebellum, particularly cerebellar white matter (Figure 3B), but not in other brain regions examined (Figure S2B,C), suggesting a potential role of the cerebellum in the motor impairments of Tango2 KO mice.

FIGURE 3.

Oligodendroglial TANGO2 deficiency impairs motor behavior. (A) Schematic diagram of the rotarod coupled c‐Fos expression and staining. (B) Representative images and quantitative results of c‐Fos immunostaining in the cerebellum. Arrowheads indicate aberrant c‐Fos positive cells in the WM. n = 12 sections/4 mice. Scale bar: 200 μm. GL, granular layer; ML, molecular layer; WM, white matter. (C) Diagram of cells in cerebellar cortex. (D) Rotarod test results of NG2‐Cre^ERT; Tango2 cKO mice with tamoxifen injection at P7. TANGO2 was ablated in oligodendrocyte progenitor cells (OPCs) and mature oligodendrocytes (ODCs). (E) Rotarod test results of Aldh1l1‐Cre^ERT2; Tango2 cKO mice with tamoxifen injection at P7. (F) Rotarod results of Gad2‐Cre; Tango2 cKO mice. (G) Rotarod results of Vglut2‐Cre; Tango2 cKO mice. Data are means ± SEM. *p < 0.05; ***p < 0.001.

The cerebellar white matter is primarily composed of dense fiber tracts and oligodendroglial lineage cells, including oligodendrocyte progenitor cells (OPCs) and mature oligodendrocytes (ODCs). To identify the specific cell type involved in cerebellar dysfunction, we generated conditional knockout (cKO) mouse models with Tango2 specifically deleted in OPCs (NG2‐Cre^ERT; Tango2 cKO), astrocytes (Aldh1l1‐Cre^ERT2; Tango2 cKO), inhibitory neurons (Gad2‐Cre; Tango2 cKO), or excitatory neurons (Vglut2‐Cre; Tango2 cKO), covering the major cerebellar cell types (Figures 3C and S3A,B). Behavioral analysis showed that Tango2 deletion in both sexes in OPCs, induced at postnatal day 7 (P7) using NG2‐Cre^ERT, resulted in significant impairment in rotarod performance (Figure 3D). In contrast, deletion of Tango2 in astrocytes, inhibitory neurons, or excitatory neurons had minimal to no effect on motor performance in rotarod tests (Figures 3E–G and S3D). To determine whether TANGO2 deficiency in adult OPCs leads to similar motor deficits, we induced Tango2 deletion in OPCs at P56 using NG2‐Cre^ERT; Tango2 cKO mice and observed no apparent motor deficits (Figure S3C). This indicates that TANGO2 plays an essential role in early OPCs, whose differentiation and maturation govern ODC function and cerebellar myelination. Together, these results suggest that cerebellar oligodendrocytes predominantly contribute to TANGO2‐deficiency‐induced motor deficits.

3.3. TANGO2 Deficiency Induces Cerebellar Myelin Loss

Oligodendrocytes are the myelinating cells of the CNS; we next examined whether Tango2 deletion affects cerebellar myelin levels. Using electron microscopy to analyze the myelin sheath in Tango2 KO mice, we observed a significant reduction in myelin sheath thickness (Figure 4A,B) and a decreased density of myelinated axons in the cerebellar white matter, particularly among small‐diameter axons (Figure 4C). Consistent with these findings, immunohistochemical staining for myelin basic protein and myelin oligodendrocyte glycoprotein revealed reduced myelin in the cerebellar white matter and granule layer of Tango2 KO mice (Figure 4D,E). Additionally, mRNA levels of myelin protein genes (Mog and Plp1) and myelin lipid‐related genes (Ugt8 and Aspa) were all downregulated in cerebellar lysates from Tango2 KO mice (Figure S4A). A similar reduction in cerebellar myelin was also observed in female Tango2 KO mice (Figure S4B,C). Together, these data demonstrate a substantial loss of myelin in the Tango2 KO cerebellum.

FIGURE 4.

TANGO2 deficiency induces cerebellar myelin loss. (A) Representative electron micrographs of myelinated fibers in the cerebellar WM from WT and Tango2 KO mice. Scale bars: up, 5 μm; down, 2 μm. (B) Quantification of mean g‐ratio (n = 15 sections/5 mice) and g‐ratio distribution across different axonal diameters (WT with 4146 axons; KO with 2789 axons; linear regression analysis). (C) Mean density of myelinated axons and density of myelinated axons grouped by axonal diameters. n = 15 sections/5 mice. (D) Representative images of cerebellar sections immunostained for MBP and MOG from WT and KO mice. (E) Quantification of MBP‐positive and MOG‐positive areas in the GL and WM of cerebellar lobule IV/V. (F) Representative images of cerebellar sections immunostained for CC1, PDGFRα, and OLIG2 from WT and KO mice. (G) Quantification of CC1⁺OLIG2⁺ and PDGFRα⁺OLIG2⁺ cells in the GL and WM of cerebellar lobule IV/V. (H, I) Representative images (H) and quantification (I) of MBP‐positive and MOG‐positive areas in the GL and WM of cerebellar lobule IV/V from Ctrl and NG2‐Cre^ERT; Tango2 cKO mice. (J, K) Representative images (J) and quantification (K) of CC1⁺OLIG2⁺ and PDGFRα⁺OLIG2⁺ cells in the GL and WM of cerebellar lobule IV/V from Ctrl and NG2‐Cre^ERT; Tango2 cKO mice. D, F, H, J: Scale bar 100 μm. E, G, I, K: n = 12 sections/4 mice. Data are means ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001.

We then investigated whether this myelin deficit results from impaired oligodendrocyte differentiation or a specific defect in myelin formation. To distinguish between these possibilities, we quantified the numbers of ODCs (CC1⁺‐OLIG2⁺) and OPC (PDGFRα⁺‐OLIG2⁺) in the Tango2 KO cerebellum (Figure 4F). Quantitative analysis showed that Tango2 deletion did not alter the differentiation trajectory of oligodendrocytes (Figure 4G), indicating no differentiation defect. Consistently, oligodendrocyte‐specific Tango2 KO mice exhibited cerebellar myelin reduction without changes in oligodendrocyte numbers (Figure 4H–K). Moreover, we did not observe significant myelin loss in other brain regions examined (Figure S4D), supporting the selective susceptibility of cerebellar oligodendrocytes to TANGO2 deficiency. Overall, these findings demonstrate that TANGO2 deficiency in oligodendrocytes induces defective cerebellar myelination.

3.4. TANGO2 Deficiency Triggers Synapse Remodeling

Previous reports have shown that myelin deficiency can regulate synapse number [24, 25, 26]. We therefore investigated whether Tango2 KO mice exhibit alterations in synapse number. In the cerebellar cortex, Purkinje cells receive excitatory inputs from vGluT1‐positive parallel fibers and vGluT2‐positive climbing fibers, as well as inhibitory inputs from vGAT‐positive interneurons [27, 28, 29]. By analyzing these synaptic markers in the cerebellar cortex of Tango2 KO mice, we found that vGluT1 intensity and vGluT2 puncta density were increased (Figure 5A,B), indicating a rise in excitatory presynaptic terminals. The size of vGluT2 puncta was also larger (Figure 5B). Consistent with the essential role of oligodendrocyte‐specific Tango2 deletion in motor deficits, we observed a similar enhancement in vGluT1 intensity and vGluT2 puncta density in NG2‐Cre^ERT; Tango2 cKO mice (Figure 5C,D). In contrast, no changes were detected in the size or density of vGAT puncta in the cerebellar cortex of either Tango2 KO or NG2‐Cre^ERT; Tango2 cKO mice (Figure 5E). Interestingly, we noted an increase in glutamate–aspartate transporter‐1 (GLAST‐1) (Figure S5A), likely reflecting a compensatory response to maintain glutamate homeostasis.

FIGURE 5.

TANGO2 deficiency triggers synapse remodeling. (A) Representative images of vGluT1 immunostaining and quantification of vGluT1 intensity in WT and KO mice. (B) Representative images of vGluT2 immunostaining and quantification of vGluT2 puncta intensity, size, and density in WT and KO mice. (C, D) Representative images of vGluT1 and vGluT2 immunostaining and quantification of puncta intensity, size, and density in Ctrl and NG2‐Cre^ERT; Tango2 cKO mice. (E) Representative images of vGAT immunostaining and quantification of vGAT puncta intensity, size, and density in WT/KO and Ctrl/NG2‐Cre^ERT; Tango2 cKO mice. All quantifications are performed separately for the superficial (0%–50%) and deep (50%–100%) molecular layers. Scale bar: 20 μm. Data are means ± SEM; n = 12 sections/4 mice. *p < 0.05; **p < 0.01; ***p < 0.001.

Additionally, we observed prominent reactive astrogliosis in the cerebellum of both Tango2 KO mice and NG2‐Cre^ERT; Tango2 cKO mice (Figure S5B,C), without evidence of obvious microglial activation, elevated inflammatory markers, or disruption of overall cerebellar architecture (Figure S5D–F). Together, these findings indicate that TANGO2 deficiency in oligodendrocytes triggers remodeling of excitatory synapses, which likely contributes to the motor deficits observed in Tango2 KO mice and individuals with TDD.

3.5. Disrupted Transcription of Genes Involved in Phospholipid Metabolism After Tango2 Deletion

To investigate the mechanisms underlying TANGO2‐mediated myelination, we performed single‐nucleus RNA sequencing (snRNA‐seq) on TANGO2‐deficient cerebellar tissue (Figure 6A). We acquired 65 644 single‐nucleus transcriptomic profiles. We identified 12 major cerebellar cell types using t‐SNE clustering with established cell markers (Figure 6B,C). Analysis of differentially expressed genes (DEGs) within each cell type revealed the highest number of DEGs in granule cells (862), followed by mature oligodendrocytes (ODCs; 537 DEGs) (Figure 6D). Gene Ontology (GO) analysis of DEGs in granule cells, Bergmann glia, and astrocytes indicated disruptions in multiple processes (Figure S6A–C). Cell communication analyses further revealed altered cell crosstalk between granule cells and glial cells (Figure S6D–F). Enrichment analysis of DEGs in the ODC population showed profound dysregulation in biological processes (BPs), molecular functions (MFs), cellular components (CCs), and KEGG signaling pathways that converge on defective myelination (Figure 6E). Among the significantly enriched biological processes, multiple lipid‐related pathways were prominently overrepresented (Figure 6F). Analysis of enrichment scores (ESs) for these lipid‐related processes demonstrated that phospholipid‐related pathways—including phospholipid metabolism, phospholipid catabolism, phospholipid translocation, and phospholipid transport—showed the most significant degree of downregulation in TANGO2‐deficient ODCs (Figure 6G). Thus, snRNA‐seq profiling identifies ODCs as among the most vulnerable cell types to Tango2 deletion, with selective impairment of phospholipid metabolic processes.

FIGURE 6.

TANGO2 deficiency disrupts transcription of genes involved in phospholipid metabolism. (A) Schematic of snRNA‐seq for the mouse cerebellum. n = 2 mice. (B) t‐SNE representation of single nuclei of cerebella colored by cell type identity. (C) Expression and specificity of marker genes used for cell type identification. (D) Cluster‐specific volcano plot of DEGs across 12 cell types. (|log2FC| > 0.25, adjusted p < 0.01). (E) GO and KEGG analysis of DEGs in the ODC cluster (adjusted p < 0.05). (F) Enrichment analysis of lipid‐related processes in the ODC cluster. (G) Enrichment scores of phospholipid‐related biological processes across all cell types. Gray dots indicate no significant difference (p > 0.05). For granule cells, WT with 31 875 nuclei and KO with 27 135 nuclei. For ODCs, WT with 511 nuclei and KO with 608 nuclei. For Golgi cells, WT with 97 nuclei and KO with 77 nuclei.

3.6. TANGO2 Is Required for Phosphatidylethanolamine De Novo Synthesis

To confirm phospholipid disruption in Tango2 KO oligodendrocytes, we performed lipidomic analysis to characterize alterations in lipid profiles (Figure 7A). Lipidomic profiling of cerebellar tissue revealed that phospholipids constitute the predominant metabolite class and the three most abundant phospholipids were phosphatidylcholine (PC), phosphatidylethanolamine (PE), and phosphatidylserine (PS) (Figure 7B). There was no significant difference in the relative abundance of lipid subclasses between WT and KO samples (Figure S7A). Tango2 deletion resulted in 54 downregulated and 51 upregulated lipid metabolites (Figure 7C). At the lipid subclass level, we observed significantly reduced levels of diglyceride (DG), PE, PS, and phosphatidylmethanol (PMe), alongside increased levels of lyso‐phosphatidylinositol (LPI), ganglioside GM3 (GM3), and ganglioside GT1a (GT1a) (Figure 7D). DG, PE, and PS are primary constituents of myelin lipids, whereas the low‐abundance PMe, LPI, GM3, and GT1a primarily function as metabolic byproducts or lipid signaling molecules. In Tango2 KO mice, lipids containing saturated and monounsaturated fatty acids (SFA/MUFA) were generally downregulated, whereas phospholipids enriched with polyunsaturated fatty acids (PUFA) were significantly upregulated (Figure 7E). A coordinated downregulation was observed not only in the total content of DG, PE, and PS, but also in individual lipids sharing the same fatty acyl chain length and degree of unsaturation (Figure 7E), indicating that TANGO2 deficiency disrupts the DG–PE–PS metabolic axis.

FIGURE 7.

TANGO2 is required for phosphatidylethanolamine de novo synthesis. (A) Schematic of lipidomic profiling procedure for the mouse cerebellum. WT, n = 13 mice; KO, n = 11 mice. (B) Percentage of lipid subclasses in WT and KO cerebella. (C) Volcano plot of all lipids detected. The differential lipids between KO and WT with p < 0.05 are highlighted in blue and red. The top 3 downregulated and upregulated lipids are labeled. (D) Quantitative comparison of lipid subclasses between KO and WT mice. (E) Heatmap of differential lipids in DG, PE, and PS subclasses. (F) Summary of PE metabolic pathways. (G) Transcriptional levels of enzymes in the CDP‐Etn pathway for PE and PS synthesis. (H) Density plot of the PE biosynthetic process enrichment scores in ODCs. WT, n = 511 ODCs; KO, n = 608 ODCs. (I) Potential PE targets for reversing ODC transcriptome alterations identified through pharmacogenomics (score > 1). (J) Motor performance of VB5‐treated mice on the rotarod at 8 to 80 rpm. Male: WT PBS, n = 11; KO PBS, n = 7; KO VB5, n = 9. Female: WT PBS, n = 9; KO PBS, n = 14; KO VB5, n = 12. Acyl‐CoA, acyl‐coenzyme A; CDP‐Cho, cytidine diphosphate choline; CDP‐Etn, cytidine diphosphate ethanolamine; Cho, choline; DG, diacylglycerol; Etn, ethanolamine; G3P, glycerol‐3‐phosphate; LPA, lysophosphatidic acid; LPE, lysophosphatidylethanolamine; P‐Cho, Phosphocholine; P‐Etn, phosphoethanolamine; PA, phosphatidic acid; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine. Data are means ± SEM; *p < 0.05; **p < 0.01.

In eukaryotic cells, DG, PE, and PS form a sequential metabolic cascade: DG serves as a direct precursor for de novo synthesis of PE via the CDP‐ethanolamine (Kennedy) pathway, and PE is further converted to PS through a base‐exchange reaction (Figure 7F). Quantitative transcriptional analysis of enzymes in the DG–PE–PS metabolic axis revealed that enzymes mediating ethanolamine (Etn) activation to CDP‐ethanolamine (CDP‐Etn), DG biosynthesis from glycerol‐3‐phosphate (G3P), the condensation of DG (as the acyl backbone donor) with CDP‐Etn to generate PE, as well as phosphatidylserine synthase 2 (PSS2), which catalyzes the direct conversion of PE to PS, were all significantly reduced—demonstrating a collapse of the DG–PE–PS metabolic axis (Figure 7G). Within the PE metabolic network, the CDP‐Etn pathway serves as the major contributor to PE biosynthesis, while three additional pathways contribute to salvage synthesis and interconversion of PE: lysophosphatidylethanolamine (LPE) acylation, PS decarboxylation, and PC‐to‐PE conversion via a PS intermediate (Figure 7F). However, quantification of enzymes involved in these alternative pathways showed no significant alterations in response to TANGO2 deficiency (Figure S7B). Notably, the lipid abundance of PC and the PC biosynthetic pathway—which shares DG with the CDP‐Etn pathway—remained unaffected, supporting a specific regulatory role for TANGO2 in de novo PE synthesis. Transcriptomic profiling of ODCs revealed that TANGO2 deficiency severely inhibits PE biosynthesis (Figure 7H), indicating a functional breakdown of the PE biosynthetic pathway.

Furthermore, we screened pharmacological treatments and genetic interventions that could reverse transcriptional abnormalities in Tango2‐deleted ODCs (Figure S7C). We identified multiple drug or intervention targets that are highly enriched in PE‐related metabolic processes (Figure 7I). Vitamin B5 (VB5), a precursor of coenzyme A (CoA), has been reported to rescue multiple defects in TANGO2‐deficient Drosophila by potentially increasing CoA metabolite levels and restoring lipid homeostasis [30]. We found that two months of VB5 treatment significantly improved motor performance on the rotarod in Tango2 KO mice, particularly during the high‐speed stage of trials (Figures 7J and S7D). Moreover, immunofluorescence of cerebellar myelin showed a partial restoration of myelin density in the VB5‐treated Tango2 KO mice (Figure S7E). Together, lipidomic analyses reveal an abnormal lipid profile in the TANGO2‐deficient cerebellum, characterized by defects in the DG–PE–PS axis. Specific regulation of de novo PE synthesis by TANGO2 indicates that this process is a potential therapeutic target for TDD, as exemplified by VB5 treatment here.

4. Discussion

In this study, we established TDD mouse models with Tango2 deleted either globally or in distinct CNS cell types. We found that mice with global or oligodendrocyte‐specific deletion of Tango2 exhibit motor deficits and cerebellar myelin loss. Further investigation revealed impaired lipid metabolism in oligodendrocytes from TANGO2‐deficient mice. Finally, we demonstrated that vitamin B5 supplementation ameliorates myelin defects and motor deficits in TANGO2‐deficient mice. Together, our data suggest that TDD is a disorder with myelin disruption.

4.1. TDD and TANGO2‐Deficient Mouse Models

In this study, we found that TANGO2‐deficient mice exhibit robust motor deficits that recapitulate the neurological abnormalities observed in TDD—namely, dyskinesia characterized by impaired motor coordination and balance, as well as ataxia [1, 2]. The TANGO2‐deficient mice also show myelin deficiency and axon loss in the cerebellar white matter, mirroring the white matter hyperintensity and cerebellar white matter volume reduction observed in some TDD individuals [1, 2, 3, 5]. Although declines in motor capacity have been reported in TANGO2‐deficient fruit flies and zebrafish [13, 30], our study provides the first evidence that deletion of Tango2 within the CNS drives motor deficits, offering a valuable tool for further investigation into the mechanisms underlying TDD and its treatment. Under basal conditions, Tango2 KO mice did not develop spontaneous metabolic crises, consistent with the stress‐triggered nature of acute TDD episodes in patients, nor did they exhibit overt convulsive seizures throughout the study. Notably, TANGO2‐deficient mice exhibit minimal cognitive deficits, in contrast to those described in clinical TDD. This discrepancy may be attributed to the fact that intellectual deficits in patients are typically mild to moderate [1], rendering them difficult to detect using standard rodent behavioral batteries. In addition, species‐specific differences in the regional vulnerability of neural circuits to TANGO2 deficiency may further contribute to this divergence. Importantly, adaptive myelination underlying learning and memory is generally circuit‐dependent and occurs in cognitive processing brain regions, such as the medial prefrontal cortex (mPFC), ACC, hippocampus, and corpus callosum [31]. In contrast, myelination deficits in Tango2 KO mice are predominantly confined to the cerebellum, potentially accounting for the lack of overt cognitive impairment.

4.2. TDD Is a Disorder With Myelin Deficiency

Approximately 60% of individuals with TDD present with MRI abnormalities, with most reports documenting increased white matter signal intensity on T2‐weighted sequences and isointense or hyperintense signal intensity on T1‐weighted sequences [1, 2, 3]. These findings are consistent with features observed in hypomyelinating leukodystrophies (HLDs), suggesting potential deficits in myelin development [32]. Here, we demonstrate that deletion of Tango2 specifically in early oligodendrocyte progenitor cells (OPCs) results in severe myelin loss, providing the cytopathological evidence for oligodendrocyte dysfunction and myelin impairment in TDD. Therefore, we propose that TDD represents a myelin deficiency disorder [33]. Compared with global Tango2 knockout mice, NG2‐Cre^ERT; Tango2 cKO mice exhibit relatively milder motor deficits and myelin defects. One possible explanation is that the Tango2 deletion at postnatal day 7 (P7) used in this study may not sufficiently affect all oligodendrocytes, because a subset of wild‐type OPCs had already differentiated and formed standard myelin sheaths by that time [34, 35].

The cerebellar white matter contains myelinated axons from climbing fibers and mossy fibers. Demyelination of these axons reduces conduction speed and action potential amplitude [36], which may trigger presynaptic remodeling by increasing the number of presynaptic terminals and enhancing glutamate loading. The observed alterations in synapse number and astrocyte hyperplasia are likely compensatory responses to defective myelination in TANGO2‐deficient mice. Reactive astrocytes can promote the formation and maturation of excitatory synapses by secreting neurotrophic factors such as BDNF and GDNF, or by transporting glucose and lactate to excitatory axon terminals [37]—thereby supplementing presynaptic energy metabolism, supporting synaptic genesis, and collectively contributing to the increase in excitatory synapses. Interestingly, the number of inhibitory synapses is minimally affected in TANGO2‐deficient mice, suggesting that excitatory synaptic remodeling may represent a preferential adaptive response to impaired myelination.

4.3. TDD and Lipid Metabolism

TDD‐related imbalances in lipid homeostasis have been observed across multiple models. For example, knockdown of TANGO2 in HepG2 cells led to increased levels of lysophosphatidic acid (LPA) and reduced phosphatidic acid (PA), suggesting a defect in acyl‐CoA availability [7]. TANGO2‐deficient zebrafish exhibited a decrease in total lipids, particularly DG, TG, PC, and PE [13]. Primary fibroblasts from individuals with TDD showed impaired lipid depletion and elevated levels of unsaturated free fatty acids, neutral lipids, sphingomyelins, and phospholipids [14]. Our snRNA‐seq and lipidomic analyses revealed a collapse of the DG–PE–PS metabolic axis after Tango2 deletion in mice (Figures 6 and 7). The specific lipid perturbations vary among these models, likely reflecting differences in cellular membrane composition in different preparations.

DG serves as a carrier for the glycerol backbone and is a common substrate for the synthesis of both PE and PC. The reduction in DG levels does not affect PC content or the expression of enzymes specifically involved in PC synthesis (Figure 7D and S7B), highlighting the particular vulnerability of PE synthesis under conditions of insufficient DG availability. PE and its major subclass, ethanolamine plasmalogen, are predominant and significant components of myelin phospholipids. Oligodendrocytes are likely particularly vulnerable to TANGO2 deficiency due to the high demand for PE in myelin synthesis. In contrast, neurons and astrocytes have a relatively low demand for PE and likely possess compensatory pathways that may protect them from damage caused by Tango2 deletion.

A recent study identified TANGO2 as an acyl‐CoA‐binding protein [8]. Vitamin B5 supplementation—which may replenishes intracellular CoA pools—has been shown to exert beneficial effects in TANGO2‐deficient cells and fruit flies [14, 30]. Oral vitamin B5 intake for two months improved motor, language, and cognitive function in individuals and reduced white matter hyperintensities [15]. In this study, supplementation of TANGO2‐deficient mice with VB5 for two months led to significant improvements in motor performance and myelin recovery. Vitamin B5 likely supports myelin restoration by improving CoA‐dependent lipid metabolism. Our data not only support the beneficial effects of VB5 on human TDD symptoms but also identify oligodendrocyte myelination as a potential cellular substrate underlying its neurological benefits.

Author Contributions

Jiewen Chen: data curation, formal analysis, investigation, methodology, validation, visualization, writing – original draft. Zhili Liu and Fengling Chen: data curation, formal analysis, investigation, methodology, validation, visualization. Yongfei Cui, Liming Qin, Mengdi Wang, Xiangbin Zhu, Na Li, and Can Huang: investigation, validation. Bo Zhang and Kunfu Ouyang: conceptualization, funding acquisition, project administration, resources, supervision, writing – original draft, writing – review and editing.

Funding

This work was supported by Shenzhen Medical Research Fund (B2402022), National Science Foundation of China (91439130, 82170235, 82470245, 82022018, 32070958, and 82161138025), Major Program of Shenzhen Bay Laboratory (S241101002), Shenzhen‐Hong Kong Institute of Brain Science‐Shenzhen Fundamental Research Institutions (2019SHIBS0004), Guangdong Pearl River Funding.

Ethics Statement

All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) at Peking University Shenzhen Graduate School under ethical approval number 20190306‐01 and animal protocol approval number AP0017002.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: TANGO2 deficiency does not affect gross development or basal metabolism and does not cause overt muscle dysfunction in mice. (A) Body weight of mice at 2, 4, and 8 weeks of age. (B) Oxygen consumption, respiratory quotient, energy expenditure, and total movement distance were measured by metabolic cages over 60 h. (C) Serum levels of creatine kinase (CK), CK‐MB, lactate dehydrogenase (LDH), and aspartate aminotransferase (AST) in WT and Tango2 KO mice under basal conditions. (D) Muscle mass of the whole body. (E) Total distance traveled, average movement speed, distance traveled in the center zone, and time spent in the center zone for mice in the open‐field test. Data are means ± SEM.

Figure S2: TANGO2 deficiency selectively increases rotarod‐induced c‐Fos activation in the cerebellum. (A) Representative western blot analysis of TANGO2 expression in distinct brain regions. CB, Cerebellum; CTX, Cortex; HPF, Hippocampal formation; MB, Midbrain; OLF, Olfactory areas; P&MY, Pons& Medulla; STR, Striatum; TH, Thalamus; WB, Whole brain. (B) Representative immunostaining of c‐Fos in coronal brain sections, including ACC, Anterior cingulate cortex; M2, Secondary motor cortex; M1, Primary motor cortex; CP, Caudoputamen; ACB, Nucleus accumbens; HPF, TH, HY, MB, P, and MY. Scale bar: 200 μm. Data are means ± SEM; n = 18 sections/6 mice.

Figure S3: Adult‐stage Tango2 deletion in OPCs or astrocytes does not impair motor performance. (A) Schematic diagram of the Tango2 cKO strategy. (B) Representative western blot analysis of TANGO2 protein expression in the brains of Tango2 cKO mice. (C) Rotarod test results of NG2‐Cre^ERT; Tango2 cKO mice after tamoxifen injection at P56. (D) Rotarod test results of Aldh1l1‐Cre^ERT2; Tango2 cKO mice after tamoxifen injection at P56. Data are means ± SEM.

Figure S4: Tango2 deletion reduces cerebellar myelin while sparing non‐cerebellar myelin. (A) mRNA levels of Plp1, Mog, Ugt8, and Aspa in the cerebella of male mice. (B) Representative images and quantification of MBP and MOG immunostaining in the cerebella of female mice. n = 12 sections/4 mice. (C) mRNA levels of Plp1 and Mog in the cerebella of female mice. (D) Representative images of MBP and MOG immunostaining and quantification of MBP‑positive and MOG‑positive areas in the corpus callosum, motor cortex, caudoputamen, and hippocampus. n = 12 sections/4 mice. Scale bar: 200 μm. Data are means ± SEM. *p < 0.05; **p < 0.01.

Figure S5: Tango2 deletion induces reactive astrogliosis without microglial activation or gross cerebellar architectural disruption. (A) Western blot images and quantification of GLAST‐1 expression levels in WT and KO cerebellar lysates. (B) Representative images of GFAP immunostaining and quantification of GFAP‐positive areas in the GL and WM of cerebellar lobule IV/V in WT and KO mice. n = 12 sections/4 mice. Scale bar: 100 μm. (C) Representative images of GFAP immunostaining and quantification of GFAP‐positive areas in the GL and WM of cerebellar lobule IV/V in Ctrl and NG2‐Cre^ERT; Tango2 cKO mice. n = 12 sections/4 mice. Scale bar: 100 μm. (D) Representative images of IBA1 immunostaining and quantification of IBA1‐positive areas in the GL and WM of cerebellar lobule IV/V. Scale bar: 100 μm. (E) mRNA levels of inflammatory genes in WT and KO cerebella. (F) H&E staining of cerebellar sagittal sections and quantification of Purkinje cell density, ML thickness, and GL thickness of cerebella from WT and KO mice. n = 15 sections/5 mice. Scale bar: 500 μm. Data are means ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001.

Figure S6: Tango2 deletion alters biological processes and cell–cell communication associated with synaptic remodeling. Gene ontology analyses show significantly changing biological pathways in (A) cerebellar granule cells, (B) Bergmann glia, and (C) astrocytes. Cellular communication analysis revealed altered crosstalk between (D) ODC‐granule, (E) Bergmann‐granule, and (F) astrocyte‐granule.

Figure S7: VB5 treatment improves motor performance and cerebellar myelination in TANGO2‐deficient mice. (A) Percentage stacked bar chart of the relative abundance of lipid subclasses. (B) mRNA levels of PE metabolic enzymes that are not involved in de novo PE synthesis. (C) Top 15 candidate compounds, knockdown/knockout targets, and overexpression targets to rescue TANGO2‑deficient ODC alterations via pharmacogenomic screening. (D) Rotarod performance of VB5‑treated mice at speeds of 4 to 40 rpm. Male: WT PBS, n = 11; KO PBS, n = 7; KO VB5, n = 9. Female: WT PBS, n = 9; KO PBS, n = 14; KO VB5, n = 12. (E) Quantification of MBP‑positive and MOG‑positive areas in the GL and WM of cerebellar lobule IV/V. n = 24 sections/8 mice. Data are means ± SEM; *p < 0.05; **p < 0.01.

Table S1: RT‐PCR primers.

Table S2: Antibodies.

Data Availability Statement

The single‐nucleus RNA sequencing data have been deposited in the National Genomics Data Center (NGDC) under BioProject accession number PRJCA062186. The data that support the findings of this study are available on request from the corresponding author.