USP11 drives stress-induced synaptic structural deficits and depression-like behaviors through GSK3β/mTOR signaling

Abstract

Major depressive disorder (MDD) is a highly disabling psychiatric condition characterized by profound synaptic dysfunction—particularly in the prefrontal cortex—which constitutes a core neuropathological hallmark of the disease. Nevertheless, the molecular mechanisms linking ubiquitination dynamics to stress-associated synaptic structural deficits remain poorly defined. In this study, we identified USP11, an X-linked deubiquitinating enzyme, as a key upstream regulator of the GSK3β/mTOR signaling cascade and the resultant synaptic structural impairments induced by chronic stress. In mice exposed to chronic unpredictable mild stress (CUMS), USP11 expression in the prefrontal cortex was markedly elevated, accompanied by aberrant phosphorylation of GSK3β and mTOR. Immunoprecipitation–mass spectrometry (IP–MS) and co-immunoprecipitation analyses verified a specific interaction between USP11 and GSK3β. Subsequent biochemical and cellular assays demonstrated that USP11 directly deubiquitinates GSK3β, modulates its Ser9 phosphorylation level, and consequently alters its enzymatic activity. Functional investigations using transgenic mice and primary neuronal cultures revealed that USP11 acts as a negative regulator of synaptic integrity. Strikingly, USP11 deficiency conferred robust protection against stress-induced synaptic injury and behavioral impairments by restoring synaptic protein expression, preserving neuronal ultrastructure, and ameliorating depressive-like behaviors. Collectively, these findings delineate a critical USP11–GSK3β/mTOR–synaptic plasticity axis underlying depression-related neuropathology and underscore USP11 as a promising therapeutic target for synaptic preservation and neuroprotection in MDD.

1. Introduction

Major depression disorder (MDD) is a highly disabling psychiatric illness which constitutes a formidable global health burden (Yang et al., 2024; Lépine and Briley, 2011). The pathophysiology of depression is exceptionally complex. Although classical frameworks such as the monoamine hypothesis have guided drug discovery for decades, they fail to fully capture the disorder's complexity, and a substantial proportion of patients remain refractory to conventional treatments (Cowen and Browning, 2015). The emergence of rapid-acting antidepressants such as ketamine has shifted attention toward synaptic-plasticity mechanisms (Kim et al., 2024; Tartt et al., 2022; Liao et al., 2025), implicating synaptic dysfunction as one of central pathogenic substrates underlying depression.

Emerging evidence indicates that post-translational modifications (PTMs), particularly ubiquitination and deubiquitination, crosstalk to collectively regulate protein stability (Lee et al., 2023), subcellular localization (Zhu et al., 2022), catalytic activity and synaptic function (Feng et al., 2019; Sager et al., 2024; Cajigas et al., 2010; Mabb and Ehlers, 2010; Patrick et al., 2023). However, the specfic governing molecular logic remains elusive. Deubiquitinating enzymes, especially the ubiquitin-specific protease (USP) family which is the largest deubiquitinase family (Yan et al., 2022), regulate protein function through a variety of molecular mechanisms (Gao et al., 2023; Chen et al., 2021). Beyond preventing proteasomal degradation, these enzymes can fine-tune substrate activity via non-proteolytic deubiquitination (Anckar and Bonni, 2015). Given the recognized role of deubiquitinating enzymes in synaptic regulation (Yan et al., 2022), we hypothesized that the X-linked deubiquitinase USP11 may be dysregulated in the prefrontal cortex under chronic stress and contribute to synaptic pathology in depression. However, its role in depression remains largely unclear.

Synaptic plasticity encompasses both functional changes in synaptic efficacy and structural remodeling of synaptic architecture (Gipson and Olive, 2017; Nugent et al., 2023); among the regulators of synaptic function and remodeling, glycogen synthase kinase-3β (GSK3β) has emerged as a key node that integrates multiple signaling cascades with activity-dependent neuronal remodeling (Liao et al., 2025; Duda et al., 2020; Cuesto et al., 2015; Llorens-Martín et al., 2013). Our mass spectrometry analysis identified GSK3β as a binding partner of USP11. Accumulating evidence indicates that the therapeutic efficacy of diverse antidepressants—including ketamine—converges on the inhibition of GSK3β activity (Pham and Gardier, 2019). GSK3β is a constitutively active serine/threonine kinase whose activity is principally restrained by inhibitory phosphorylation at Ser9 mediated by upstream kinases such as Akt (Atkins et al., 2012). Yet this phosphorylation-centric view is incomplete and insufficient to explain the dynamic fluctuations in GSK3β activity (Nagini et al., 2019).

Given the high heritability of depression and the consistently reported dysregulation of GSK3β activity in patients and animal models (Karege et al., 2007; Tao et al., 2023; Liu et al., 2018a), we hypothesize that USP11 regulates GSK3β activity through deubiquitination. Specifically, USP11-mediated deubiquitination may modulate Ser9 phosphorylation, thereby enhancing kinase activity. We propose that USP11 potentiates GSK3β function via ubiquitination-phosphorylation crosstalk, ultimately driving synaptic structural damage and depression-like behaviors. This study aims to elucidate the role of the USP11-GSK3β axis in chronic stress-induced synaptic structural deficits, providing a mechanistic framework for understanding stress-related synaptic dysfunction and identifying novel therapeutic targets.

2. Materials and methods

2.1. Cells

Human embryonic kidney 293T (HEK293T) cells were obtained from Procell Life Science & Technology Co., Ltd. (Wuhan, China). Cells were maintained in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 10 % fetal bovine serum (FBS) and 1 % penicillin-streptomycin (P/S) at 37 °C in a humidified atmosphere with 5 % CO₂.

Primary cortical neurons were isolated from embryonic day 18 (E18) mouse pups, as previously described (Wang et al., 2018). Briefly, cortical tissues were dissociated in Neurobasal medium containing B27 supplement, GlutaMAX, and 1 % P/S. Neurons were plated onto poly-D-lysine-coated plates and maintained at 37 °C, 5 % CO₂ in a humidified incubator. Half of the culture medium was replaced every 3–4 days.

2.2. Transfection

2.2.1. Plasmid and siRNA transfection

Cells were seeded into appropriate culture dishes and allowed to reach 70–80% confluency prior to transfection. Plasmids encoding Flag-tagged USP11 (Wide-type and C318S mutation) and His-tagged GSK3β were purified by standard endotoxin-free purification protocols. For transient transfection. Transfections were performed using Neofect™ DNA transfection reagent (Cat. No. TF201201) per manufacturer instructions. For the small interfering RNA (siRNA) treatment experiments, USP11-targeting or scramble siRNAs (Sangon Biotech) were transfected at 20 nM using the CALNP™ RNAi in vitro transfection kit (D-Nano, Cat. No. DN001-10). Cells were collected 48–72 h post-transfection for downstream analysis.

2.2.2. Adeno-associated virus (AAV) infection of primary neurons

Primary cortical neurons were prepared from embryonic day 18 (E18) mouse pups and plated at the indicated density. At DIV4–DIV5, cultures were infected with AAV particles encoding USP11-overexpression construct. The AAV viral suspension was gently added to the culture medium and incubated for at least 48–72 h. Transduction efficiency was verified by immunoblotting analysis of USP11 protein expression.

2.2.3. USP11-overexpressing HEK293T cell lines

HEK293T cells were seeded in 6-well plates and cultured until reaching 50–70 % confluence. Cells were then infected with lentiviral particles encoding USP11. After 24 h, the infection medium was replaced with fresh complete culture medium. At 48 h post-infection, cells were subjected to selection with 7 μg/mL blasticidin S (specified by vector resistance marker) for 7–14 days, with medium replaced every 2–3 days, the selected cells were maintained in growth medium containing blasticidin for routine passage. Successful establishment of stable cell lines was confirmed by immunoblotting analysis of USP11 protein expression.

2.3. Dot blot

Purified USP11 recombinant protein was diluted in phosphate-buffered saline (PBS) and spotted onto nitrocellulose membranes. The membranes were irradiated with a 245 nm UV lamp for 1 h to crosslink and immobilize proteins. After air-drying, membranes were blocked in 5 % non-fat milk for 1 h at room temperature and washed three times with TBST. For protein-binding evaluation, membranes were incubated overnight at 4 °C with GSK3β protein solution, and incubated with primary antibody against GSK3β (1:500, CST) overnight at 4 °C. After three TBST washes, membranes were incubated with HRP-conjugated secondary antibody (1:5000, SAB) for 1 h at room temperature. Signal detection was performed using enhanced chemiluminescence (ECL).

2.4. Mice

All animal experiments were conducted in strict accordance with the Guidelines for the Care and maintained under specific pathogen-free (SPF) conditions. The study was approved by the Institutional Animal Care and Use Committee (IACUC) of Renmin Hospital of Wuhan University (Approval No: WDRM20230603C). We have exerted maximum efforts to minimize the suffering of animals and reduce the number of animals used. Additionally, we endeavor to employ in vivo cell whenever feasible to conduct certain experiments. All efforts were made to minimize the number of animals used and their suffering.

Wild-type (WT) C57BL/6J male mice were purchased from Hunan SJA Laboratory Animal Co., Ltd. The USP11 knockout (USP11-KO) mice in C57BL/6N background were generated by Cyagen Company (Suzhou, China). Cas9 protein, two gRNAs (gRNA-1:TTTAGTTGTGCAGGATGGCGGGG,RNA-2:GGCTACCCATTAAAGCTACATGG) flanked Exon 2∼9 of mouse USP11 were co-injected into fertilized eggs. The embryos were transferred to recipient female mice to obtain F0 mice. The genotype of knockout mice was confirmed by PCR using two pairs of primers (F1: 5′-AACACATTAGATGGCTGACAAACAC-3′,R1:5′-TTCCTGAGCCACTTCCTGTTGAC-3’; F2:5′-TCTTATCTCATGCTCACTCTCCC-3′,R1:5′-TTCCTGAGCCACTTCCTGTTGAC-3′) and sequencing.

Male mice aged 8-10 weeks with weighing 22-25 g were used for the experiments and allowed to acclimate in the animal facility for at least one week before experiments. The mice were housed in a specific pathogen-free (SPF) facility under standard laboratory conditions, with a controlled temperature of 22 ± 2 °C, relative humidity of 55 ± 5 %, and a 12-h light/dark cycle. All mice had ad libitum access to standard chow and water, except where noted as part of the CUMS procedure. Control animals were maintained under these standard conditions throughout the experimental period.

For the experiments, WT and USP11-KO mice were randomly assigned to experimental groups. At the end of the experimental period, mice were deeply anesthetized with isoflurane. Upon confirmation of a surgical level of anesthesia, tissues were rapidly harvested and stored at −80 °C for subsequent analysis.

2.5. CUMS

The CUMS procedure was adapted from previously described protocols (Wang et al., 2024). Mice were first allowed to acclimate for one week to the standard housing conditions. Subsequently, mice in the CUMS group were subjected to a variety of mild, unpredictable stressors for a period of 4 consecutive weeks. The stressors were categorized as either long-term or short-term.

Long-term stressors included: 1) 24-h food deprivation; 2) 24-h water deprivation; 3) 24-h cage tilt (45°); 4) 24-h empty cage (removal of all bedding and enrichment); 5) 24-h odor exposure (e.g., soiled bedding from a different cage). Short-term stressors included: 1) 6-h restraint in a 50 ml centrifuge tube; 2) 5 min of tail pinch (1 cm from the tip); 3) 5 min of cold-water swim (4 °C); 4) 5 min of cage oscillation. The daily stress protocol involved the random application of one long-term and one short-term stressor, with food deprivation and water deprivation never scheduled on consecutive days. To prevent habituation, the specific types of stressors were not repeated within any 3-day period. The control group mice were housed in a separate room and were not exposed to any stressors, although they were handled daily to simulate the conditions of the CUMS group. The body weight and general health of all mice were monitored weekly throughout the CUMS procedure.

2.6. Schedule of behavioral testing

All behavioral experiments were performed during the light phase between 09:00 and 17:00. Behavioral tests were conducted within one week, with at least 24 h between tests. To minimize potential carryover effects, assays with higher acute stress load (FST/TST) were scheduled after less stressful assessments (SPT/OFT).

2.7. Sucrose preference test (SPT)

SPT was used to assess anhedonia (Liu et al., 2018b). As described previously with minor modifications (Yin et al., 2023), mice were housed in single cages and given two bottles (1 portion with 1 % sucrose solution, 1 portion with water) for 48 h before the test (switching the position every 12 h). After 24 h of water deprivation (standard chow available ad libitum), each mouse was given two pre-weighed bottles (1 portion with 1 % sucrose solution, 1 portion with water) for 24 h. Standard chow remained available ad libitum during the 24-h two-bottle measurement period, and bottle positions were switched at 12 h to avoid side bias. The calculation formula of sucrose preference rate is: sucrose preference rate (%) = sucrose intake/(sucrose intake + water intake) × 100 %.

2.8. Open field test (OFT)

The OFT was used to assess locomotor activity and anxiety-like behaviors in mice. The apparatus consisted of a square arena (50 cm × 50 cm × 40 cm) made of opaque gray PVC. The arena was divided into two zones by the analysis software: a central zone (25 cm × 25 cm) and a peripheral zone. At the beginning of each trial, a mouse was gently placed in the center of the arena and allowed to explore freely for 5 min. A camera mounted above the arena recorded the activity, which was subsequently analyzed. The total distance traveled, the time spent in the central zone, and the number of entries into the central zone were quantified. The arena was thoroughly cleaned with 75 % ethanol between each trial to eliminate olfactory cues. All tests were conducted in a dimly lit and sound-attenuated room.

2.9. Force Swimming test (FST)

Mice were individually placed in a transparent cylinder (height: 25 cm, diameter: 10 cm) filled with water (22–25 °C, 14–20 cm depth) to prevent tail touching the bottom. Each test lasted for 6 min, with the total immobility time recorded during the final 4 min. Immobility was defined as the absence of any motion except those necessary to keep the mouse afloat. Water was changed between trials to avoid olfactory cues.

2.10. Tail suspension test (TST)

For the TST, mice were suspended individually by the tail from a horizontal bar (distance from tip: ∼1 cm) fixed 30–50 cm above the testing surface, using adhesive tape. The test lasted 6 min. Immobility, defined as absence of limb/body movement except those required for respiration, was recorded during the last 4 min.

2.11. Pharmacological inhibition of mTOR (rapamycin)

USP11^−/− male mice were randomly assigned to three groups: KO-CON (vehicle), KO-CUMS (vehicle), and KO-CUMS + Rapa (rapamycin). CUMS procedure was performed as described above. Rapamycin (MedChemExpress, Cat. No. 53123-88-9) was formulated in a mixed vehicle containing 10% DMSO, 40% PEG300, 5% Tween-80, and 45% sterile 0.9% saline and prepared fresh before injection(Yang et al., 2025). Starting on day 14 of CUMS, mice received rapamycin (3 mg/kg at a final injection volume of 0.2 mL administered i.p.) or an equal volume of vehicle three times per week until 24 h before tissue collection (Hadamitzky et al., 2018). The same dosing schedule was maintained during behavioral testing, and injections were timed ≥24 h before each behavioral test to minimize potential acute drug effects.

2.12. Paraffin sections

After deep anesthesia, mice were transcardially perfused with 0.9 % saline followed by 4 % paraformaldehyde. The brains were quickly removed and post-fixed in 4 % paraformaldehyde at 4 °C overnight. The tissues were then dehydrated in a graded ethanol series, cleared in xylene, and embedded in paraffin. Serial coronal sections (thickness: 4 μm) were cut using a microtome and mounted onto glass slides. Sections were dried at 60 °C for 1 h and stored at room temperature until further use.

2.13. Immunofluorescence

Paraffin sections were dewaxed in xylene and rehydrated through graded ethanol. For antigen retrieval, sections were incubated in 0.01 M sodium citrate buffer (pH 6.0) at 94 °C for 30 min and cooled to room temperature naturally. After washing with PBS, sections were blocked with 3 % BSA containing 0.1 % Triton X-100 at room temperature for 1 h. Next, a hydrophobic pen was used to circle the sections, and the primary antibodies were added and incubated at 4 °C overnight. After washing, sections were incubated with fluorophore-conjugated secondary antibodies at room temperature in the dark for 1–2 h. Nuclei were counterstained with DAPI. Finally, sections were covered with antifade mounting medium and visualized using a confocal microscope. Image analysis was performed with ImageJ software.

2.14. Golgi-cox staining and morphometric analysis

Golgi-cox staining was performed as previously described. Briefly, mouse brain tissue was harvested and immersed in fixation solution for at least 48 h. The tissue was cut into 2–3 mm blocks, rinsed with physiological saline, then transferred to Golgi staining solution for 14 days. After three rinses with distilled water and overnight soaking in 80 % glacial acetic acid, the tissue was further rinsed, immersed in 30 % sucrose, sectioned at 100 μm on an oscillating slicer, and mounted on gelatin-coated slides. Sections were air-dried overnight in the dark, treated with concentrated ammonia, immersed in hardening fixative, washed and finally sealed for microscopic analysis of dendritic spine density.

For morphometric quantification, neurons were sampled from layer V of the prelimbic (PL) region of the mPFC based on classical pyramidal-like morphology. Only well-impregnated neurons with intact, non-truncated dendritic arbors and minimal overlap/occlusion were included. Spine density was quantified on clearly visualized distal segments of second–third order dendritic branches. Dendritic complexity was assessed by Sholl analysis (ImageJ/Neuroanatomy plugin) by counting intersections with soma-centered concentric circles at 10 μm intervals. For each mouse, 5 neurons were analyzed (Wilson et al., 2017). Because no genetic or immunohistochemical cell-type labeling was performed, neurons are referred to as pyramidal-like based on morphology.

2.15. Transmission electron microscopy (TEM) staining

For TEM, Fresh brain tissue blocks (1 mm^3) were collected within 3 min to minimize mechanical damage. After fixation in 1 % osmium tetroxide prepared with 0.1 M phosphate buffer (pH 7.4) at room temperature for 2 h, the tissue was processed for dehydration, osmotic embedding, polymerization, ultrathin sectioning (60–80 nm), dual staining with uranyl acetate and lead citrate, and finally observed and imaged using a TEM (Hitachi HT7800/HT7700). Synaptic structures were quantified using ImageJ Pro Plus.

2.16. Western blot

Tissue or cell lysates were prepared with RIPA buffer containing protease and phosphatase inhibitors. Protein concentrations were determined using a BCA kit (Thermo, USA). Equal amounts of protein were separated by SDS-PAGE and transferred to PVDF membranes. After blocking with 5 % BSA, membranes were incubated overnight at 4 °C with primary antibodies (e.g., anti-USP11, anti-GSK3β, anti-mTOR, anti-PSD95, etc.), followed by HRP-conjugated secondary antibody for 1 h at room temperature. Bands were visualized using ECL reagents and quantified with ImageJ or Image Lab.

2.17. Protein structure and protein–protein docking predictions

In this study, the AlphaFold Server (https://alphafoldserver.com/) was applied to predict the three-dimensional conformations of USP11 and GSK3β. The predicted structures were aligned with known templates, and their reliability was assessed based on the C-score, which evaluates model accuracy. To explore potential protein–protein interactions, docking of full-length USP11 and GSK3β was performed using the HDOCK web server (Yan et al., 2020) (http://hdock.phys.hust.edu.cn/). This platform generated multiple candidates binding conformations, which were then ranked through the ITScorePP scoring function. The docking process utilized a fast Fourier transform–based search algorithm for computational efficiency. Subsequently, possible interaction interfaces between USP11 and GSK3β were analyzed using the PDBePISA web tool (https://www.ebi.ac.uk/pdbe/pisa/). The ten highest-ranked docking models were preserved, and these computational predictions were integrated with experimental findings to determine the most credible interaction configuration.

2.18. Statistical analysis

All data were expressed as mean ± standard error of mean (SEM) and analyzed using GraphPad Prism software. Data with normal distribution and homogeneity of variance were analyzed by two-tailed t-test or one-way ANOVA (LSD post hoc). If not, non-parametric tests such as Kruskal-Wallis were applied. P < 0.05 was considered statistically significant. ∗P < 0.05, ∗∗P < 0.01, and ∗∗∗P < 0.001.

3. Results

3.1. CUMS elicits depression-related phenotypes and dysregulates GSK3β/mTOR signaling with increased USP11 expression

The figure provides a schematic overview of the experimental design (Fig. 1A), which includes a 1-week adaptation phase, a 4-week chronic unpredictable mild stress (CUMS) protocol, followed by a series of behavioral tests. Mice exposed to CUMS displayed a significant reduction in sucrose preference (SPT; Fig. 1B), traveled shorter distance in the open field test (OFT; Fig. 1C), and demonstrated increased immobility time in both the forced swim test (FST) and tail suspension test (TST; Fig. 1D and E) compared to control mice, thereby confirming the successful induction of depression-like phenotypes. Western blot analysis revealed significantly decreased phosphorylation of p-mTOR (Ser2448) and p-GSK3β (Ser9) relative to their total protein levels in the medial prefrontal cortex (mPFC) of CUMS mice (Fig. 1F and G), consistent with dysregulation of GSK3β/mTOR signaling observed in human MDD. Notably, USP11 protein levels were substantially increased, as determined by immunoblot densitometry and immunofluorescence analysis (Fig. 1H–K) indicating robust upregulation of USP11 in the mPFC following chronic stress exposure.

Fig. 1.

Chronic unpredictable mild stress promotes depression-like behaviors and upregulates USP11 in mouse prefrontal cortex.

(A) Schematic overview of the experimental timeline: male C57BL/6J mice underwent 1-week adaptation, followed by 4 weeks of chronic unpredictable mild stress (CUMS) and subsequent behavioral tests.

(B-E) SPT, OFT, FST, TST results in control (Ctrl) and CUMS groups (n = 8, SPT, Welch's t-test, p = 0.0204; OFT, p = 0.0101; FST, p = 0.0020; TST, p = 0.0078).

(F) Western blot of p-mTOR (Ser2448) (289 kDa), total mTOR (289 kDa), p-GSK3β(Ser9) (47 kDa), total GSK3β (47 kDa), and Tubulin (55 kDa) in mPFC tissue (n = 6).

(G) Quantification of p-mTOR/t-mTOR and p-GSK-3β/t-GSK-3β ratios (p-mTOR, Welch's t-test, Pp= 0.0023; p-GSK-3β, p = 0.0075).

(H) Western blot of USP11 (110 kDa) and Tubulin (55 kDa) in mPFC (n = 6).

(I) Quantification of USP11 protein normalized to Tubulin (p = 0.002).

(J) Representative immunofluorescence images for DAPI (blue, nuclear stain), USP11 (red), and merged panels in mPFC of control and CUMS mice. Scale bar: 50 μm.

(K) Mean USP11 immunofluorescence intensity quantification (n = 3, p = 0.0142).

Data are shown as mean ± SEM. Statistical analysis used two-tailed unpaired Student's t-test unless otherwise indicated. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

3.2. Deubiquitinase USP11 interaction with GSK3β

Immunoprecipitation coupled with mass spectrometry (IP-MS) identified GSK3β as a high-confidence USP11-interacting protein in the mouse mPFC (Fig. 2A). Co-immunoprecipitation assays using mPFC lysates (Fig. 2B) further validated this interaction, confirming physical association between USP11 and GSK3β in vivo. Due to extremely low endogenous USP11 expression in HEK293T cells, we transfected cells with Flag-USP11 alone and performed co-immunoprecipitation for GSK3β. The results demonstrated that exogenous USP11 efficiently co-precipitated GSK3β, confirmed correct cellular localization and structural integrity of exogenous USP11 (Fig. 2C). Heterologous co-expression of Flag-USP11 and His-GSK3β in HEK293T cells further validated this robust interaction through reciprocal pull-down assays (Fig. 2D–F). Dot blot analysis revealed a direct interaction between USP11 and GSK3β in vitro (Fig. 2G). Immunofluorescence studies in transfected HEK293T cells and primary neurons exhibited pronounced co-localization of USP11 and GSK3β (Fig. 2H), quantitatively substantiated by fluorescence intensity profiles and scatter plot analyses (Fig. 2I and J).

Fig. 2.

USP11 directly interacts with GSK3β.

(A) Volcano plot of proteins detected after USP11 immunoprecipitation from mouse mPFC. Log2 fold change (x-axis) shows enrichment versus control; log2 intensity (y-axis) reflects normalized quantitation in experimental samples. USP11 served as bait; GSK3β is highlighted as an interactor (log2 intensity USP11 = 22.9, log2FC = 2.38).

(B) Immunoprecipitation (IP) with anti-USP11 antibody, immunoblot (IB) detection for USP11 (110 kDa) and GSK3β (47 kDa). IP with anti-GSK3β or anti-USP11 antibody. Input: whole lysate; IgG: isotype control.

(C) Validation in HEK293T transfection system: lysates of vector control or Flag-USP11 transfected cells (Flag tag, 110 kDa) subjected to IP (anti-GSK3β), IB for anti-USP11.

(D) Cell lysate analysis of HEK293T single His-GSK3β, single Flag-USP11, or co-transfected groups, immunoblotted for His-GSK3β (47 kDa) and Flag-USP11 (110 kDa).

(E, F) Reciprocal Co-IP verification from HEK293T co-transfection. Immunoblot analysis for His and Flag tag in His-GSK3β, Flag-USP11, and co-transfected samples. (E) Lane 1: His-GSK3β group (IP-His), Lane 2: Flag-USP11 group (IP-Flag), Lane 3: Co-transfection (IP- His) (F) Lane 1: His -GSK3β group (IP- His), Lane 2: Flag-USP11 group (IP-Flag), Lane 3: Co-transfection (IP-Flag).

(G) Dot blot analysis showing specific binding between USP11 and GSK3β. BSA (100/200/500 ng) served as negative control, and purified USP11 (100/200/500 ng) was spotted on the same nitrocellulose membrane. After incubation with GSK3β protein solution, binding was detected by fluorescence imaging.

(H) Immunofluorescence analysis of co-localization: Exogenous expression in HEK293T cells demonstrates USP11 (red) and GSK3β (green); endogenous expression verified in primary neurons. Nuclei stained with DAPI (blue), scale bar = 25 μm.

(I) Fluorescence intensity profiles along linear ROIs: Gray values of USP11 (red) and GSK3β (green) measured with ImageJ. Dual-channel curves plotted in GraphPad Prism using exported data.

(J) Pearson's correlation scatter plots for USP11(red) and GSK3β(green) fluorescence, generated using ScatterJ plugin for ImageJ. Pearson's r value shown.

(K) Schematic of Flag-tagged USP11 fragment constructs used for pulldown mapping.

(L) HEK293T cells were co-transfected with Flag-USP11 or its deletion mutant and His- GSK3β, followed by immunoprecipitation and immunoblot analysis for Flag and His.

(M) Computational molecular docking predicts multiple direct contact sites between USP11 and GSK3β.

To delineate the interacting site, a series of USP11 truncation mutants were constructed (Fig. 2K). Domain mapping using His-pulldown with USP11 truncation mutants revealed that only the C-terminal fragments of USP11, particularly the region spanning amino acids 503–963, mediated strong binding to GSK3β; in contrast, the N-terminal fragment (1–502) failed to interact with GSK3β (Fig. 2L). Molecular docking simulations predicted multiple putative binding interfaces within the C-terminal segment of USP11 (Fig. 2M). Collectively, cellular, biochemical, and structural studies demonstrate a direct physical interaction between USP11 and GSK3β.

3.3. USP11 deubiquitinates GSK3β and inhibits activation of the mTOR signaling pathway

To investigate the enzyme–substrate relationship between USP11 and GSK3β, a catalytically inactive USP11 point mutant (USP11-C318S) was generated. HEK293T cells were transfected with Flag-USP11 (wild-type), catalytically inactive USP11-C318S, or an empty vector control. Immunoprecipitation-GSK3β (IP-GSK3β) assays showed that overexpression of USP11-WT significantly reduced GSK3β ubiquitination, while USP11-C318S exhibited no effect. These findings indicate that USP11 catalytic activity is necessary for deubiquitinating GSK3β, supporting GSK3β as a direct substrate of USP11 (Fig. 3A). The protein levels of total GSK3β and phosphorylated GSK3β at Ser9 remained unchanged across all treatment groups.

Fig. 3.

USP11 regulates GSK3β ubiquitination, phosphorylation, and synaptic protein homeostasis in neural cells

(A) Western blot analysis of GSK3β ubiquitination in HEK293T cells co-transfected with Flag-vector (control), Flag-USP11 (wild-type, 110 kDa), or Flag-USP11-C318S (catalytically inactive mutant). Endogenous GSK3β and phosphorylated GSK3β at Ser9 were immunoprecipitated from cell lysates using anti-GSK3β antibody, and ubiquitination levels were detected by immunoblotting with anti-ubiquitin antibody. GSK3β: 47 kDa; ubiquitin bands detected as smear.

(B) Western blot analysis of GSK3β phosphorylation in three 293T cell groups: wild-type (Ctrl), stable USP11-overexpressing line generated by lentiviral transduction (USP11-OE), and USP11-overexpressing cells subjected to siRNA knockdown (USP11-OE + siUSP11). siUSP11 was transfected to silence USP11 in the stable overexpressing cell line. Whole cell lysates were analyzed for endogenous USP11 (110 kDa), phosphorylated GSK3β at Ser9 (p-GSK3β, 47 kDa), total GSK3β (47 kDa), and GAPDH (35 kDa) as loading control. Representative results from n = 3 biological replicates per group.

(C) Gray value quantification of p-GSK3β/t-GSK3β in 293T cells (n = 3, F (2, 6) = 35.38, p = 0.0005).

(D) Western blot analysis of USP11 (110 kDa), phosphorylated mTOR (p-mTOR, Ser2448, 289 kDa), total mTOR (289 kDa), p-GSK3β (Ser9, 47 kDa), total GSK3β (47 kDa), and Tubulin (55 kDa) in primary neurons upon USP11 siRNA knockdown (n = 3).

(E, F) Gray value quantification of p-GSK3β/t-GSK3β, and p-mTOR/t-mTOR ratios in neurons upon USP11 siRNA knockdown (n = 3, p-GSK3β, p = 0.0213, p-mTOR, p = 0.0047).

(G) Immunoblot of USP11 (110 kDa), p-GSK3β (Ser9, 47 kDa), total GSK3β (47 kDa), SYN (77 kDa), and Tubulin (55 kDa) in primary neurons infected with adeno-associated virus (AAV) (n = 3).

(H, I) Gray value quantification of p-GSK3β/t-GSK3β, and SYN/Tubulin ratios in neurons transduced with vector or AAV-USP11 viruses (n = 3, p-GSK3β, p = 0.0078, SYN, Welch's t-test, p = 0.0031).

(J) Representative immunofluorescence of primary neurons transduced with vector or AAV-USP11 viruses, showing DAPI (blue, nuclei), SYN (green, synaptophysin), and USP11 (magenta); merged panels display synapse integrity. Scale bar: 50 μm.

(K, L) Quantitative analysis from three independent biological replicates in primary neurons transduced with vector or AAV-USP11 viruses (K) Mean USP11 immunofluorescence intensity (p = 0.0416), (L) Mean SYN immunofluorescence intensity (p = 0.0035).

Data are shown as mean ± SEM. Determined by t-test (baseline comparisons) or one-way ANOVA (multiple groups) unless otherwise indicated. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

To further examine the effect of USP11 on GSK3β activity while avoiding non-physiological alterations from transient overexpression, stable USP11-overexpressing cell lines and corresponding USP11 knockdown lines (via siRNA) were established. In stably transfected HEK293T cells, USP11 overexpression led to a marked reduction in GSK3β Ser9 phosphorylation, subsequent knockdown of USP11 restored Ser9 phosphorylation to baseline levels observed in parental 293T cells, which inherently express USP11 at very low levels (Fig. 3B and C). Further validation utilizing primary neuronal cultures was performed. Both siRNA knockdown and adeno-associated virus (AAV)-mediated USP11 overexpression in primary neurons confirmed a strong negative correlation between USP11 expression and the p-GSK3β(Ser9)/total GSK3β ratio. In primary neurons, USP11 knockdown led to increased GSK3β Ser9 phosphorylation and mTOR activation (Fig. 3D–F). In parallel, USP11 overexpression in neurons decreased p-GSK3β(Ser9), along with reduced synaptophysin (SYN) levels, according to immunoblotting and immunofluorescence analyses (Fig. 3G–L). Collectively, these results identify USP11 as a critical regulator of GSK3β activity and synaptic protein homeostasis.

3.4. USP11 knockout alleviates depression-like behaviors and associated with mTOR signaling

These findings suggest that USP11 negatively regulates synaptic integrity. To further investigate the effects of USP11, USP11 knockout (USP11-KO) mouse models were employed to assess molecular and behavioral outcomes. Molecular analyses showed that p-GSK3β (Ser9) levels were significantly elevated in USP11-KO mice compared to wild-type (WT) mice, accompanied by enhanced mTOR signaling (p-mTOR) and increased expression of the cortical synaptic marker PSD95 (Fig. 4A–E).

Fig. 4.

USP11 knockout alleviates stress-induced depressive-like behaviors and associated with mTOR Signaling

(A) Western blot analysis of USP11 (110 kDa), p-mTOR (Ser2448, 289 kDa), total mTOR (289 kDa), p-GSK3β (Ser9, 47 kDa), total GSK3β (47 kDa), PSD95 (95 kDa), and Tubulin (55 kDa) in mouse mPFC from wild-type (WT) and USP11 knockout (USP11^−/−) male mice (n = 6, Tubulin as loading control).

(B–E) Quantification of baseline protein band intensity in wild-type control (WT-CON) and USP11 knockout control (KO-CON) groups: (B) USP11 (relative to Tubulin, p < 0.0001), (C) p-GSK3β (relative to total GSK3β, p = 0.0072), (D) p-mTOR (relative to total mTOR, p = 0.0028), (E) PSD95 (relative to Tubulin, p = 0.0159). n = 6/group.

(F–I) Behavioral results for four groups: WT-CON, KO-CON, WT-CUMS, and KO-CUMS (OFT, distance [cm], F [3, 28] = 8.234, p = 0.0004; OFT, velocity [cm/s], F [3, 28] = 8.233, p = 0.0004; FST, F [3, 28] = 8.721, p = 0.0003; TST, F [3, 29] = 5.378, p = 0.0046). n = 8/group.

(J) Western blot analysis of USP11 (110 kDa), p-mTOR (Ser2448, 289 kDa), total mTOR (289 kDa), SYN (synaptophysin, 77 kDa), and Tubulin (55 kDa) in mPFC from all four groups (n = 3).

(K-M) Quantification of (K) USP11 (relative to Tubulin, F (3, 8) = 139.5, p < 0.0001), (L) p-mTOR (relative to total mTOR, F (3, 8) = 8.298, p = 0.0077), (M) SYN (relative to Tubulin, F (3, 8) = 8.811, p = 0.0065). n = 3/group.

(N) Schematic overview of the experimental design, including a 7-day acclimation period, a 28-day chronic unpredictable mild stress (CUMS) procedure, the rapamycin dosing regimen (3 mg/kg, i.p., three times per week; from day 14 of CUMS until 24 h before tissue collection), and the behavioral test battery in male USP11^−/− mice.

(O-R) Behavioral results for three groups in USP11^−/− mice: CON + Veh, CUMS + Veh and CUMS + Rapa. (SPT, F (2, 18) = 7.019, p = 0.0056; OFT, center time [s], F [2, 18] = 8.788, p = 0.0022; OFT, velocity [cm/s], F [2, 18] = 0.09090, p = 0.9135; TST, F [2, 18] = 7.797, p = 0.0036). n = 7/group.)

(T) Quantification of p-mTOR (relative to total mTOR, F (2, 6) = 38.49, p = 0.0004)

Data are shown as mean ± SEM. Determined by t-test (baseline comparisons) or one-way ANOVA (multiple groups) unless otherwise indicated. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

(S) Representative immunoblots of p-mTOR (Ser2448, 289 kDa), total mTOR (289 kDa) in USP11^−/− mice under the indicated conditions. (n = 3, Tubulin as loading control).

Behaviorally, following CUMS procedure, WT mice exhibited characteristic depressive-like behaviors, including reduced speed and distance in the OFT, increased immobility in the FST, confirming the reliability of the CUMS model. Remarkably, USP11 knockout mice exposed to CUMS exhibited relatively normal behavioral performance in the OFT, FST and TST, which markedly differed from the responses of WT mice under CUMS. Notably, KO-CON mice did not differ significantly from WT-CON mice in baseline behavioral measures, however, the behavioral differences between KO-CON and WT-CUMS mice were considerably greater than those observed between WT-CON and WT-CUMS groups, evidenced by higher distance and velocity in the OFT and lower immobility times in the FST and TST for KO-CON mice. This pronounced divergence demonstrates that USP11 knockout confers robust genetic resilience to chronic stress, specifically protecting against stress-induced depressive-like behaviors without altering baseline behavior (Fig. F–I).

Protein analysis of mPFC tissues showed that following CUMS exposure, USP11-KO mice retained higher levels of p-mTOR and synaptophysin (SYN) compared to WT mice, suggesting resistance to stress-induced decreases in synaptic protein expression. (Fig. 4J–M).

To further probe whether mTOR signaling contributes to this stress-resilient phenotype, USP11^−/− mice undergoing CUMS were treated with rapamycin (Fig. 4N). Rapamycin decreased p-mTOR (p-mTOR/total mTOR) and was associated with worsened behavioral outcomes (SPT, OFT center time, and TST immobility) without affecting OFT distance (Fig. 4O–T), indicating that the behavioral effects in USP11^−/− mice are mTOR-dependent.

Together, these data suggest that USP11 deficiency preserves synaptic marker expression under stress, at least partly through mTOR-linked signaling.

3.5. Ultrastructural and morphological analyses reveal synaptic protection by USP11 knockout

To directly visualize the pathological impact of USP11 on synapses, prefrontal cortex tissues from each group were examined by electron microscopy and quantitatively analyzed (Fig. 5A). Synaptic density in USP11-KO mice, both before and after CUMS exposure, was significantly greater than in WT control mice (Fig. 5B). Quantitative measurement of postsynaptic density (PSD) thickness indicated that USP11-KO mice exhibited increased PSD thickness compared to WT mice under both basal and CUMS conditions, with more pronounced differences following chronic stress (Fig. 5C). These findings support enhanced postsynaptic stability under chronic stress mediated by USP11 knockout. Golgi staining analysis focused on layer V pyramidal-like neurons in the mPFC revealed that dendritic spine density in USP11-KO mice was substantially higher than in WT mice following CUMS, indicating preserved dendritic spine structure in the absence of USP11 (Fig. 5D and E). Consistently, Sholl analysis revealed that USP11 knockout attenuated the CUMS-induced reduction in dendritic intersections across radial distances, indicating preserved dendritic arbor complexity, with the area under the Sholl curve providing a single, intuitive summary metric of overall dendritic arbor complexity. (Fig. 5F and G).

Fig. 5.

Ultrastructural and dendritic morphological analysis reveals preservation of synaptic integrity and neuronal complexity in USP11 knockout mice under chronic stress

(A) Representative transmission electron micrographs of the prefrontal cortex from WT-CON, KO-CON, WT-CUMS, and KO-CUMS mice, showing typical synaptic structures. Scale bar: 2 μm.

(B) Quantification of synapse number per field from electron micrographs (n = 3, F (3, 8) = 33.5, p < 0.0001). Synaptic density was significantly reduced in WT-CUMS compared to WT-CON, while KO-CUMS mice showed partial rescue.

(C) Measurement of postsynaptic density (PSD) thickness (nm) using ImageJ Pro Plus software on high-resolution electron micrographs (n = 3, 3 synapses per mouse, Brown-Forsythe ANOVA test, p = 0.0003).

(D) Golgi staining images of prefrontal cortical neurons (magnifications: 20 × , 60 × , 100 × ) displaying dendritic arborization and spine morphology for each group. Scale bar: 50 μm. Representative circular diagrams illustrate dendritic arborization complexity of typical prefrontal cortical neurons in each group (spacing: 5 μm) for visualization.

(E) Quantification of dendritic spine density (spines/μm) from Golgi-stained neurons (n = 3, F (3, 8) = 13.33, p = 0.0018). Each data point represents one independent biological sample, calculated as the within-mouse mean of spine measurements from 5 randomly selected neurons.

(F, G) Sholl analysis of Golgi-stained mPFC neurons. (F) The number of dendritic intersections as a function of radial distance from the soma (step size = 10 μm; maximum radius = 120 μm). Statistical analysis was performed using two-way repeated-measures ANOVA (group × radius), followed by post hoc multiple comparisons, as appropriate. (Distance: F (1.590, 13.88) = 45.80, p < 0.0001; Treatment: F (1.257, 40.21) = 39.54, p < 0.0001, Asterisks denote P values for WT-CUMS vs. KO-CUMS at each radius.) (G) Area under the Sholl curve was calculated from the intersection–radius profiles as an integrated, single-metric summary of overall dendritic arbor complexity. (n = 3, F (3, 8) = 40.72 p < 0.001, each data point represents one independent biological sample, calculated from 5 randomly selected neurons.)

Data are shown as mean ± SEM, statistical analysis by one-way ANOVA with Tukey's post hoc test unless otherwise indicated. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

In summary, ultrastructural and morphological analyses provide direct evidence that USP11 knockout mitigates stress-induced synaptic and dendritic structural damage, highlighting USP11 as a key negative regulator of synaptic structure and stress-induced synaptic remodeling. The maintained synapse number, PSD thickness, and dendritic spine density indicate that targeting USP11 could help preserve neural circuitry and cognitive function in stress-related depression.

4. Discussion

Numerous studies have reported altered ubiquitination in the brain tissues of patients with depression and relevant animal models, with the greatest alterations observed in the prefrontal cortex and hippocampus (Yan et al., 2025; Minelli et al., 2015; Bousman et al., 2010; Imai et al., 2012; Choi et al., 2018). Recently, genome-wide association studies (GWAS) and systematic proteomic analyses of postmortem human brain tissue have directly identified significant ubiquitin-related changes in the dorsolateral prefrontal cortex (DLPFC) of individuals with major depressive disorder (MDD) (Belaish et al., 2021). Cross-regional whole genome and transcriptome studies have further implicated the prefrontal cortex, revealing numerous differentially expressed genes and pathway enrichments, thus emphasizing the pivotal role and molecular pathological burden of the PFC in the biology of depression (Yan et al., 2025). Moreover, the extent of ubiquitin dysregulation is positively correlated with depression severity (Minelli et al., 2015; Mouri et al., 2016; Golan et al., 2013). However, despite the recognized regulatory potential of USP11 in neuropsychiatric disorders such as Alzheimer's disease (Yan et al., 2022), the precise roles of key ubiquitin system enzymes in synaptic pathology and synaptic structural plasticity impairment in MDD remain to be elucidated. Some progress has been made in elucidating the roles of ubiquitin-specific proteases (USPs) in depression. For example, USP7 has been shown to promote activation of the NLRP3 inflammasome, increasing secretion of interleukin-1β (IL-1β) and interleukin-18 (IL-18), changes that correlate with depression severity (Zhang et al., 2025). Another study reported that USP25 induction reduces hippocampal neurogenesis in mice and is associated with depressive-like behaviors (Cai et al., 2023). Collectively, these findings implicate the USP system in the pathophysiology of depression; however, given the large and functionally diverse USP family, current investigations represent only the tip of the iceberg. This study investigates the pathological role of the deubiquitinating enzyme USP11 in modulating synaptic integrity and stress-related structural remodeling in depression.

Based on the CUMS animal model, we demonstrated that USP11 expression is significantly upregulated in chronic stress–induced depression and is accompanied by dysregulation of the GSK3β/mTOR signaling axis. Protein interaction screening is highly consistent with various verification evidence, which clarifies the direct binding between USP11 and GSK3β molecules, and the key domain of this interaction is located at the C-terminus of USP11. Mechanistically, USP11 functions as a specific deubiquitinating enzyme for GSK3β, regulating its ubiquitination and Ser9 phosphorylation, which consequently modulates mTOR activity and downstream synaptic protein expression (e.g., PSD95, SYN). This study further expands our understanding of GSK3β regulation in vivo, revealing that its ubiquitination is closely linked to phosphorylation, and together, these modifications play a critical role in regulating synaptic protein homeostasis and stress-related synaptic structural remodeling.

Experiments in primary neurons and USP11 knockout mice further confirmed that modulation of this pathway is essential for stress-induced synaptic impairment and neuroprotection: elevated USP11 expression disrupts neuroprotective homeostasis and exacerbates synaptic damage under stress. In vivo pharmacological inhibition of mTOR with rapamycin further indicates that the neuroprotective phenotype conferred by USP11 loss is at least partly mTOR-dependent, rather than a coincidental parallel effect. Golgi staining and electron microscopy data further show the protective changes of synaptic/dendritic structure in USP11 knockout model mice: the number of synapses, dendritic branches, dendritic spine density and other structural integrity of USP11 deletion animals are more complete. From molecular to structural to behavioral phenotypes, multi-dimensional verification has proposed that the USP11-GSK3β-mTOR axis is a critical molecular mechanism regulating stress- and depression-associated synaptic pathology (Fig. 6).

Fig. 6.

USP11 knockout conferred synaptic protection under chronic stress. (with Figdraw).

Interestingly, although transient overexpression of USP11 in HEK293T cells significantly reduced the ubiquitination level of GSK3β, it failed to change the phosphorylation of GSK3β at Ser9 site, while in USP11 stably overexpressing cell lines, variations in USP11 abundance led to significant changes in the phosphorylation of GSK3β. This difference suggests that USP11-mediated deubiquitination and phosphorylation regulation show obvious time-course and cell homeostasis dependence: the former is a rapid response, while the latter depends on the adaptive establishment of USP11 continuous expression and cell signal homeostasis. Therefore, relying solely on the transient expression system may underestimate the dynamic complexity of the ubiquitin-kinase regulatory axis, emphasizing the value of long-term models in revealing the post-translational regulatory mechanisms within cell signal transduction networks.

The regulation of GSK3β signaling axis by USP11 is time/steady-state dependent, suggesting that the basal expression level of USP11 is essential for regulatory sensitivity. As an X-linked deubiquitinating enzymes, USP11 is overexpressed in female populations due to the random inactivation of the X chromosome. Future studies should investigate the influence of hormonal and genetic factors and include multi-group and diverse populations to advance understanding of specific intervention strategies. This study has several limitations. First, our study primarily assessed synaptic pathology at the molecular and structural levels, we did not perform in vivo or ex vivo electrophysiology to directly test how USP11 affects synaptic function or systematically profile synaptic receptor composition. Future electrophysiological studies will help link the USP11–GSK3β/mTOR mechanism to circuit-level synaptic function and behavioral phenotypes. Second, both the constitutive systemic USP11 knockout and systemic rapamycin treatment were performed at the whole-animal level; therefore, developmental compensation and contributions from non-neuronal cell types or peripheral tissues cannot be fully excluded, cell-type- and region-specific manipulations will be required to establish where USP11 acts in vivo to regulate stress susceptibility and synaptic pathology. Third, the molecular mechanisms underlying how USP11-mediated deubiquitination of GSK3β facilitates its phosphorylation remain to be elucidated. Finally, due to the current lack of potent and highly selective USP11 inhibitors, pharmacological validation of the potential antidepressant effects of USP11 inhibition could not be conducted at the animal level.

Overall, this study expands the understanding of depression pathogenesis and clarifies the specific contribution of ubiquitination to neuropsychiatric disorders. Importantly, it is the first to propose USP11 as a novel therapeutic target for depression, providing valuable insights for the development of targeted treatment strategies.

Funding

This work was supported by grants from the National Natural Science Foundation of China (grant number: U21A20364 and 82501827). This work has not received funding/assistance from any commercial organizations. The funding sources had no roles in the design of this study and will not have any roles during the execution, analyses, interpretation of the data, or decision to submit results.

CRediT authorship contribution statement

Ningyuan Li: Investigation, Project administration, Writing – original draft. Yuqi Feng: Investigation. Qian Gong: Data curation. Hao Duan: Investigation. Hanchun Yan: Investigation. Honghan Zhang: Investigation. Chao Wang: Funding acquisition, Project administration, Writing – review & editing. Zhongchun Liu: Funding acquisition, Project administration, Writing – review & editing.

Declaration of competing interest

The authors declare that there is no conflict of interest in this work.

Handling Editor: Prof R Lawrence Reagan

Data availability

Data will be made available on request.