Shank3 related oligodendrocyte alterations in autism are restored by Erk pathway inhibition

Abstract

White matter abnormalities are consistently observed in Shank3-related autism spectrum disorders (ASD), yet the mechanisms underlying oligodendrocyte dysfunction and myelination deficits remain poorly characterized. Here, we demonstrate that Shank3 deficiency disrupts oligodendrocyte development by promoting oligodendrocyte precursor cell (OPC) proliferation while impairing functional maturation and myelination. Mechanistically, Shank3 deficiency induced hyperactivation of the Erk signalling pathway, which compromised oligodendrocyte maturation and contributes to hypomyelination. Pharmacological inhibition of the Erk pathway effectively restored oligodendrocyte maturation in vitro, rescued myelination deficits in vivo, and partially improved autism-related behaviors and motor function in Shank3-deficient mice. Transcriptomic analyses furtherly revealed dysregulation of Wnt signalling, particularly the upregulation of Wnt5a, a key ligand of the non-canonical Wnt pathway, in Shank3-deficient oligodendrocytes. Consistently, Wnt5a treatment was found to activate Erk signalling in primary oligodendrocytes and replicate the observed myelination deficits. These findings uncover the Wnt5a-Erk axis as a critical regulator of oligodendrocyte dysfunction in Shank3-related ASD and highlight its therapeutic potential for addressing associated white matter deficits.

Introduction

White matter abnormalities are a hallmark of autism spectrum disorders (ASD), with neuroimaging consistently revealing alterations in white matter integrity and connectivity across multiple brain regions [1–3]. Among the genetic contributors to ASD, mutations in the Shank3 gene stand out as one of the most robust and replicated findings [4–7]. Shank3-deficient individuals and animal models exhibit severe white matter deficits, including reduced corpus callosum volume and disrupted myelination [8, 9]. These findings suggest that white matter dysfunction may represent a critical nexus linking the diverse neuropsychiatric symptoms observed in Shank3-associated ASD [10–12].

Corpus callosum abnormalities, in particular, have been well-documented in individuals with ASD [13–15]. While a reduced corpus callosum size is consistently reported in children and adults with ASD, emerging evidence indicates that early developmental stages might feature transient increases in corpus callosum volume in individuals [16–18]. Interestingly, adult Shank3-deficient patients exhibit less pronounced white matter impairments, whereas younger individuals display more significant structural abnormalities [8]. These observations underscore the dynamic and age-dependent nature of white matter alterations in ASD and highlight the importance of a developmental trajectory as a key component of ASD pathophysiology.

At the cellular level, deficits in oligodendrocyte (OL) proliferation and reduced expression of myelin basic protein (MBP) have been implicated in the pathogenesis of ASD [19]. Oligodendrocytes undergo a highly regulated developmental process, transitioning from oligodendrocyte progenitor cells (OPCs) to mature, myelinating OLs [20, 21]. This process is critical for the precise timing of axonal myelination, a hallmark of functional white matter development [22]. Disruption at any stage of this process can have profound effects on contributing to ASD pathology [23, 24].

Several signalling pathways regulate oligodendrocyte development and function in both the developing and adult brain. The Erk(p44/42 MAPK, Erk1/2) pathway plays a pivotal role in shaping oligodendrocyte morphology and regulating myelin synthesis [25, 26]. This pathway also influences the cytoskeletal organization that is essential for oligodendrocyte maturation [27]. Additionally, non-canonical Wnt signalling via Wnt5a/JNK directs asymmetric cytoskeletal organization, modulating cytoskeletal dynamics in diverse cell types, including neural progenitors and Schwann cells [28, 29]. Cross-regulation between the RAS-Erk pathway and Wnt/β-catenin signalling has been shown to influence cell proliferation and differentiation [30–32], suggesting a potential intersection of these pathways in the regulation of oligodendrocyte maturation.

Dysregulation of the Erk pathway has been implicated in numerous syndromic forms of autism, including Fragile X syndrome, 16p11.2 deletion syndrome, tuberous sclerosis, Angelman syndrome, and Phelan–McDermid syndrome [33–35]. In human studies, transcriptomic analyses of post-mortem brain tissue from individuals with ASD have revealed significant alterations in Erk signalling pathways, highlighting the central role of Erk dysregulation in ASD pathophysiology [36]. These findings are further supported by recent mechanistic studies demonstrating that Erk signalling regulates Shank3 stability, with a kinome-wide RNAi screen identifying Erk2 as a druggable target for modulating Shank3 [37]. Furthermore, Shank3 depletion has been associated with hyperactivation of the Erk pathway and Erk-dependent cell death, particularly in KRAS-mutant cancers [38]. Together, these findings suggest that targeting the Erk pathway could offer therapeutic potential, as evidenced by studies showing Erk modulation rescued core ASD-like phenotypes in animal models [39, 40].

However, it remains unclear how Shank3 deficiency disrupts the molecular pathways regulating oligodendrocyte development, or whether targeting these pathways can rescue myelination deficits. This gap in understanding is crucial, as oligodendrocyte dysfunction may be a key link between white matter abnormalities and the diverse symptoms of Shank3-related autism spectrum disorder (ASD).

In this study, we investigated the impact of Shank3 deficiency on oligodendrocyte development and myelination, focusing on the molecular and cellular mechanisms that underlie white matter abnormalities in Shank3-related ASD. By combining in vivo mouse models, primary oligodendrocyte cultures, and transcriptomic analyses, we are presenting a comprehensive assessment of how Shank3 disruption might affect oligodendrocyte lineage progression and myelination. We further identified potential therapeutic targets that may reverse myelin deficits. This multi-pronged approach enables us: firstly, to capture the age-dependent dynamics of oligodendrocyte maturation in Shank3-deficient mice; secondly, to dissect cell-specific signalling disruptions; and thirdly, to leverage transcriptomic data to uncover novel therapeutic targets. By addressing the glial contributions to ASD pathology, this study could lay the groundwork for innovative treatment strategies that address both neuronal and glial dysfunction in Shank3-related disorders.

Materials and methods

Ethical statement

Ethical approval for the animal experiments was obtained by the review board of the Land Baden-Württemberg, Permit Number o.103-17 and 1595 TschB:W. The experiments were performed in compliance with the guidelines for the welfare of experimental animals issued by the Federal Government of Germany, the National Institutes of Health, and the Max Planck Society.

Animal housing

Pro2 KO GVO (Shank3Δ11(−/−)) mice were generated as elucidated by [41]. Pro2 KO GVO (Shank3Δ11(−/−) and wild-type mice from the same mouse line were housed under standard laboratory conditions with food and water ad libitum, along with an average temperature of 22 °C and dark/light cycle as 12/12 rhythm. Mice were genotyped and grouped as wild-type Shank3(+/+)) or knock-out (Shank3Δ11(−/−)). For biochemical analysis 7-, 21- and 140-day-old male mice were used.

Cultivation and differentiation of primary oligodendrocytes

Primary oligodendrocyte cultures were established from the cortices of postnatal day 0–2 (P0-P2) wild-type Shank3(+/+) mouse pups. Mixed glial cultures were initially prepared following the protocol described by [42]. Briefly, cortical tissue was dissociated and seeded in poly-L-lysine-coated (0.1 mg/ml, Sigma) 75 cm² flasks. Cells were maintained in OPC medium consisting of DMEM (supplemented with L-glutamine, 4.5 g/l glucose, and 1% sodium pyruvate), 2% B27, 1% FBS, 1% Pen-Strep, 10 ng/mL human FGF-basic (Peprotech, INC., 100-18B), and 30 ng/mL rhPDGF-AA (R&D systems, 221-AA). Medium was replenished every 2–3 days until confluence. At 7 days (DIV), the OPC medium was supplemented with insulin (5 μg/ml final concentration). At 9 DIV, oligodendrocyte precursor cells (OPCs) were isolated using 0.25% trypsin, centrifuged, and seeded at a density of 2.5 × 10⁴ cells/cm² onto laminin-coated (Sigma, L 2020-1MG) ibidi dishes (ibidi, 81156) for subsequent immunocytochemistry experiments. One day post-purification, the medium was completely replaced with differentiation medium composed of DMEM (supplemented as above) with 2% B27, 1% FBS, 1% Pen-Strep, and the following additives: apo-transferrin (100 μg/ml, Sigma T1147), BSA (100 μg/ml, Neofroxx 1126GR100), sodium selenite (40 μg/ml, Sigma S5261-10G), progesterone (60 ng/ml, Sigma P8783), putrescine (16 μg/ml, Sigma P7505-25G), insulin (5 μg/ml, Merck), bFGF (10 ng/ml, PeproTech Inc), T3 (30 ng/ml, Sigma T6397-100MG), and thyroxine (40 ng/ml, Merck). After 4 or 10 days in differentiation medium, the cultures were deemed ready for immunocytochemical analysis and pharmacological treatments.

Recombinant mouse Wnt3a/5a treatment in vitro

For in vitro Wnt3a (Wnt3a: 1324-WN-010/CF, R&D systems) and Wnt5a (Wnt5a: 645-WN-010/CF, R&D systems) treatment, at DIV 18 medium was replaced by a mix of fresh oligodendrocytes differentiation medium and final concentration of 100 ng/ml, 200 ng/ml, 300 ng/ml Wnt3a and wnt5a. A treatment duration of 24 h and 48 h was adapted.

Ozuriftamab (Anti-ROR2) treatment in vitro

For in vitro (Anti-ROR2: A3162, Selleckchem) treatment, at DIV 18 medium was replaced by a mix of fresh oligodendrocytes differentiation medium and final concentration of 20 µM. A treatment duration of 48 h was adapted.

Mirdametinib treatment in vivo

For in vivo Mirdametinib treatment, male Shank3Δ11(−/−) and wild-type mice were housed individually under standard laboratory conditions (22 ± 1 °C, 50% ± 10% humidity, 12-h light/dark cycle). Mice were randomly assigned to receive either Mirdametinib or vehicle treatment. Mice received Mirdametinib (30 mg/kg body weight in saline with 1% DMSO) or vehicle (saline with 1% DMSO), administered as intraperitoneal injection once daily for six days per week over 4–5 weeks. Treatment started on postnatal day (P) 27-P29, and mice were sacrificed on P60-P62 via cervical dislocation, with brains collected for biochemical analysis.

Mirdametinib treatment in vitro

For in vitro Mirdametinib treatment, at DIV 18 medium was replaced by a mix of fresh oligodendrocytes differentiation medium and final concentration of 20 μM PD0325901. A treatment duration of 48 h was adapted. The optimal concetration of PD0325901 for primary oligodendrocytes ranges between 10 μM and 30 μM [43]. For bulk RNA-seq analysis, two independent culture preparations (Exp1 and Exp2) were performed for the DMSO-KO and PD-KO groups to ensure drug-induced transcriptional effects are robust to experimental replication.

Behavioral experiments

Inverted grid test

One test to assess the animals muscle function was the grid hanging test. The mouse was placed on a horizontal grid on top of a box made of plexiglass in 15 cm above the ground. After turning the grid by 180°, the latency to fall was measured. The cutoff time was set to 300 s and the test was three times repeated, if the mouse fell within the first 10 s.

Open field and object exploration test

The paradigm for the open field test during the treatment study was modified from the previously described. The mouse was placed into the empty arena. The explorative behavior of the mouse was video recorded for 20 min. Then, the mouse was returned into its home cage for 5 min. Three Duplo bricks (LEGO®, total measures 3 × 3 × 6 cm) were placed into the center of one quadrant of the arena. The mouse was returned into the arena and video tracked for another 10 min. After the test, the arena and the bricks were cleaned with 70% ethanol and whipped dry. Analysis of the videos was performed with EthoVision XT software. For the open field test, the arena was separated into a center and a border zone. Data was collected for the total distance traveled, time spent in the center or the border zone, and the number of entries into the respective zones. For the object exploration test, the latency to the first object contact, contact events and time in contact as well as the time spent in the quadrant containing the object or in the non-object area was measured.

Marbles burying test

For the marbles burying test, a clean standard cage was filled with fresh litter (approx. 5 cm in height). Then 18 glass marbles (15 mm in diameter) were evenly distributed on the flattened litter surface. The testing mouse was carefully placed into the cage and filter lid was placed on the cage to prevent the mouse from escaping. After 30 min, the mouse was returned into its home cage, and the number of marbles which were buried to at least 2/3 depth into the litter were counted.

Immunocytochemistry

The cells were washed with DPBS, fixed with 4% PFA (Merck)/0.1 M sucrose (Roth) in 1 × DPBS (Gibco) for 20 min at RT and washed three times for 5 min with 1 × DPBS. They were permeabilized with 0.2% Triton X-100 (Roche) diluted in 1 × DPBS for 10 min at RT. After blocking with 10% goat-serum (Millipore), 5% FBS (Gibco) in 1 × DPBS at RT for 2 h, the primary antibodies were incubated at 4 °C for 24 h. After a short, 5-min, 10-min and 20-min washing steps, the secondary antibodies were incubated light protected at RT for 1 h. Washing was performed as described previously and coverslips were mounted with ProLong Gold Antifade reagent with DAPI (Invitrogen).

Immunohistochemistry

For immunostaining of the brain, mice were anesthetized by intraperitoneal injection of ketamine (WDT, 10%) and xylazine (Rompun 2%, Bayer, 20 µl/g body weight) in saline solution. Mice were trans-cardially perfused with 4% PFA (pH 7.4, Merck) in DPBS. Collected tissue was post-fixed in 4% PFA overnight and cryoprotected with 30% sucrose in 1 × DPBS at 4 °C for 18 h. Tissues were afterward lodged in O.C.T. (Tissue-Tek), cut in transversal 40 µm thick sections (brain) using a Leica CM1950 cryostat. Free-floating brain slices were suspended first in 1 × DPBS, then in cryoprotectant (50% 1 × DPBS, 30% ethylene glycol, 20% glycerol). To use the sections, the cryoprotectant was removed and sections were washed three times with DPBS. The procedure was followed either by ice-cold methanol fixation at 4 °C for 20 min (for Anti-ERK1/2 and Anti-phospho-ERK1/2) or were immediately blocked with 10% goat-serum (Millipore), 5% FBS (Gibco) and 0.1% Triton X-100 (Roche) diluted in blocking solution at RT for 4 h. Sections were then immuno-labelled with corresponding antibodies at 4 °C for 24 h. The tissue was gently washed four times with 1 × DPBS: short, 5 min, 10 min and 20 min. Secondary antibodies were obtained from Invitrogen, Jackson ImmunoResearch Laboratories or Chromotek (ms568 IgG1 and ms647 IgG2), respectively. The incubation was light protected at RT for 2 h. Following the same washing steps, tissue sections were mounted with ProLong Gold Antifade reagent with DAPI (Invitrogen). Overview images were acquired using a Leica DMi8 microscope. Detailed images were acquired with a resolution of 1024 × 1024 pixels by utilizing the Leica TCS SPE II confocal microscope (Wetzlar, Germany).

In situ hybridization

In situ hybridization was performed on free floating coronal brain sections (40 μm) using RNAscope® Fluorescent Multiplex Assay as described by vendor’s protocol (ACDbio, Cat#320850). In situ probes include: MBP (Cat#451491-C2), and Shank3 (Cat#417371). Overview images were acquired using a Leica DMi8 microscope. Detailed images were acquired with a resolution of 1024 × 1024 pixels by utilizing the Leica TCS SPE II confocal microscope (Wetzlar, Germany).

Image analysis

ImageJ 1.52p was used for z-projection, ROI extraction, manual cell counting, and measurement of fluorescence intensity and area.

Tissue analysis (Corpus Callosum)

Cell Counting: For each genotype, coronal sections at similar bregma positions were analyzed. Two images per hemisphere (left and right) were quantified for each marker. Positive cells were manually counted, normalized to DAPI+ nuclei, and averaged across N = 3 animals. MBP+ cells (P7): MBP+ cells were counted and normalized to DAPI+ nuclei. Myelinating vs. Non-myelinating Oligodendrocytes: MBP+ and NFHSMI32+ cells were classified based on morphology, counted, and expressed as a ratio to total MBP+ cells. Olig2+, CC1+, Olig2+CC1+, Ki67+, and Cleaved Caspase 3+ cells (P7, P21, P140): Positive cells were counted and normalized to DAPI+ nuclei. Fluorescence Intensity Measurements: MBP Protein and RNA Scope Signals: The soma area of MBP+ cells was manually selected, and mean intensity was measured, normalized, and averaged. ERK Intensity in O4+, GFAP+, and Iba1+ Cells: ROIs were created for each marker, and mean ERK intensity was measured, normalized, and averaged. Phospho-ERK+ Cells: Phospho-ERK+ cells co-labeled with GFAP, Iba1, or O4 were counted and normalized to DAPI+ nuclei.

Quantitative analysis to distinguish myelinating from non-myelinating oligodendrocytes: To distinguish myelinating from non-myelinating oligodendrocytes, we performed quantitative analysis using immunohistochemical markers for myelin basic protein (MBP) and SMI32. For each sample, 25 axons were randomly selected. For each axon, we measured the total axon length and the length of axonal segments overlapped by MBP immunoreactivity (“MBP overlap length”). The overlap ratio was calculated for each axon as the MBP overlap length divided by the total axon length.

To objectively determine the threshold separating myelinating from non-myelinating axons, we plotted the density distribution of overlap ratios across all axons, which revealed a bimodal pattern. The minimum point (valley) between the two modes was identified using the first local minimum of the density curve as the threshold candidate. Axons with an overlap ratio greater than this threshold were classified as myelinating; those below were classified as non-myelinating. Applying this approach, the minimum MBP overlap length among axons classified as myelinating was 120.9 µm. This value was used as the operational threshold for myelination status in subsequent analyses.

Shank3 RNA scope analysis

Shank3+ Cells: Shank3 RNA Scope signals in GFAP+, Iba1+, O4+, and NeuN+ cells were counted and normalized to DAPI+ nuclei. Shank3 Expression Level: The threshold for detecting Shank3 RNA Scope signals was determined using negative (DapB) and positive (PPIB) controls from the RNA Scope kit. The number of Shank3 RNA Scope puncta within marker-positive cells (GFAP, Iba1, O4) was quantified. All puncta were measured within DAPI+ nuclei. The number of puncta in GFAP+, Iba1+, and O4+ cells was normalized to NeuN+ cells and expressed as a percentage, representing the relative Shank3 expression level in each cell type compared to neurons.

Primary oligodendrocyte cultures analysis

Cell Counting: Ten images per coverslip (20X/40X) from two coverslips per genotype were analyzed across N = 3–5 independent cultures. Positive cells (Olig2+, Ki67+, O4+, MBP+) were manually counted, normalized to DAPI+ nuclei, and averaged. Fluorescence Intensity and Area Measurements: MBP Protein Intensity: The soma area of MBP+ cells was manually selected, and mean intensity was measured, normalized, and averaged. MBP+ Membrane Sheet Area: The membrane sheet area of MBP+ cells was manually selected, measured, normalized, and averaged.

Western blot

Brain was mechanically homogenized in modified RIPA buffer (10 mM Tris–HCl pH 7.4 (AppliChem), 0.1% sodium dodecyl sulfate (SDS, Roth), 1% Triton X-100 (Roche), 1% sodium-deoxycholate (Merck), 5 mM EDTA (Sigma), protease/phosphatase inhibitor (Roche)) utilizing an electric tissue grinder and were lysed on ice for 15 min. The lysate was sonicated 10 times for 10 s and further incubated on ice for 30 min. Lysates were clarified by centrifuging at 13,000 rpm at 4 °C for 10 min. Protein concentration was determined via Bradford assay. Therefore, 20 µl of 150 mM NaCl (Merck), 2 µl of the vortexed sample and 200 µl Bradford solution ((ethanol (95%, Roth), phosphoric acid (85%, VWR), Serva Blue (Serva) solved in distilled water) were pipetted as duplicates into a 96-well plate (Sarstedt). The plate was placed into a microplate reader and the absorbance was measured at 595 nm. The measured values were used to calculate the protein concentration of the sample and were prepared in water and sodium dodecyl sulfate ACS reagent (SDS,  ≥  99.0%, Roth) according to the desired load. Samples were boiled at 95 °C for 10 min. Gel electrophoresis was performed at 90 V for about 15 min and then at 110 V for further 45 min. The protein was blotted onto a nitrocellulose membrane by a Trans-Blot® Turbo™ Transfer System (BioRad), for the total protein, after the transfer, the total protein was imaged using the ChemiDoc MP machine, the membranes were blocked with 5% skim milk powder (Sigma) or 5% BSA in Tris-buffered saline with 0.1% TWEEN-20 (TBST). The membranes were incubated with primary antibodies at 4 °C overnight. The next day, the membranes were washed three times for 20 min with 0.1% TBST and were incubated with HRP-conjugated secondary antibodies (Dako) at RT for 1 h. Membranes were washed three times for 20 min with 0.1% TBST. Protein was visualized using the Chemiluminescent Western Blot Reagent (Thermo Fischer). Bands were analyzed using Gel-analyzer Software 2010a.

RNA isolation and bulk RNA sequencing

Qiagen RNeasy Mini kit was used to isolate total RNA according to the manufacturer’s description including all purification steps. RNA was eluted in 50 μl RNAse-free water. Quality control (QC), library construction, sequencing, and read mapping was performed by Novogene Munich. In brief, messenger RNA was purified from total RNA and library cDNA synthesis was performed. Libraries were quality checked with Qubit and pooled for sequencing on Illumina platform. Raw data (raw reads) of fastq format were cleaned by removing reads containing adapter, ploy-N and low-quality reads. Reference genome index was built and paired-end clean reads were aligned to the reference genome using Hisat2 v2.0.5. Read counts were calculated with featureCounts v1.5.0-p3.

DEG analysis

All computational analyses were conducted using R version 4.4.1 and RStudio 2024.04.2 + 764. The differential gene expression (DGE) analysis primarily followed the methodology described by utilizing functions from the limma and edgeR packages [44–46]. Gene Selection and Normalization: Only genes with assigned entrezIDs (derived from Ensemble IDs using the biomaRt R package) and RNA biotypes categorized as protein_coding, lincRNA, Mt_rRNA, snoRNA, miRNA, snRNA, scaRNA, rRNA, ribozyme, or macro_lncRNA were included in the analysis. Read counts were normalized using the trimmed mean of M-values (TMM) method implemented in edgeR::calcNormFactors().Filtering and Batch Effect Correction: Genes were filtered using edgeR::filterByExpr(), retaining those with a minimum of 10 counts in 25% of the samples. Batch effects were mitigated either through limma::removeBatchEffect() for graphical representation in Linear Discriminant Analysis (LDA) plots or by inclusion as a factor in the linear model for differential expression analysis. Model Design and Contrast Generation:A voom object was generated from the filtered genes, and a design matrix was constructed based on the formula: ∼0+group+batch where the group factor encoded WT_DMSO_D14, KO_DMSO_D14, WT_DMSO_D20, KO_DMSO_D20 and the batch factor encoded the four experimental batches. Contrasts were derived from this design using limma::makeContrasts().Statistical Analysis: Linear models were fitted to each gene using limma::lmFit(), and contrasts were computed from these models using limma::contrast.fit(). Empirical Bayes statistics were calculated using limma::eBayes() to moderate the standard errors of the estimated log-fold changes. Identification of Differentially Expressed Genes: Differentially expressed genes were identified using thresholds of p < 0.05 and |log2 fold change| >0.

Gene ontology

Gene ontology (GO) analysis on differentially expressed genes (DEGs) was conducted using the ClusterProfiler R package. The analysis incorporated all previously included genes as the background set. P-values were adjusted using the Benjamini-Hochberg (BH) method, and q-values were computed to control for false discovery rate. Enrichment Criteria: Terms were considered significantly enriched when both p-values and q-values were below the threshold of 0.01. AveExpr, mean normalized expression (TMM-adjusted log2(CPM)) was calculated across biological replicates within each condition. Semantic Similarity and Term Reduction: The rrvgo package was employed to identify and reduce redundant GO terms. Semantic similarity between terms was calculated using the “Wang” method. Terms were weighted by -log10(q-value) during the reduction process of all Biological Process (BP) terms. Similarity Metrics and Community Detection: Jaccard similarities between terms were computed using the stats::dist() function. To identify coherent groups of related terms, community detection was performed using the Louvain algorithm implemented in the tidygraph::group_louvain() function.

EGSEA

RNA-seq raw counts were normalized using the TMM method (Trimmed Mean of M-values) to account for library size differences. Experimental design matrix was constructed to model group differences (e.g., genotype and treatment effects). Contrasts were defined for: Genotype contrast: (DMSO_KO_D20) .vs. (DMSO_WT_D20); Treatment contrast: (PD_KO_D20) .vs. (DMSO_KO_D20) Differential expression analysis was performed using the limma-voom pipeline. Multiple GSEA methods (ORA, GAGE, camera, GSVA, PADOG, SAFE, PLAGE, zscore, ssGSEA, globaltest, fry) were integrated using the EGSEA package to robustly identify enriched gene sets for each contrast. Significant gene sets (FDR-adjusted p-value < 0.05) were identified for both genotype and treatment contrasts. Overlapping gene sets between genotype and treatment contrasts were determined to identify pathways affected by both genotype and treatment. Rescue effects were defined as gene sets downregulated in the genotype contrast (logfc <0) but upregulated in the treatment contrast (logfc >0), and upregulated in the genotype contrast (logfc >0) but downregulated in the treatment contrast (logfc <0). Jaccard similarity was computed between significant gene sets based on gene overlap to cluster pathways with similar gene content. Clustering and visualization of gene set relationships were performed using the simplifyEnrichment package, aiding functional interpretation. Genes from selected clusters (e.g., upregulated Wnt signaling genes, downregulated myelin/skeletal genes) were extracted for further analysis.

Primary antibodies list

Antibody Species	Company	Country State	Reference	Method	Dilution	Blocking
SHANK3 (rb)	Homemade	Germany	Homemade	ICC	1:500	10% DS + 5% FBS in PBS-/-
MBP (ms)	Atlas Antibodies	Sweden, Stockholm	AMAb91062	ICC	1:200	10% DS + 5% FBS in PBS-/-
OLIG2 (rb)	Millipore	United States, Massachusetts	AB9610	ICC	1:500	10% DS + 5% FBS in PBS-/-
GFAP (gp)	Synaptic Systems	Germany	173004	ICC	1:500	10% DS + 5% FBS in PBS-/-
ERK1/2 (rb)	Cell Signaling	United States, Massachusetts	#9102	ICC	1:1000	10% DS + 5% FBS in PBS-/-
Phospho ERK1/2 (rb)	Cell Signaling	United States, Massachusetts	#4370	ICC	1:1000	10% DS + 5% FBS in PBS-/-
Ki67 (ms)	Abcam	United States, Massachusetts	ab238020	ICC	1:500	10% DS + 5% FBS in PBS-/-
PLP (ms)	Abcam	United States, Massachusetts	Ab9311	ICC	1:500	10% DS + 5% FBS in PBS-/-
CNPase (ms)	Abcam	United States, Massachusetts	Ab6319	ICC	1:200	10% DS + 5% FBS in PBS-/-
MBP (rb)	Millipore	United States, Massachusetts	AB5864	IHC	1:200	10% DS + 5% FBS + 0.3% Triton in PBS-/-
NFH (ck)	Antibodies-online	Germany	ABIN361351	IHC	1:1000	10% DS + 5% FBS + 0.3% Triton in PBS-/-
GFAP (gp)	Synaptic Systems	Germany	173004	IHC	1:500	10% DS + 5% FBS + 0.3% Triton in PBS-/-
Iba1 (ms)	Synaptic Systems	Germany	234011	IHC	1:500	10% DS + 5% FBS + 0.3% Triton in PBS-/-
Ki67 (rb)	Abcam	United Kingdom	Ab15580	IHC	1:500	10% DS + 5% FBS + 0.3% Triton in PBS-/-
CC1 (ms)	Merck	Germany	MABC200	IHC	1:500	10% DS + 5% FBS + 0.3% Triton in PBS-/-
Sox9 (rb)	Merck	Germany	AB5535	IHC	1:500	10% DS + 5% FBS + 0.3% Triton in PBS-/-
Cleaved caspase3 (rb)	Cell signaling	United States, Massachusetts	#9661	IHC	1:500	10% DS + 5% FBS + 0.3% Triton in PBS-/-
O4 (ms)	R&D Systems	United States, Minneapolis	MAB1326	IHC	1:500	10% DS + 5% FBS + 0.3% Triton in PBS-/-
ERK1/2 (rb)	Cell Signaling	United States, Massachusetts	#9102	WB	1:1000	5% BSA in 0.1% TBST
Phospho ERK1/2 (rb)	Cell Signaling	United States, Massachusetts	#4370	WB	1:1000	5% BSA in 0.1% TBST
Wnt5a/b (rb)	Cell Signaling	United States, Massachusetts	#2530	WB	1:1000	5% BSA in 0.1% TBST
Phospho-LRP6(rb)	Cell Signaling	United States, Massachusetts	#2568	WB	1:1000	5% BSA in 0.1% TBST
Dvl2 (rb)	Cell Signaling	United States, Massachusetts	#3224	WB	1:1000	5% BSA in 0.1% TBST
LRP6 (rb)	Cell Signaling	United States, Massachusetts	#3395	WB	1:1000	5% BSA in 0.1% TBST
Dvl3 (rb)	Cell Signaling	United States, Massachusetts	#3218	WB	1:1000	5% BSA in 0.1% TBST
Naked1 (rb)	Cell Signaling	United States, Massachusetts	#2262	WB	1:1000	5% BSA in 0.1% TBST
Naked2 (rb)	Cell Signaling	United States, Massachusetts	#2073	WB	1:1000	5% BSA in 0.1% TBST
ß-Catenin (rb)	Abcam	United Kingdom	Ab6302	WB	1: 2000	5% milk in 0.1% TBST
Frizzled-3 (rat)	R&D systems	United States, Minneapolis	MAB1001	WB	1: 1000	5% BSA in 0.1% TBST
Frizzled-6 (goat)	R&D systems	United States, Minneapolis	AF1526	WB	1: 1000	5% BSA in 0.1% TBST
JNK (ms)	R&D systems	United States, Minneapolis	MAB1846	WB	1: 1000	5% BSA in 0.1% TBST
NLRC5 (rb)	Abcam	United Kingdom	Ab105411	WB	1: 1000	5% BSA in 0.1% TBST
TLR2 (rb)	Abcam	United Kingdom	EPR20303	WB	1: 1000	5% BSA in 0.1% TBST
Phospho-JNK1/JNK2(rb)	ThermoFisher	United States, Massachusetts	44682 G	WB	1: 1000	5% BSA in 0.1% TBST
β-ACTIN (ms)	Sigma-Aldrich	United States, Massachusetts	A5411	WB	1:250,000	5% BSA in 0.1% TBST

Secondary antibodies list

Antibody	Company	Country/State	Reference	Dilution	Blocking
AlexaFlour® 488 donkey anti-rabbit	Jackson Immuno research	United Kingdom	711-545-152	1:500	10% DS + 5% FBS 0.2% Triton in PBS-/-
AlexaFlour® 594 donkey anti-rabbit	Jackson Immuno research	United Kingdom	711-585-152	1:500	10% DS + 5% FBS 0.2% Triton in PBS-/-
AlexaFlour® 647 donkey anti-rabbit	Jackson Immuno research	United Kingdom	711-605-152	1:500	10% DS + 5% FBS 0.2% Triton in PBS-/-
AlexaFlour® 488 donkey anti-mouse	Jackson Immuno research	United Kingdom	715-545-151	1:500	10% DS + 5% FBS 0.2% Triton in PBS-/-
AlexaFlour® 594 donkey anti-mouse	Jackson Immuno research	United Kingdom	715-585-151	1:500	10% DS + 5% FBS 0.2% Triton in PBS-/-
AlexaFlour® 647 donkey anti-mouse	Jackson Immuno research	United Kingdom	715-605-151	1:500	10% DS + 5% FBS 0.2% Triton in PBS-/-
AlexaFlour® 488 donkey anti-chicken	Jackson Immuno research	United Kingdom	703-545-155	1:500	10% DS + 5% FBS 0.2% Triton in PBS-/-
AlexaFlour® 594 donkey anti-chicken	Jackson Immuno research	United Kingdom	703-585-155	1:500	10% DS + 5% FBS 0.2% Triton in PBS-/-
AlexaFlour® 647 donkey anti-chicken	Jackson Immuno research	United Kingdom	703-605-155	1:500	10% DS + 5% FBS 0.2% Triton in PBS-/-
AlexaFlour® 488 donkey anti-gp	Jackson Immuno research	United Kingdom	705-545-147	1:500	10% DS + 5% FBS 0.2% Triton in PBS-/-
AlexaFlour® 594 donkey anti-gp	Jackson Immuno research	United Kingdom	705-585-147	1:500	10% DS + 5% FBS 0.2% Triton in PBS-/-

Plots and statistical analysis

All figures for RNA sequencing were generated in R with ggplot2, pheatmap, ggVennDiagram, tidygraph and ggraph packages. Statistical analyses of data and graphs were performed using GraphPad Prism 9 Software and/or Microsoft Excel. Data display the mean ± Standard Error of Mean (SEM). Data were tested for normality using Shapiro-Wilk test. For parametric data, Student’s t test and one-way analysis of variance (ANOVA) followed by post hoc analysis (Tukey’s multiple comparisons test) were used. Nonparametric data were analyzed using Mann-Whitney test or Kruskal-Wallis test followed by post hoc analysis (Dunn’s multiple comparisons test). Significance levels (p values) were set to 0.05 (p  ≤  0.05*, p  ≤  0.01**, p  ≤  0.001***, p  ≤  0.001****) with 95% confidence interval.

Results

Age dependent dysregulation of oligodendrocyte maturation and myelination in the Corpus Callosum of Shank3-Deficient mice

To investigate the oligodendrocytes in the corpus callosum (CC) of Shank3 knockout (KO) mice, we employed RNA scope and anti-MBP immunostaining. We first analysed the total number of MBP positive mature oligodendrocytes (MBP + ) at postnatal day 7 (P7). Our analysis revealed an increased total number of MBP+ cells and enhanced MBP intensity in Shank3 KO mice compared to wild-type (WT) controls (Fig. 1A). MBP mRNA levels showed increased levels, but this difference did not reach statistical significance (Fig. 1A). To further assess the myelination in the corpus callosum, we performed double immunolabeling for non-phosphorylated neurofilament H (SMI32), and MBP, a marker for myelinated axons [47]. Myelinating and non-myelinating oligodendrocytes were distinguished by calculating the MBP overlap ratio for 25 randomly selected axons per genotype. The classification threshold was defined by the first local minimum of the bimodal overlap ratio distribution, corresponding to a minimum MBP overlap length of 120.9 µm (Suppl. Fig. 1A, see Methods for details). Surprisingly, Shank3 KO mice showed a clear decrease in the proportion of myelinating oligodendrocytes (37.16% vs 65.53%, p < 0.001) and accordingly an increase in non-myelinating oligodendrocytes (62.84% vs 34.47%, p < 0.001) (Fig. 1B). This indicates that despite the increased total number of MBP+ cells, a significant portion remains non-myelinating in KO mice at P7.

Fig. 1. Age-dependent dysfunction of oligodendrocytes in the Corpus Callosum of Shank3 KO mice.

Coronal brain sections from Shank3 WT and KO mice of corpus callosum (CC) have been used for all analyses. A Brain sections from P7 WT and Shank3KO mice stained for MBP and MBP RNA scope. MBP protein and RNA intensity were normalized to mean value. MBP positive cells were shown in the percentage of DAPI. Mean ± SD. Student’s Unpaired t test, p** < 0.01, n = 3 animals. Scale bar = 200 μm. B Immunostaining for MBP and NFH SMI32 and analysis of myelinating and non-myelinating oligodendrocytes in P7 CC. Myelinating and non-myelinating cells were shown in the percentage of total MBP positive cells. Mean ± SD. Student’s Unpaired t test, p**** < 0.0001, n = 3 animals. Scale bar = 100 μm. C P7, P21 and P140 brain section were stained for Olig2 and CC1. Cell number of Olig2 + CC1+ cells were analysed in CC. Data were shown in the percentage of DAPI and Olig2 respectively. Mean ± SD. Student’s Unpaired t test, p* < 0.05, n = 3 animals. Scale bar = 20 μm. D P7 brain section were stained for Ki67, Iba1 and GFAP and proliferation of Ki67+cells was analysed in medial and lateral CC separately. Ki67 positive cells were shown in the percentage of DAPI. Mean ± SD. Student’s Unpaired t test, p* < 0.05, n = 3 animals. Scale bar = 20 μm.

To examine the developmental progression of oligodendrocyte lineage cells, we analysed the proportion of cells co-expressing Olig2+ (a marker for oligodendrocyte progenitor cells) and CC1+ (a marker for mature oligodendrocytes) at three time points: postnatal days 7 (P7), 21 (P21), and 140 (P140). At P7, we found a significant upregulation in the proportion of CC1+Olig2+ cells relative to the total Olig2+ population in Shank3 KO mice compared to WT controls. In WT controls, 3.54% Olig2+ cells expressed CC1, while in Shank3 KO mice, this proportion was 10.63% (Fig. 1C). At P21, we observed a different pattern. The proportion of CC1+Olig2+ cells in WT controls was 52.75% compared to 36.87% of CC1+Olig2+ cells in Shank3 KO mice (Fig. 1C). The distinction between genotypes persisted into adulthood. At P140, 71.66% of Olig2+ cells were CC1+ in WT controls, while this proportion was 16.35% in Shank3 KO mice (Fig. 1C). These findings suggest a complex pattern of oligodendrocyte maturation characterized by early acceleration followed by a decline at later stages. Our RNAscope in situ hybridization analyses, combined with immunostaining for Olig2 and CC1, demonstrate that Shank3 expression in the corpus callosum is initiated at postnatal day 7 (P7), peaks during the critical period of myelination at P21, and subsequently declines in the mature stage at P140. Notably, Shank3 expression is highest in oligodendrocyte precursor cells (Olig2⁺) and progressively decreases as these cells mature into CC1⁺ oligodendrocytes (Suppl. Fig. 1B).

To investigate the source of the increased mature oligodendrocytes at P7, we examined Ki67 expression, a marker for cell proliferation. We found an increased number of Ki67+ cells in the corpus callosum of Shank3 KO mice at P7, which did not colocalize with astrocytes (GFAP+) or microglia (Iba1+) (Fig. 1D). This suggests enhanced proliferation of Shank3 deficient oligodendrocyte lineage cells. Additionally, there was no significant difference in the expression of cleaved caspase 3, a marker for apoptosis, neither in the corpus callosum at P7 (Suppl. Fig. 1C) nor of P140 mice (Suppl. Fig. 1D), indicating that the observed changes are not due to altered cell death. These results unveil a complex pattern of oligodendrocyte development in Shank3 KO mice, characterized by an early increase in the number of myelinating oligodendrocytes at P7, followed by a decrease in the proportion of mature oligodendrocytes at later stages.

Shank3 deficiency alters the development and maturation of primary oligodendrocytes in culture

To furtherly elucidate the developmental consequences of Shank3 deficiency underlying the observed in vivo phenotypes, we established a primary oligodendrocytes cell culture systems derived from P0-P2 cortices of Shank3 WT and KO mice (Fig. 2A). We used the expression of Olig2, O4, and MBP as markers to define the developmental stages of oligodendrocytes (Fig. 2B).

Fig. 2. Impaired proliferation, differentiation, and maturation of oligodendrocyte lineage cells in primary oligodendrocytes cultures of Shank3 KO.

A Experimental timeline of primary cell culture of mixed glia cells from P0-P2 cortices of WT and Shank3 KO mice. B Olig2 was used as a marker of oligodendrocyte progenitor cells, O4 as marker of premature oligodendrocytes and MBP as marker of mature oligodendrocytes in mouse primary cell culture. C Cells were stained for Olig2 and Ki67, and proliferation (Ki67+) of oligodendrocytes lineage cells (Olig2+) was analyzed. The percentage of Olig2 to DAPI and the percentage of Olig2 + Ki67+ to DAPI were shown after normalized to mean value. Mean ± SD. Student’s paired t test, p* < 0.05, n = 3–4 independent cultures. Scale bar = 50 μm. D Cells were stained for O4 and the percentage of premature oligodendrocytes (O4+ cells) to DAPI were shown after normalized to mean value. Mean ± SD. Student’s paired t test, p* < 0.05, n = 3 independent cultures. Scale bar = 20 μm. E Cells were stained for MBP, and the number of MBP+ cells, MBP intensity and MBP membrane sheet area in WT and KO cells were analyzed. Data were shown after normalized to mean value. Mean ± SD. Student’s paired t test, p**  <  0.01, p*** < 0.001, p**** < 0.0001, n = 5 independent cultures. Scale bar = 20 μm.

The ratio of Ki67-positive Olig2-positive cells was significantly higher in KO cultures compared to WT, indicating enhanced proliferation of oligodendrocyte progenitor cells (OPCs) (Fig. 2C). This finding suggests that Shank3 deficiency may lead to an expansion of the OPC population. Examination of premature oligodendrocytes using O4 as a marker revealed an increased number of O4-positive cells in KO cultures. The representative morphology of these O4-positive cells showed distinct differences between WT and KO conditions, with KO cells exhibiting less continuous O4 puncta morphology (Fig. 2D). This observation indicates that there is an increase in the number of premature oligodendrocytes. Focusing on mature oligodendrocytes, while we observed an increase in the number of MBP-positive cells in KO cultures, we saw a significant reduction in somatic MBP expression compared to WT cells. Concomitantly, the cell surface area of MBP+ cells was decreased (Fig. 2E). This suggests that although more cells are expressing MBP, the overall production of myelin and the extent of oligodendrocyte maturation are compromised in Shank3-deficient cells. To further investigate the maturation process, we examined MBP-positive cells at two time points: DIV14 and DIV20. We identified two distinct morphological types of MBP-positive cells and quantified their percentages at both time points. The results showed a shift in the distribution of these morphological types between DIV14 and DIV20, indicating ongoing maturation processes. Notably, at both DIV14 and DIV20, we consistently observed fewer cells with membrane sheets in the Shank3-KO cultures compared to WT cultures (Suppl. Fig. 2A. a). The time-course analysis at DIV14, DIV16, DIV18, and DIV20 revealed that progressively lower MBP intensity in KO cultures compared to WT, further supporting impaired maturation (Suppl. Fig. 2A. b). To further characterize our model, we examined Shank3 expression patterns within this culture system. Immunocytochemical analysis revealed that Shank3 protein was expressed in oligodendrocytes. Interestingly, Shank3 did not directly colocalize with oligodendrocyte markers such as MBP, PLP, or CNPase (Suppl. Fig. 2B), suggesting a potential regulatory role for Shank3 in oligodendrocyte development that may not involve direct interaction with these myelin-associated proteins.

Transcriptomic analysis revealed altered myelin-related genes and wnt signalling in Shank3-deficient primary oligodendrocytes

To further investigate the molecular mechanisms underlying the observed phenotypes in Shank3-deficient oligodendrocytes, we performed transcriptomic analysis of primary oligodendrocyte cultures derived from Shank3 WT and KO mice. The cells were cultured for 14 and 20 days in vitro, resulting in four experimental groups: WTD14, KOD14, WTD20, and KOD20 (each n  =  4) (Fig. 3A). For a comprehensive analysis of differentially expressed genes (DEGs), we utilized a pipeline that included the limma package for DEG analysis, clusterProfiler for enrichment analysis of gene ontology (GO) terms, and rrvgo to identify semantic similarities between GO terms (Fig. 3A). We employed linear discriminant analysis (LDA) to further reduce dimensionality and clarify group distinctions between phenotypes, as LDA is effective for enhancing the separation of predefined groups and improving interpretability (Fig. 3B).

Fig. 3. Transcriptomic analysis reveals downregulated myelin sheath and upregulated WNT signalling genes in Shank3 KO oligodendrocytes.

A Experimental design of groups and sample size. R package pipeline. B Linear discrimination analysis (LDA) plot of 4 groups. LD1 and LD2 represent the primary and secondary principal component. C Top 10 enriched GO terms from up- and downregulated DEGs by -log10 (qvalue) of DIV20. Size of points corresponds to the ratio of DEGs to all genes of the GO term. BP: Biological process, MF: molecular function, CC: cellular component. D Extracted genes from the downregulated CC term myelin sheath and filtered genes that are at least shared by 3 clusters of Fig. 3E. AveExpr as the average log2 counts per million (logCPM) expression value across samples. E Hierarchical clustering of up- regulated BP terms based on the jaccard similarity.

Focusing on the DIV 20 samples, which comprise a more mature culture system characterized by an increased presence of mature oligodendrocytes with well-developed membrane sheets (Suppl. Fig. 3A, statistical analysis was shown in Suppl. Fig. 2A), we identified a total of 694 downregulated genes and 510 upregulated genes through differential expression analysis. Gene Ontology (GO) enrichment analysis revealed 445 downregulated and 408 upregulated terms, with a significance threshold set at q.value < 0.01. The top 10 terms from both the upregulated and downregulated categories are presented in Fig. 3C. In our analysis of downregulated terms, the “myelin sheath” category emerged from the cellular component (CC) analysis. This category encompasses nine genes predominantly associated with membrane proteins expressed in mature oligodendrocytes, including gap junction proteins such as Gjc2 and Cldn14, myelin proteolipid proteins Pllp and Mal, as well as cytoskeletal proteins Mobp and Ermn, and the cell adhesion protein Tmem125. Additionally, Ppp1r14a was identified as being highly upregulated during myelination (Fig. 3D). These findings align with the reduced MBP intensity and smaller membrane sheet area observed in Shank3-deficient oligodendrocytes.

The upregulated biological process (BP) terms were predominantly associated with developmental processes, encompassing key activities such as proliferation, differentiation, and morphogenesis. Clustering based on semantic similarity further categorized these terms into five primary groups (Fig. 3E). Notably, the largest cluster was identified as Wnt signalling, a pathway critically involved in oligodendrocyte development and myelination [47]. By filtering genes shared by at least three clusters, we identified key regulators of Wnt signalling, including Fzd3, Fzd6, Wnt5a, Sfrp1, and Vangl2 (Fig. 3D). Among these, Wnt5a activates ERK via the Ror2 receptor, a key component of non-canonical Wnt signalling [48]. Additionally, Sox9 is a transcription factor known to regulate oligodendrocyte differentiation and myelination [49], and has been shown to directly bind to the Wnt5a promoter. We confirmed elevated Sox9 expression in primary oligodendrocyte cultures (MBP+ oligodendrocytes) (Suppl. Fig. 3B).

These transcriptomic findings provide a molecular context for the phenotypic changes observed in Shank3-deficient oligodendrocytes. The downregulation of myelin sheath genes, particularly those related to cytoskeletal proteins, aligns with the reduced MBP intensity and smaller membrane sheet area we previously observed. Additionally, the upregulation of genes associated with Wnt signalling suggests a potential mechanism for the altered proliferation and maturation of oligodendrocytes in the absence of Shank3.

Shank3 deficiency alters Wnt5a signalling and Erk activation in primary oligodendrocytes

Transcriptomic analysis identified the upregulation of key genes involved in Wnt signalling, which was of particular interest given the stage-specific effects of Wnt signalling on oligodendrocyte development reported in previous studies [47]. Wnt signalling pathway can be divided in two types, canonical and non-canonical. Among Wnt ligands, Wnt3a induces phosphorylation of the co-receptor LRP6, which is a key step in activating canonical Wnt signalling, and results in accumulation of β-catenin in the cytoplasm. In the non-canonical pathway, WNT5a activates JNK/phospho-JNK and Erk/phospho-Erk [50–52] (Fig. 4A).

Fig. 4. Wnt5a-induced ERK signalling impairs oligodendrocytes differentiation in Shank3 KO cultures.

A Illustration of Wnt3a and Wnt5a associated signalling pathways. B Western blot of Wnt5a expression from the lysate of WT and Shank3 KO derived primary oligodendrocytes. Mean ± SD. Student’s paired t test, p**  <  0.01, n = 6 independent cultures. C Primary oligodendrocytes of WT were treated with vehicle (PBS) and Wnt5a 300 ng/ml for 48 h and stained for MBP. The membrane sheet of MBP positive cells were shown after normalized to the mean value. Violin plot, Mann-Whitney test, p**** < 0.0001, n = 20 MBP+ cells from 3 independent cultures. Scale bar = 20 μm. D Primary oligodendrocytes of WT were treated with vehicle (PBS) as Control (CTRL), Wnt3a and Wnt5a (100 ng/ml, 200 ng/ml, 300 ng/ml, 48 h), respectively. Western blots were immunolabeled for LRP6, phospho-LRP6, β-catenin, JNK, phospho-JNK, Erk and phospho-Erk and normalized to total protein (Shown in Suppl Fig. 4B). Western blot bands of control, Wnt3a (300 ng/ml), Wnt5a (300 ng/ml) were analysed. Western blot bands of phospho-JNK and phospho-Erk under all the conditions were analysed. Mean ± SD, Dunnett´s multiple comparsion test, p* < 0.05, p** < 0.01, p*** < 0.001, n = 3 independent WT cultures. E Primary oligodendrocytes of WT and Shank3 KO were lysed and Western blots of LRP6, phospho-LRP6, β-catenin, JNK, phospho-JNK, Erk and phospho-Erk were analyzed. Mean ± SD. Student’s paired t test, p* < 0.05, n = 3 independent cultures. F Primary oligodendrocytes of WT and Shank3 KO were immunostained for Erk and MBP, and Erk expression within MBP+ cells was analyzed, and normalized to mean value. Mean ± SD. Student’s paired t test, p** < 0.01, n = 3 independent cultures. Scale bar = 20 μm.

To validate the identified key genes involved in the WNT signalling pathway, we performed western blots with the Wnt Signalling Antibody sampler Kit (Cell signalling #2915), which detects integral proteins within the Wnt signalling pathway. We found an upregulation of Wnt5a in Shank3 KO primary oligodendrocytes compared to WT controls, corroborating our transcriptomic findings (Fig. 4B). Notably, β-catenin levels were also significantly elevated in KO cultures, consistent with our previous observations [53]. However, other Wnt pathway proteins showed no significant changes (Suppl. Fig. 4A). To elucidate the functional consequences of elevated Wnt5a, we treated WT oligodendrocytes with recombinant Wnt5a (300 ng/ml for 48 h). This treatment resulted in inhibition of MBP+ cell differentiation and reduced surface area (Fig. 4C), recapitulating the phenotype observed in Shank3 KO cells. These results suggest that increased Wnt5a signalling may contribute to the altered morphology and impaired maturation of Shank3-deficient oligodendrocytes.

To delineate the downstream signalling effects of upregulated Wnt5a, we conducted a comparative analysis of Wnt3a and Wnt5a treatments on WT cells. Dose-response experiments (100 ng/ml, 200 ng/ml, and 300 ng/ml) at 24 and 48 h revealed that both Wnt3a and Wnt5a induced β-catenin accumulation at both time points. However, significant activation of phospho-Erk was observed only at 48 h, prompting us to focus on this time point for subsequent analyses (Suppl. Fig. 4B). As anticipated, Wnt3a treatment, serving as a positive control for canonical Wnt signalling, resulted in increased phosphorylation of LRP6 and accumulation of β-catenin. Intriguingly, Wnt5a treatment activated both JNK and Erk signalling, as evidenced by elevated levels of phospho-JNK and phospho-Erk (Fig. 4D).

Comparative analysis of WT and Shank3 KO oligodendrocytes revealed increased Erk and phospho-Erk levels in KO cells, while JNK and phospho-JNK levels remained unchanged (Fig. 4E). This suggests that in the context of Shank3 deficiency, Erk signalling may be the primary downstream effector of increased Wnt5a. Immunocytochemical analysis confirmed the elevated Erk expression in Shank3 KO MBP+ cells (Fig. 4F).

To determine whether inhibition of Wnt5a signalling could normalize aberrant ERK activation in Shank3-deficient oligodendrocytes, we targeted ROR2, a classical receptor mediating noncanonical WNT5A signalling [54]. RNA scope analysis confirmed that Ror2 is robustly expressed in MBP+ oligodendrocytes (Suppl. Fig. 4C). Treatment of Shank3 KO primary oligodendrocyte cultures with a selective ROR2 inhibitor (20 μM, 48 h) resulted in a marked reduction of ERK phosphorylation, as assessed by western blot (Suppl. Fig. 4D). These results indicate that the elevated ERK activation observed in Shank3-deficient oligodendrocytes is mediated, at least in part, through ROR2-dependent Wnt5a signalling.

These findings provide a molecular link between Shank3 deficiency, increased Wnt5a signalling, and altered Erk activation in oligodendrocytes. The dysregulation of this pathway may contribute to the impaired oligodendrocyte maturation and myelination observed in Shank3-deficient conditions.

Temporal and cell-specific alterations in Erk signalling in Shank3-deficient corpus callosum

To further investigate the Erk signalling pathway in Shank3 KO mice in vivo, we performed Western blot analyses across three developmental time points (P7, P21, and P140) in three brain regions: corpus callosum, cortex, and striatum. Our results revealed that the Erk signalling pathway was affected in the Shank3 KO model, with a temporal pattern that closely mirrored the previously observed MBP changes (Suppl. Fig. 5A). Specifically, in the corpus callosum, we observed a trend towards increased phospho-Erk (pErk) and pErk/Erk ratio at P7 (Fig. 5A). This was followed by a significant downregulation of total Erk, pErk, and pErk/Erk ratio at P21(Fig. 5B), which persisted into adulthood at P140 (Fig. 5C).

Fig. 5. Upregulation of Erk and Phospho-Erk in the Corpus Callosum of P7 Shank3 KO Mice.

A–C Western blot of Erk and phospho-Erk in corpus callosum at postnatal day 7 (A), 21 (B), and 140 (C), relative to actin, and then normalized to the mean value. Mean ± SD. Student’s unpaired t test, p* < 0.05, p** < 0.01, p*** < 0.001, n = 6 animals. D Brain sections of P7 WT and Shank3 KO were stained for Erk, O4 and GFAP. Corpus callosum was analyzed for total Erk intensity, Erk intensity in immature oligodendrocytes (O4+) and astrocytes (GFAP+). Mean ± SD. Student’s unpaired t test, p*< 0.05, p** < 0.01, p****  < 0.0001, n = 3 animals. Scale bar = 50 μm. E Brain sections of P7 WT and Shank3 KO were stained for phospho-Erk, O4 and GFAP. Corpus callosum was analyzed for total phospho-Erk positive cell number, the number of phospho-Erk positive immature oligodendrocytes (O4+) and astrocytes (GFAP+) were analyzed. Mean ± SD. Student’s unpaired t test, p** < 0.01, n = 3 animals. Scale bar = 50 μm.

The corpus callosum is primarily composed of oligodendrocytes, astrocytes and microglia. To elucidate the cell type-specific expression of Shank3 in the corpus callosum, we employed RNA scope in conjunction with cell type-specific markers. We utilized GFAP as a marker for astrocytes, Iba1 for microglia, and O4 for immature oligodendrocytes. Our RNA scope analysis revealed the presence of Shank3 mRNA in all three glial cell types examined (Suppl. Fig. 5B), indicating a broader expression pattern of Shank3 across various glial populations in the corpus callosum.

To further characterize the cell type-specific contributions to the observed Erk signalling alterations, we performed immunostaining analyses on P7 corpus callosum sections. Our findings revealed a significant increase in total Erk immunoreactivity in the corpus callosum of Shank3 KO mice compared to WT controls. Notably, this elevation was predominantly observed in immature oligodendrocytes (O4+) and astrocytes (GFAP+) (Fig. 5D), but not in microglial cells (Suppl. Fig. 5C). Consistent with our Western blot analysis, immunostaining revealed a marked increase in the number of phospho-Erk positive cells in the corpus callosum of Shank3 KO mice at P7 (Fig. 5E). Intriguingly, we observed a dichotomous regulation of phospho-Erk across different glial populations. While there was a reduction in phospho-Erk-positive O4+ cells, we noted a significant increase in phospho-Erk-positive GFAP+ astrocytes (Fig. 5E). Microglial cells showed no detectable phospho-Erk signal in either genotype (Suppl. Fig. 5D).

In parallel, we examined Wnt5a protein levels in vivo across the same developmental time points. Western blot analysis revealed that Wnt5a expression in the corpus callosum of Shank3 KO mice was elevated at P7, but significantly decreased at P21 and P140, mirroring the temporal profile observed for ERK and pERK (Suppl. Fig. 5E). This coordinated regulation suggests that Wnt5a may act as an upstream modulator of aberrant ERK activity in the Shank3-deficient corpus callosum. To further illustrate the relationship among these key molecular markers, we summarized the temporal expression patterns of Wnt5a, ERK, pERK, and MBP in a comparative table (Suppl. Fig. 5F). Notably, all four proteins displayed similar trends across the three developmental stages, supporting the existence of a Wnt5a–ERK–MBP (myelination) axis in vivo. These results provide valuable insights into the molecular mechanisms underlying the myelin abnormalities observed in Shank3-deficient models. The temporal and cell type-specific dysregulation of Erk signalling may contribute to the altered oligodendrocyte development and myelination deficits associated with Shank3 deficiency.

ERK pathway inhibition partially rescues behavioral and oligodendrocyte deficits in Shank3-deficient models

To investigate whether ERK pathway inhibition could ameliorate behavioral deficits associated with Shank3 deficiency, we chronically treated WT and Shank3 KO mice with the selective MEK inhibitor Mirdametinib for a period of 4 to 5 weeks. Behavioral performance was assessed using a battery of tests including marble burying, open field, object exploration, and the inverted grid test (Fig. 6A). In the marble burying test, vehicle-treated Shank3-KO mice buried fewer marbles than WT controls, consistent with reduced avoidance of novel objects. Mirdametinib treatment did not significantly alter marble burying in KO mice, but slightly reduced marble burying in WT animals. In the open field test, Mirdametinib increased total locomotion in both WT and KO mice, with KO mice exhibiting a more pronounced enhancement, indicating improved exploratory behavior. Anxiety-like behavior, measured by entries into the center of the arena, was reduced in veh-KO mice compared to WT, whereas Mirdametinib treatment partially normalized center entries in KO mice. Furthermore, Mir-KO mice spent more time investigating a novel object in the object exploration test, suggesting enhanced novelty-driven exploration. In the inverted grid test, KO mice displayed consistently shorter latencies to fall relative to WT, indicative of impaired muscle function. Notably, Mirdametinib treatment increased latency to fall in KO mice from day 7 onwards, with effects maintained through day 28, demonstrating partial rescue of motor deficits. Collectively, these data indicate that ERK pathway inhibition selectively improves autism-related behaviors and motor function in Shank3-deficient mice. To investigate whether these behavioral improvements are associated with enhanced myelination, we examined MBP expression in the corpus callosum of WT, Shank3 KO, and Mirdametinib-treated Shank3 KO mice (treatment starting at P28, analyzed at P60–62). Western blot analysis revealed that ERK inhibition significantly restored MBP levels in KO mice, suggesting that the observed behavioral improvements are likely linked to enhanced oligodendrocyte maturation and myelin integrity in vivo (Fig. 6B)

Fig. 6. ERK Pathway Inhibition Partially Rescues Behavioral and Oligodendrocyte Deficits in Shank3-Deficient Models.

A Behavior tests: The marble burying test, conducted in the fifth week of treatment, assessed avoidance behavior toward novel objects. The number of buried marbles is shown for wild-type and Shank3-KO mice treated with Mirdametinib or vehicle. General locomotion was measured as the total distance traveled during a 20-min trial in the open field arena. Anxiety-like behavior was assessed by quantifying the number of entries into the center of the open field arena. In the object exploration test, the total time spent sniffing a novel object is shown. Data were analyzed using ordinary one-way ANOVA followed by Tukey’s multiple comparisons test. All data are presented as mean ± SEM. Significance thresholds were set as follows: p ≤ 0.1 (#), p ≤ 0.05 (*), p ≤ 0.01 (**), p ≤ 0.001 (***). Sample size: n = 8–10 animals per group. Muscle function and endurance were assessed by measuring grid-hanging time three times per week from the start of treatment. The latency to fall from the inverted grid is shown. Data were analyzed using two-way ANOVA with repeated measures (factor: time). Statistical comparisons are shown only for veh-WT vs. veh-KO, and veh-KO vs. Mir-KO. All data are presented as mean ± SEM. Significance thresholds were set as follows: p ≤ 0.1 (#), p ≤ 0.05 (*), p ≤ 0.01 (**), p ≤ 0.001 (***), p ≤ 0.0001 (****). Sample size: n = 8–10 animals per group. Abbreviations: SEM: standard error of the mean; veh-WT, vehicle treated wild-type mice; veh-KO, vehicle treated Shank3-knockout mice; Mir-WT, Mirdametinib treated wild-type mice; Mir-KO, Mirdametinib treated Shank3-knockout mice; ns, not significant. B Western blot analysis of MBP expression in corpus callosum in WT and Shank3 KO mice, KO_DMSO and KO_PD treated conditions. Mean ± SD. Student’s unpaired t test, *p < 0.05, n = 5–6 animals. C Primary oligodendrocytes of WT and Shank3 KO were treated with 20 μM PD0325901 or vehicle (DMSO) for 48 h. MBP intensity and surface area were analyzed. Data normalized to DMSO treatment. Mean ± SD. Student’s paired t test, *p < 0.05, n = 3 independent cultures. Scale bar = 20 μm. D Proliferating (Ki67 + ) cells were analysed in DMSO and PD0325901 treated cells, relative to DMSO. Mean ± SD. Student’s paired t test, *p < 0.05, n = 3 independent cultures. Scale bar = 20 μm. E Experimental design of groups and sample size. R package pipeline. F Venn diagram depicting the number of GSs enriched in both contrasts and uniquely in genotype or treatment contrast and overlapping GSs by the contrast based on the direction of regulation (up = 1, down = 0). G Clusters 1–4 of hierarchical clustering of shared GS based on inter-GS JC including only genes with logFC > 0 for genotype and <0 for treatment contrast.

Building on these in vivo findings, we next investigated whether ERK inhibition could directly rescue cellular deficits in primary Shank3-deficient oligodendrocytes, reasoning that improved oligodendrocyte function may underlie the observed behavioral and myelination effects. We treated the cells with 10 μM and 20 μM PD0325901 for 48 h. Western blot analysis revealed significant reduction in pErk expression at both concentrations, leading us to select 20 μM for 48 h as our treatment protocol for subsequent experiments (Suppl. Fig. 6A). Primary oligodendrocytes derived from Shank3 KO mice were treated with PD0325901, and myelin basic protein (MBP) expression was assessed. PD0325901 treatment significantly restored MBP expression in Shank3-deficient oligodendrocytes to levels comparable with WT cells (Fig. 6C). This rescue effect was further confirmed by western blot analysis (Suppl. Fig. 6B). This finding indicates that Erk pathway inhibition can ameliorate the myelin deficits observed in Shank3 KO oligodendrocytes. We previously observed increased proliferation in primary oligodendrocytes derived from Shank3 KO mice. To determine whether Erk pathway inhibition could mitigate this proliferative effect, we treated Shank3-deficient oligodendrocytes with PD0325901. Immunocytochemical analysis with Ki67 revealed that PD0325901 treatment successfully rescued the increased proliferation phenotype in Shank3 KO oligodendrocytes, restoring proliferation rates to levels similar to WT cells (Fig. 6D). Together, these findings demonstrate that ERK inhibition corrects both myelination and proliferation deficits in Shank3-deficient models.

To elucidate the mechanisms underlying ERK inhibitor-mediated restoration, we performed transcriptomic analysis on ERK inhibitor-treated samples. Cells were cultured in vitro for 20 days, resulting in three experimental groups: WT_DMSO_D20 (n = 4), KO_DMSO_D20, and KO_PD_D20 (each n = 8). Two main contrasts were analyzed: genotype (KO_DMSO_D20 vs. WT_DMSO_D20) and treatment (KO_PD_D20 vs. KO_DMSO_D20) (Fig. 6E). Principal component analysis (PCA) of global gene expression revealed that WT_DMSO, KO_DMSO, and KO_PD samples formed three distinct clusters (Suppl. Fig. 6C). This indicates that PD treatment induces broad transcriptomic shifts rather than simply normalizing KO expression toward WT. We therefore turned to pathway-level analyses to better capture potential rescue effects relevant to oligodendrocyte biology.

Gene set enrichment analysis was conducted using the Ensemble of Gene Set Enrichment Analyses (EGSEA) framework, which integrates multiple gene set enrichment algorithms to robustly identify biologically relevant pathways and processes perturbed in the dataset. Complete ranked gene lists for both contrasts were analyzed, incorporating gene sets from Biological Process (BP), Molecular Function (MF), and Cellular Component (CC) Gene Ontology (GO) categories.

In total, with an adjusted P value < 0.05, 3,249 gene sets (GSs) were enriched in the treatment contrast, with 1,084 overlapping with the 1,933 enriched GSs identified in the genotype contrast (Fig. 5F). Of these overlapping GSs, 55 were downregulated in the genotype contrast and upregulated in the treatment contrast, while 268 showed the opposite pattern—upregulated in the genotype contrast and downregulated following treatment—indicative of a rescue-like effect of the ERK inhibitor on these gene sets (Fig. 5F).

To identify meaningful clusters among all GSs, we calculated the proportion of shared genes between gene sets, focusing on genes inversely expressed across contrasts (logFC >0 in the genotype contrast and logFC <0 in the treatment contrast). Hierarchical clustering revealed top clusters related to Wnt signalling, extracellular matrix (ECM), catenin, and proliferation (Fig. 6G). A parallel analysis of gene sets with logFC <0 in the genotype contrast and logFC >0 in the treatment contrast identified clusters associated with microtubule organization, immune response, ion channel activity, calcium signalling, and potassium ion transport (Suppl. Fig. 6D).

We also assessed the average expression n (logCPM) of upregulated Wnt signalling genes and downregulated myelin-associated genes (as identified in Fig. 3D), finding that myelin-related genes such as Mal and Mobp were significantly rescued following treatment (Suppl. Fig. 6E). Furthermore, changes in oligodendrocyte lineage progression after treatment were evaluated using specific cell stage markers as defined by [55]. Although WT and KO groups showed only subtle differences in OPC and COP marker expression in our bulk RNA-seq data, PD treatment demonstrated a shift from OPCs to more mature oligodendrocyte states (Suppl. Fig. 6F). This pattern suggests a pro-differentiation effect, consistent with our cellular observations that Shank3 deficiency expands the OPC pool and impairs terminal maturation.

Collectively, our findings underscore the central role of ERK signalling not only in mediating the oligodendrocyte proliferation and maturation observed in Shank3-deficient models, but also improved autism-related behaviors and motor function in vivo. These results establish a mechanistic link between oligodendrocyte dysfunction and behavioral abnormalities in Shank3-deficient models, and underscore the therapeutic potential of ERK pathway inhibition for ameliorating both cellular and behavioral deficits in Shank3-related disorders.

Discussion

This study provides new mechanistic insight into the age-dependent dysregulation of oligodendrocyte (OL) maturation and myelination in Shank3-deficient mice. By integrating in vivo and in vitro analyses, we identify the Wnt5a–ERK signalling axis as a central mediator of aberrant OL lineage progression and impaired myelin formation. Critically, pharmacological inhibition of ERK signalling rescues myelination deficits, suggesting a potential therapeutic strategy for white matter abnormalities in SHANK3-related neurodevelopmental disorders.

Mechanistic insights: the Wnt5a–ERK axis

We observed that in Shank3-deficient mice, the expression patterns of Wnt5a, phosphorylated ERK (pERK), and myelin basic protein (MBP) follow a similar trajectory in vivo: all are elevated at postnatal day 7 (P7), but reduced at later stages (P21 and P140) compared to wild-type controls. In contrast, our in vitro primary oligodendrocyte cultures from Shank3 knockout (KO) mice consistently show upregulation of Wnt5a and ERK/pERK, yet a marked reduction in MBP expression. While this initially appears inconsistent, a closer examination of cellular phenotypes resolves the discrepancy.

At P7 in vivo, despite an increase in MBP-positive cells, there is a lower proportion of mature, myelinating oligodendrocytes and a significant rise in proliferating (Ki67 + ) cells. This indicates that early ERK hyperactivation primarily drives OPC proliferation but simultaneously blocks their differentiation into mature, myelinating cells. This impaired differentiation persists or becomes more pronounced at later stages (P21, P140), as evidenced by a significant reduction in Olig2 + CC1+ mature oligodendrocyte numbers in the KO model. These findings are mirrored in vitro, where ERK hyperactivation also leads to increased proliferation but reduced differentiation. Thus, the observed reduction in ERK/pERK at P21 and P140 in vivo likely reflects a maladaptive response to prior ERK hyperactivation at P7, resulting in persistent deficits in oligodendrocyte maturation and myelination. This dynamic is consistent across both in vivo and in vitro systems.

Importantly, our results from primary oligodendrocyte cultures—where non-oligodendroglial influences are excluded—demonstrate that impaired differentiation in the absence of Shank3 is a cell-autonomous effect. This interpretation is strongly supported by recent studies using human iPSC-derived OPCs from individuals carrying SHANK3 mutations, which revealed intrinsic deficits in calcium signalling, glutamate responsiveness, and myelin-related gene expression, independent of the surrounding neural environment [54]. Furthermore, in vivo and organoid studies confirm that SHANK3 is expressed in oligodendrocyte lineage cells and that its deficiency leads to maturation-dependent myelin defects, further substantiating a cell-autonomous mechanism [9]. While a conditional, OPC-specific Shank3 knockout model would provide definitive proof, the convergence of our in vitro findings and recent human and mouse data provides compelling evidence that Shank3 deficiency intrinsically disrupts oligodendrocyte development and myelination.

Notably, both systems reveal that Shank3 expression closely parallels that of MBP, peaking at the critical period of myelination (P21 in vivo; DIV14 in vitro) and declining at mature stages (P140 in vivo; DIV20 in vitro). In vitro, the expression pattern of Shank3 at DIV14 and DIV20 is supported by transcriptomic data (Suppl. Fig. 3C). This temporal alignment suggests that Shank3 plays an important role in oligodendrocyte maturation and myelin formation. Our RNAscope in situ hybridization analyses, combined with immunostaining for Olig2 and CC1, provide robust evidence that Shank3 expression is highest in oligodendrocyte precursor cells (Olig2+) and diminishes as cells mature into CC1+ oligodendrocytes. This pattern is consistently observed across developmental time points in the corpus callosum and recapitulated in primary oligodendrocyte cultures (Suppl. Fig. 2C). The higher expression of Shank3 in Olig2+ cells suggests a stage-dependent regulatory function, potentially influencing the transition from precursor proliferation to differentiation and myelination. This interpretation is reinforced by recent findings, providing evidence that Shank3 deficiency leads to increased OPC proliferation and impaired maturation [54], directly aligning with our observations. Although white matter pathology has been recognized as a predominant feature of SHANK3 deficiency, our recent studies indicate that defects in oligodendrocyte maturation and myelination are not confined to white matter. We observed a significant decrease in MBP expression in the cortex (gray matter) at both P21 and P140 [9], and other work using the InsG3680 Shank3 mutant mouse model has demonstrated reduced expression of myelin-related genes and proteins in both frontal cortex (gray matter) and striatum [54], alongside deficits in oligodendrocyte differentiation and morphology. These findings highlight that Shank3’s regulatory role in oligodendrocyte lineage cells extends to both white and gray matter regions, broadening the scope of its impact on neural circuit integrity and brain function.

To further dissect the molecular mechanisms underlying these phenotypes, we considered both direct and indirect pathways by which Shank3 deficiency might activate ERK signalling. Recent work has shown that SHANK3 functions as a scaffold protein that directly interacts with MAPK pathway molecules [38, 56, 57]. Thus, it is possible that at P7, ERK activation in Shank3-deficient oligodendrocytes may also be influenced by this KRAS-mediated mechanism, when Shank3 expression is low and the system is especially vulnerable to dysregulation.

However, our data in the oligodendrocyte context support a multi-layered regulatory cascade involving β-catenin and Sox9 as the predominant mechanism. Previous studies have demonstrated that Shank3 can bind directly to β-catenin [58], and our findings showed that upregulation of β-catenin in Shank3 deficient oligodendrocytes, while gene sets associated with β-catenin binding were downregulated following treatment. Sox9 has been proposed to act through direct binding to β-catenin or the β-catenin destruction complex [59]. Most crucially, Sox9 is known to regulate oligodendrocyte differentiation and myelination [49], and has been shown to directly bind to the Wnt5a promoter, enhancing its expression [60]. Our immunostaining confirmed elevated Sox9 in both MBP+ oligodendrocytes in vitro and CC1+ mature oligodendrocytes in vivo at P7 (Suppl. Fig. 5G). This suggests that increased Sox9 expression in Shank3-deficient oligodendrocytes drives Wnt5a overproduction, contributing to the observed upregulation of the Wnt5a–ERK signalling axis. Additionally, we found that Ror2, a specific Wnt5a receptor [48] highly expressed in mature oligodendrocytes, mediates ERK activation, as pharmacological inhibition of Ror2 normalized ERK activity in primary cultures.

Together, our findings delineate a mechanistic cascade in which Shank3 deficiency disrupts β-catenin and Sox9 regulation, leading to Wnt5a upregulation and subsequent hyperactivation of the ERK pathway. This, in turn, results in increased OPC proliferation but impaired differentiation and myelination, with lasting consequences for white matter integrity.

ERK inhibition as a therapeutic strategy

Our findings position ERK inhibition as a promising therapeutic strategy for addressing myelination deficits in Shank3-related autism spectrum disorder (ASD), with robust evidence spanning molecular, cellular, behavioural, and translational domains.

ERK inhibition exerts stage-specific effects on oligodendrocyte lineage progression [43, 61]. In Shank3-deficient models, hyperactivated ERK signalling drives oligodendrocyte precursor cell (OPC) proliferation while blocking differentiation—a phenotype reversed by ERK inhibitors. Interestingly, recent studies also report that Shank3 overexpression alters myelin-related gene expression [62] and perturbs ERK signalling [63]. These findings, together with our loss-of-function data, suggest that oligodendrocyte development and myelination are highly sensitive to Shank3 dosage. While Shank3 deficiency leads to hyperactive ERK signalling and impaired OPC differentiation, overexpression likewise disrupts myelin gene regulation, highlighting the critical need for balanced Shank3 levels to maintain oligodendrocyte homeostasis. Integrative RNA-seq analysis of ERK inhibitor-treated Shank3 KO cells revealed transcriptome-wide rescue effects, including: Shift from OPCs to mature oligodendrocytes, as evidenced by lineage marker analysis [55]. The rescue of ion transport and calcium signalling pathways following ERK inhibition is critical for oligodendrocyte function. Calcium signalling regulates key processes such as myelin sheath formation and axonal support [64–66], with disruptions in Shank3-deficient models impairing oligodendrocyte differentiation and myelination [54]. For instance, calcium influx through voltage-gated channels is essential for initiating myelination [67, 68], and its downregulation (Cav3.1, Cav2.2) in Shank3 KO cells aligns with deficits in myelin gene expression (Suppl. Fig. 3D). Similarly, potassium channels maintain ionic homeostasis required for oligodendrocyte maturation [69]. Therefore, restoration of related GSs following ERK inhibition likely facilitates functional recovery, as supported by the observed restoration of MBP expression. Aberrant Wnt5a overexpression drives ERK hyperactivation, which disrupts the balance between OPC proliferation and differentiation in Shank3 deficiency, ERK inhibition normalizes Wnt pathway components might restore transcriptional programs necessary for myelination. Concurrently, extracellular matrix (ECM) remodeling—particularly involving hyaluronan and laminin—is critical for providing structural and biochemical cues that guide OPC migration and myelin stability [70, 71]. Dysregulated ECM gene sets in Shank3 KO models has been reported previously [53], reversed ECM GSs after ERK inhibition, might re-establish a permissive microenvironment for myelination. While our in vivo treatment was not oligodendrocyte-specific, the concordance between in vitro and in vivo rescue of MBP expression supports a cell-autonomous mechanism. Future studies using oligodendrocyte-targeted ERK inhibition could further clarify cell-type-specific contributions.

Multiple ASD mouse models with white matter abnormalities demonstrate behavioral improvements following ERK pathway inhibition. In BTBR mice—a model of idiopathic autism with documented white matter deficits—treatment with the MEK inhibitor PD0325901 (Mirdametinib) during juvenile stages reduced ERK activation, ameliorated core ASD-like behaviors (sociability, communication, repetitive behaviors), and normalized aberrant EEG signals [39]. Similarly, acute MEK inhibition in adult Spred1-/- mice reversed enhanced social dominance and improved nesting behaviour [72]. These behavioral rescues align with our observations of restored oligodendrocyte maturation and myelination, and selectively rescued autism-related behaviors and motor function following ERK inhibition in shank3 deficient mice model, suggesting that white matter repair may contribute to functional recovery.

The selection of P28 for in vivo intervention addresses a key translational challenge: initiating treatment after pathological establishment. Although ERK hyperactivation peaks at P7, impaired oligodendrocyte differentiation becomes pronounced by P21, suggesting that maladaptive pathways sustain deficits even after initial ERK dysregulation subsides. Late intervention at P28 may “reset” these pathways, mirroring clinical scenarios where treatment often begins post-diagnosis. This approach is bolstered by studies showing that restoring Shank3 or myelination at later stages improves function [54, 73–75]. Furthermore, ERK inhibition modulates Shank3 levels in vivo, suggesting a feedback loop that could amplify therapeutic benefits [37].

Notably, human studies of SHANK3 deficiency reveal pronounced white matter alterations in long association tracts (e.g., uncinate fasciculus) with minimal gray matter changes, underscoring white matter pathology as a primary driver of neurodevelopmental deficits [8]. The observed reduction in fractional anisotropy in Shank3 KO mice reflects compromised axonal structure and myelin integrity, which can slow or desynchronize action potential conduction and disrupt neural circuit communication. As efficient long-range connectivity is crucial for normal brain development and function, the predominance of white matter disease in SHANK3 deficiency likely underlies the cognitive, social, and motor impairments observed in individuals with Phelan-McDermid syndrome (PMDS).

In this context, PD0325901—a blood-brain barrier–penetrant MEK inhibitor currently in phase II trials for neurofibromatosis type 1 (NF1) [76], is particularly promising. Its established safety profile and CNS efficacy provide a clear rationale for considering its repurposing as a therapeutic strategy in SHANK3-related ASD.

Limitations and future directions

While our study provides mechanistic clarity regarding the role of the Wnt5a–ERK axis in oligodendrocyte lineage progression and myelination deficits in Shank3-deficient models, several limitations should be acknowledged. First, although ERK inhibition partially improved motor function and exploratory behaviors, we did not comprehensively assess whether restoring myelination fully translates to broader ASD-relevant behaviors, such as social interaction or cognitive performance. Second, our findings are based on murine models; validation in human systems—such as iPSC-derived oligodendrocytes or post-mortem ASD brain tissue—will be essential to confirm translational relevance. Third, while we observed normalization of gene sets involved in ion transport, calcium signalling, and extracellular matrix pathways, the precise functional consequences of these transcriptomic changes for oligodendrocyte physiology and myelin integrity require further investigation. Finally, our focus on Shank3-related ASD raises the question of whether the Wnt5a–ERK axis is similarly dysregulated in other ASD subtypes. Testing ERK inhibition in models with distinct genetic etiologies (e.g., CHD8, CNTNAP2) will help clarify the broader applicability of this therapeutic approach.

Broader implications

Our findings reposition Shank3 as a regulator of both neuronal and glial dysfunction in ASD, extending its impact beyond synaptic roles to include oligodendrocyte maturation and myelination. By demonstrating that ERK pathway modulation can restore key molecular and cellular features of white matter integrity, this work underscores the importance of glial-specific mechanisms in ASD pathophysiology. These insights highlight the potential of targeting ERK signalling as a disease-modifying strategy for white matter pathology in neurodevelopmental disorders. Furthermore, our results support the need for future clinical trials to consider glial endpoints—such as myelination and oligodendrocyte function—when evaluating the efficacy of ERK-targeted therapies.

Conclusion

This study establishes the Wnt5a–ERK axis as a critical mechanism underlying oligodendrocyte dysfunction and myelination deficits in Shank3-related ASD. By demonstrating that ERK inhibition can rescue structural abnormalities at the molecular and cellular levels, we provide a roadmap for addressing white matter pathology in ASD and related neurodevelopmental disorders. Future work integrating glial biology and behavioral outcomes will be essential to translate these mechanistic insights into clinical breakthroughs.

Author contributions

YM, A-KL, and TB conceptualized and designed the study. MS and HFB provided the animal material and supervised the breeding process. HFB performed the in vivo Mirdametinib treatment, conducted the behavioral experiments including data collection, and provided the corresponding animal material. YM conducted the experiments, analyzed the data, and performed the statistical analyses. YM drafted the manuscript, which was critically revised by A-KL. A-KL and TB supervised the project and provided overall guidance.

Funding

Open Access funding enabled and organized by Projekt DEAL.

Data availability

Raw data sets will be made available upon reasonable request to the corresponding author.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41380-025-03333-1.