The role of Dymeclin in chronic unpredictable mild stress-induced depression: maintaining the Golgi apparatus structure and regulating NLRP3 inflammasome activation

Abstract

The Golgi apparatus plays a key role in innate immune signaling, but its exact role in depression-related neuroinflammation remains unclear. The role of Dymeclin (DYM), a Golgi structural protein, in stress-related psychiatric disorders needs further clarification. This study aims to investigate whether DYM participates in the development of chronic stress-induced depression by regulating Golgi function. A mouse depression model was developed with chronic unpredictable mild stress (CUMS), followed by a rescue experiment using stereotactic injection of DYM-overexpressing lentiviruses into the hippocampus. Depression-associated behaviors, hippocampal metabolites, Golgi apparatus function, and inflammatory markers were examined using mass spectrometry, immunofluorescence, electron microscopy, and immunoprecipitation techniques. Overexpression of DYM mitigated depressive-like behaviors induced by CUMS in mice. Additionally, it enhanced the structural integrity of the Golgi apparatus, suppressed NLRP3 inflammasome activation, reduced levels of central and peripheral inflammatory factors, and partially ameliorated hippocampal metabolic disturbances. In vitro experiments further confirmed that DYM overexpression inhibited LPS-induced NLRP3 upregulation and inflammatory factor release in microglia. This study reveals that chronic stress causes depressive-like behaviors by reducing hippocampal DYM expression, leading to Golgi dysfunction and triggering NLRP3 inflammasome-related neuroinflammation and metabolic issues.

This finding broadens the traditional framework of depression research, which has primarily focused on neurons and monoamine neurotransmitters, by underscoring the pivotal role of Golgi organelle homeostasis in mediating the interactions among stress, neuro-immune, and metabolic processes. Our research indicates that targeting the DYM/Golgi pathway could offer new treatment strategies and a theoretical foundation for reducing depression-related neuroinflammation and metabolic issues. Drugs targeting the structural stability of the Golgi apparatus may serve as potential candidates for novel antidepressants in the future.

Supplementary Information

The online version contains supplementary material available at 10.1186/s10020-026-01524-2.

Introduction

Depression (major depressive disorder, MDD) is among the most common psychiatric disorders, affecting nearly 280 million people worldwide (Yang et al. 2022), Furthermore, the COVID-19 pandemic has exacerbated this situation, leading to a 27.6% increase in depressive disorders globally (Collaborators 2021). MDD is a highly heterogeneous psychiatric syndrome. Experimental and genetic studies have yielded several mechanisms, including maladaptive responses to stress with hypothalamic–pituitary–adrenal (HPA) axis dysregulation, inflammation, reduced neuroplasticity, circuit dysfunctions, and perturbations in neuromodulatory systems, such as monoaminergic and endocannabinoid (eCB) systems (Chevalier et al. 2020). Antidepressant medications based on these prevailing theories are widely used to treat MDD; however, 50% of treated patients relapse, and 30% are drug-resistant, suggesting that current hypotheses do not fully capture the heterogeneous pathophysiology of MDD (Yun et al. 2018). However, despite recent advances in the understanding of pathophysiology and treatment, the etiology of depression remains largely unexplained. Much work remains for the understanding and management of this disorder (Wang et al. 2021). Evidence indicates that MDD is often associated with neuroinflammation in specific brain regions in response to external stress (Howe et al. 2022). Recent studies have shown that NLRP3 agonists can induce NLRP3 inflammasome assembly by disrupting or damaging several cellular structures, including lysosomes, mitochondria, and the Golgi apparatus. The Golgi apparatus, endoplasmic reticulum, and mitochondrial connections are important for NLRP3 inflammasome activation (Boncompain and Weigel 2018). Many important organelles, including the Golgi apparatus, endosomes, endoplasmic reticulum, mitochondria, and peroxisomes, play key regulatory roles in neuroinflammation (Mao et al. 2020; Song et al. 2021; Xian et al. 2019). The Golgi apparatus serves as a hub for glycan and lipid synthesis, protein post-translational modification, and vesicle transport, which is crucial for maintaining cellular homeostasis. Moreover, as the central sorting organelle, the Golgi plays a key role in the modification and transport of innate immune receptors, adaptors, and products (Tao et al. 2020). For example, several inflammatory cytokines, such as interleukin (IL)-6 and interferon (IFN)-β, are secreted through the ER–Golgi pathway (Zhang et al. 2020a). Now, however, the role of Golgi in neuroinflammation and fine-grained regulatory mechanisms remain unclear.

Variations in the splicing sites of the dymeclin (DYM) gene lead to Dyggve-Melchior-Clausen disease and hearing loss, a disorder associated with chondrodysplasia that may involve defects in intracellular vesicle transport (Geneviève et al. 2005; Osipovich et al. 2008). The DYM gene-derived circDYM plays a key role in major depressive disorder (MDD). CircDYM, a homologous human and mouse circRNA, is derived from exons 4, 5 and 6 of the DYM gene. Our previous findings and those of others have indicated that circDYM could be a novel therapeutic target for alleviating depressive-like symptoms in various animal models (Yu et al. 2022; Zhang et al. 2020b). It inhibits nuclear translocation of CEBPB by binding to it, downregulates ZC3H4 expression, and thereby suppresses LPS-induced microglial apoptosis and neuroinflammation. Additionally, circDYM can inhibit downstream gene expression by binding to the transcription factor TAF1, thus alleviating inflammatory responses. To treat MDD, targeted delivery systems for the brain have been developed, such as RVG-circDYM-extracellular vesicles (EVs) and PANE-circDYM nanoemulsions (Gao et al. 2025b; Yu et al. 2022; Zhou et al. 2024). These systems effectively alleviate depressive behaviors in mouse models and inhibit microglial activation, blood-brain barrier leakage, and astrocyte dysfunction. The DYM gene encodes a 669-amino acid protein called Dymeclin, a Golgi-associated protein involved in Golgi organization and intracellular vesicle traffic. Dymeclin has been found to play an important role in maintaining the normal structure and function of the Golgi apparatus (Denais et al. 2011; Dupuis et al. 2015; Osipovich et al. 2008). However, the role of the Golgi apparatus protein DYM in depression remains completely unknown, and this is the key question addressed in the present study.

As the core hub of the intracellular membrane system, the Golgi apparatus not only participates in fundamental functions such as the transport, sorting, and modification of proteins and lipids, but also plays a key role in innate immune regulation (Spano and Colanzi 2022; Zhao et al. 2023; Zhong et al. 2025). It serves as a platform for the activation of the STING signaling pathway (via vesicular transport between the endoplasmic reticulum and the Golgi apparatus), and can integrate cross-talk between lipid metabolic signals and NF-κB immune activation (Fei et al. 2023; He and Zhang 2024). In the nervous system, the Golgi apparatus participates in neuronal morphogenesis through dendrite-specific structures such as Golgi outposts, and its structural abnormalities (e.g., fragmentation) are associated with cognitive impairments in Alzheimer’s disease (Chen et al. 2023b; Kang et al. 2024; Liu et al. 2024). Furthermore, dysregulation of Golgi homeostasis is closely associated with various diseases, including neurodegenerative diseases, cancer, and cardiovascular diseases. Moreover, ion channels such as the CCX4 calcium transporter (which regulates pH) and stress responses critically influence Golgi function (Kanamori et al. 2024; Liu et al. 2021, 2024; Zhong et al. 2025). In depression research, abnormalities in Golgi apparatus-related innate immune signaling (e.g., downregulation of AHI1 expression) have been confirmed to be associated with impaired antiviral immune function, supporting the critical role of neuro-immune interactions in mood disorders (Lee and Wheeler 2026; McCallum and Perreault 2021; Zhang et al. 2022). Current direct evidence for the role of the Golgi apparatus in depression remains limited; however, given its regulatory functions in synaptic transmission, neuroinflammation, and stress responses, further exploration is needed (Chen et al. 2023b; Niu et al. 2025; Wang et al. 2022). Therefore, investigating the potential role of the Golgi apparatus in depression and identifying Golgi-related therapeutic targets for mental disorders such as MDD is an important research priority. In summary, building on prior findings, we sought to explore the role of DYM and the Golgi apparatus in neuroinflammation associated with depression.

Method and materials

Ethics approval

All animal procedures were performed in strict accordance with the guidelines and regulations established by the Animal Ethics Committee. The care and use of animals were reviewed and approved by the Institutional Animal Care and Use Committee at the Nanjing First Hospital, Nanjing Medical University.

Materials

The concentrations of cytokines, specifically IL-6, IL-1β, MCP-1, and TNF-α, in peripheral plasma and hippocampal extracts were quantified using commercially available ELISA kits, following the manufacturer’s protocols. Antibodies against Dymeclin (1:1000, Proteintech, 12001-1-AP), NLRP3 (1:1000, Proteintech, 27458-1-AP), GRASP65 (1:1000, Proteintech, 10747-2-AP), and GAPDH (1:3000, Proteintech, 60004-1-Ig) were purchased from Proteintech (Wuhan, China). The ELISA kits for Mouse IL-6 (KE10007), Mouse IL-1β (KE10003), Mouse MCP-1 (KE10006), and Mouse TNF-α (KE10002) were also obtained from Proteintech, Wuhan, China. Lipopolysaccharide (LPS) derived from Escherichia coli O111:B4 was obtained from Sigma-Aldrich (L2630, Sigma, USA). In vivo transduction was performed using a lentiviral system to overexpress the mouse Dymeclin (Dym) gene. The lentivirus carrying the DYM overexpression vector was purchased from Hanheng Bio (Shanghai, China). The core coding sequence (CDS) corresponds to RefSeq NM_027727.3, which is 1,476 base pairs in length and encodes a 491-amino acid protein (NP_082003.1). This sequence has been validated by the National Center for Biotechnology Information (NCBI) and the Mouse Genome Informatics (MGI) database (Gene ID: 69190). The encoded Dymeclin protein is critical for Golgi apparatus organization and skeletal development. The human homolog of this gene is associated with osteochondrodysplasias, including Dyggve-Melchior-Clausen disease.

Animals

Adult male C57BL/6J mice (20∼25 g) were purchased at 5 weeks of age from the Model Animal Research Center of Yangzhou University (Yangzhou, China). After one week of acclimatization, all mice were group-housed under a constant temperature and maintained on a 12-hour light/dark cycle (lights on at 7:00 AM, lights off at 7:00 PM). Mice were allowed to adlibitum access to water and food.

Lentivirus microinjection for DYM overexpression

Prior to the experiments, each mouse was weighed and subsequently assigned randomly to different groups. Male C57BL/6J mice were administered bilateral hippocampal injections of either Control-GFP or DYM-GFP lentivirus, with each injection consisting of 1 µl containing 1 × 10⁹ viral genomes/µl. Specific details may be referred to our previous research (Zhang et al. 2020b). The injection coordinates were precisely set at 2.06 mm posterior to the bregma, ± 1.5 mm lateral, and 2 mm in depth. To assess the impact of DYM on microglial activation induced by CUMS, the mice were subjected to the CUMS protocol for a duration of four weeks prior to undergoing tests for depression-like behaviors.

Fluoxetine injections

Fluoxetine powder was purchased from MedChemExpress (HY-B0102). Stock solutions were prepared fresh in PBS on the day of injection. Starting 14 days after CUMS treatment initiation, mice received either fluoxetine (10 mg/kg/day) or an equivalent volume of sterile PBS by intraperitoneal injection for 14 consecutive days (Tunc-Ozcan et al. 2025).

Chronic unpredictable stress (CUMS) protocol and depression-like behavior tests

To assess neurological function, mice underwent a series of behavioral tests with at least 24 h between each: Sucrose Preference Test, Open Field Test, Forced Swim Test, and Tail Suspension Test. To prevent fatigue effects, experimenters unaware of group assignments conducted all tests and analyses.

CUMS protocol

The Chronic Unpredictable Mild Stress (CUMS)-induced depressive mouse model is a widely recognized and utilized animal model for studying depression, particularly in the context of elucidating the mechanisms underlying various antidepressant agents (Liao et al. 2021; Zhang et al. 2020b). In this model, mice are subjected to two or three unpredictable mild stressors each day over a period of four weeks. These stressors encompass a range of conditions, including a 5-minute swim in ice-cold water, 6 h of physical restraint, 24-hour periods of fasting, deprivation of drinking water, placement on wet bedding, absence of bedding, reversal of the light-dark cycle, and 3 h of cage tilting. More details are provided in Supplementary information 1.

Depression-like behavior tests

The Open Field Test (OFT) assessed exploratory, anxiety-like, and depressive behaviors, as well as antidepressant efficacy, in mouse depression studies (Ma et al. 2022). The Open Field Test (OFT), the sucrose preference test, the Forced Swim Test (FST), and the Tail Suspension Test (TST) are used to evaluate depression-like behaviors in mice. Animals were tracked and recorded with SMART 3.0 software (Panlab, Spanish). More details are provided in Supplementary information 1.

Cell cultures

Microglial cells were cultured from the brains of 1–2 day old C57BL/6J mice using a standard method (Zhang et al. 2020b). Brain tissue was mechanically and enzymatically dissociated, then seeded into poly-L-lysine-coated flasks with DMEM, 10% fetal bovine serum, and 1% penicillin/streptomycin. Cultures were kept at 37 °C with 5% CO₂, with medium changes every three days. CSF2/GM-CSF was added to promote microglial growth. Microglia were collected by agitation and centrifugation at 1500 g for 5 min. More details are provided in Supplementary information 1.

Lentiviral transduction of microglia with GFP

Microglia were transduced with LV-GFP lentivirus. Specific details may be referred to our previous research (Zhang et al. 2020b).Initially, 1 × 10⁴ cells per well were cultured in DMEM with 10% FBS for 48 h. The medium was then replaced with fresh medium containing 8 µg/mL polybrene and 100 µL of lentiviral solution, and cells were incubated at 37 °C with 5% CO₂ for 24 h. Post-transduction, the medium was changed to DMEM with 10% FBS until cells were over 50% confluent. For selection, cells were treated with DMEM containing 10 µg/mL puromycin and 10% FBS at 37 °C with 5% CO₂ for 24 h, followed by two washes with fresh DMEM and 10% FBS. More details are provided in Supplementary information 1.

Western blotting (WB)

Mice were anesthetized, and their brain tissue was extracted. The hippocampus was isolated on ice, homogenized in RIPA lysis buffer with inhibitors, and centrifuged. The supernatant was mixed with sample buffer, heated, and subjected to SDS-PAGE electrophoresis. Proteins were transferred to PVDF membranes, blocked with 5% milk, and incubated overnight at 4 °C with primary antibodies for Dymeclin, NLRP3, GRASP65, and GAPDH. After washing, membranes were treated with secondary antibodies for 1 h at room temperature. Protein bands were visualized using chemiluminescence and analyzed with ImageJ software. More details are provided in Supplementary information 1.

Analysis quantitative real-time PCR

Total RNA was extracted from cells using TRIzol (Invitrogen). The quantitative real-time PCR (qRT-PCR) experiments were performed using SYBR-Green reagents (Novizan Biotechnology Co., Nanjing, China) with specific primers for DYM ( forward primer 5’- TCAGGCCCCGAGTCTATATCT-3′ and reverse primer, 5′- GATTTCCTGTTCGAGGGTTGTT-3′), GRASP65 (5’- AAACAAGGAGATTTTCCTCCGC-3′ and reverse primer, 5′- CCGTCAGGCTCTCATATCGAAT-3′) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH forward primer, 5’- ATCTGCACGGGGTACAAGAGA-3′ and reverse primer, 5’- GGGGTCGTTGATGGCAACA-3′).

Enzyme-linked immunosorbent assay (ELISA)

Male C57/BL6 mice were anesthetized via intraperitoneal injection. Hippocampal tissue (100 mg) was rinsed with PBS, homogenized in 1 mL PBS, mechanically sheared, and stored at -20 °C overnight. The homogenate underwent two freeze-thaw cycles for membrane disruption, then centrifuged at 5000 g for 5 min at 4 °C to collect the supernatant. Cell culture supernatants were also centrifuged at 1000 g for 5 min at 4 °C. Cytokine levels (IL-6, IL-1β, MCP-1, and TNF-α) in both supernatants and hippocampal extracts were measured using ELISA kits, following the manufacturer’s instructions. Further details are in the Supplementary information 1.

Metabolomic analysis

High-performance liquid chromatography-mass spectrometry (LC-MS) analysis of brain tissues and plasma, following sample processing, was conducted using a Thermo UHPLC-Q Exactive system. The detection and data processing methodologies in metabolomics are informed by prior research studies (Furman et al. 2017; Want et al. 2006, 2013; Yuan et al. 2012). Further details are in the Supplementary information 1.

Electron microscopy experiments

The hippocampal tissue was observed under a transmission electron microscope (TECNAI 10, Philips Company, Netherlands). Based on previous studies, the changes in Golgi apparatus function are quantified by calculating the percentage of Golgi fragmentation (Ferguson et al. 2017; Renna et al. 2018; Zhang et al. 2019). Further details are in the Supplementary information 1.

Immunofluorescence staining

Immunofluorescence staining was performed on paraffin brain sections as previously described (Jin et al. 2014). Further details are in the Supplementary information 1.

Analysis

Statistical analysis was performed using GraphPad Prism 9.5. Data are presented as mean ± SD. Student t-tests compared two groups, while one-way ANOVA with Bonferroni’s post hoc test compared variables across three or more groups, with significance set at P < 0.05.

Result

MDD is associated with Golgi dysfunction

Mice were exposed to a CUMS protocol for 4 weeks followed by depressive-like behavior tests. Starting 14 days after CUMS treatment initiation, mice received either fluoxetine (10 mg/kg/day) or an equivalent volume of sterile PBS by intraperitoneal injection for 14 consecutive days. The experimental procedure is shown in Fig. 1A. The Open Field Test (OFT) results revealed that the Chronic Unpredictable Mild Stress (CUMS) group exhibited a significant reduction in time spent in the central area and decreased ambulation (Fig. 1B-D). The Sucrose Preference Test (SPT), a reliable indicator of anhedonia, showed a marked decrease in sucrose consumption in the CUMS group compared to the control group (Fig. 1E). Additionally, the Forced Swim Test (FST) and Tail Suspension Test (TST) demonstrated increased immobility time (Fig. 1F and G, respectively), further corroborating the presence of depressive-like behaviors in CUMS-induced depressed mice relative to controls. Notably, fluoxetine treatment significantly ameliorated the CUMS-induced alterations observed in the OFT, SPT, FST, and TST. In the hippocampal tissue of CUMS mice, we observed that the Golgi structure was loose and irregular, with granular and fragmented Golgi visible. Electron microscopic observations simultaneously revealed that fluoxetine can significantly ameliorate the structural damage to the Golgi apparatus in the hippocampal brain region induced by CUMS, with a significant reduction in the degree of Golgi apparatus fragmentation (Fig. 1H-I). The mRNA levels of DYM and GRASP65 in CUMS group were significantly decreased (Figs. 1J-K).The protein levels of DYM and GRASP65 in CUMS group were significantly decreased, while the expression level of NLRP3 was significantly upregulated (Fig. 1L-O).

Fig. 1.

Chronic unpredictable mild stress (CUMS) induces Golgi dysfunction in the depression model of mouse. A Schematic diagram of the experimental procedure. Mice were exposed to a CUMS protocol for 4 weeks followed by depressive-like behavior tests, n = 10 per group. B-D The open field test (OFT). B Heatmaps of locomotor activity. C Distance in the center zone of the OFT. Interaction, F (1, 36) = 6.032, P = 0.0190; CUMS, F (1, 36) = 152.2, P < 0.0001; DYM, F (1, 36) = 12.18, P = 0.0013. D The time spent in the center zone of the OFT. Interaction, F (1, 36) = 4.898, P = 0.0333; CUMS, F (1, 36) = 175.8, P < 0.0001; DYM, F (1, 36) = 6.872, P = 0.0128. E Sucrose preference test (SPT) results. Interaction, F (1, 36) = 41.91, P < 0.0001; CUMS, F (1, 36) = 274.4, P < 0.0001; DYM, F (1, 36) = 40.79, P < 0.0001. F Immobility time in the forced swim test (FST). Interaction, F (1, 36) = 12.33, P = 0.0012; CUMS, F (1, 36) = 295.8, P < 0.0001; DYM, F (1, 36) = 45.34, P < 0.0001. G Immobility time in the tail suspension test (TST). Interaction, F (1, 36) = 11.64, P = 0.0016; CUMS, F (1, 36) = 168.8, P < 0.0001; DYM, F (1, 36) = 34.46, P < 0.0001. Data are presented as mean ± SD; ****P < 0.0001 vs. the Control + PBS group; ##P < 0.01 and ####P < 0.0001 vs. the CUMS + PBS group (Two-way ANOVA with Bonferroni’s post hoc test), n = 10 per group. H-I Golgi apparatus in hippocampus tissue using electron microscopy in CUMS mice. H Golgi mophology. Mitochondria (red arrow), Golgi apparatus (green arrow). Scale bar, 1 μm (n = 3 per group). I The percentage of golgi with fragmented Golgi. Interaction, F (1, 8) = 16.00, P = 0.0039; CUMS, F (1, 8) = 100.0, P < 0.0001; DYM, F (1, 8) = 16.00, P = 0.0039. Data are presented as mean ± SD; ****P < 0.0001 vs. the Control + PBS group; ##P < 0.01 vs. the CUMS + PBS group (Two-way ANOVA with Bonferroni’s post hoc test), n = 3 per group. J The mRNA levels of DYM in the hippocampus. Data are presented as mean ± SD, Student’s t-test, t = 8.141, ****p < 0.0001, n = 6 per group. K The mRNA levels of GRASP65 in the hippocampus. Data are presented as mean ± SD, Student’s t-test, t = 18.74, ****p < 0.0001. n = 6 per group. L-O Western blot analysis of NLRP3, GRASP65, and DYM protein expression, n = 6 per group. M The protein levels of NLRP3. Student’s t-test, t = 9.297. N The protein levels of DYM. Student’s t-test, t = 8.355. O The protein levels of GRASP65. Student’s t-test, t = 8.038. Data are presented as mean ± SD; Data are presented as mean ± SD, ****p < 0.0001vs Control group

Restoration of DYM expression rescues depression-related behaviors in CUMS mice

Having determined that DYM was decreased in CUMS mice, we next sought to verify the role of DYM in vivo by microinjecting either the Control-GFP lentivirus or DYM-GFP lentivirus into the hippocampus of C57BL/6J mice. The experimental procedure is shown in Fig. 2A. One week after lentivirus microinjection, mice were subjected to a 4-week CUMS protocol, followed by behavioral tests for depressive-like behaviors. As shown in Fig. 2B-D, the CUMS-treated group spent less time and traveled a shorter distance in the center area of the OFT. The SPT also decreased in CUMS treatment group compared with control (Fig. 2E). CUMS-induced depressive mice exhibited significant increases in immobility duration relative to the control group in FST (Fig. 2F) and TST (Fig. 2G). As expected, these effects were significantly ameliorated in DYM-injected mice.

Fig. 2.

Overexpression of DYM rescues CUMS-induced depressive-like behaviors. A Schematic diagram of the experimental procedure. B-D Results of the open field test (OFT). B Heatmaps of movement trajectories in the OFT. C Distance in the center zone of the OFT. Interaction, F (1, 36) = 17.13, P = 0.0002; CUMS, F (1, 36) = 208.4, P < 0.0001; DYM, F (1, 36) = 32.75, P < 0.0001. D The time spent in the center zone of the OFT (the right panel). Interaction, F (1, 36) = 6.179, P = 0.0177; CUMS, F (1, 36) = 113.5, P < 0.0001; DYM, F (1, 36) = 9.187, P = 0.0045. E The sucrose preference test (SPT) results. Interaction, F (1, 36) = 55.73, P < 0.0001; CUMS, F (1, 36) = 545.2, P < 0.0001; DYM, F (1, 36) = 37.76, P < 0.0001. F Immobility time in the FST. Interaction, F (1, 36) = 17.89, P = 0.0002; CUMS, F (1, 36) = 90.79, P < 0.0001; DYM, F (1, 36) = 4.467, P = 0.0416. G Immobility time in the TST. Interaction, F (1, 36) = 2.065, P = 0.1593; CUMS, F (1, 36) = 190.9, P < 0.0001; DYM, F (1, 36) = 19.54. P < 0.0001. Data are presented as mean ± SD; ****P < 0.0001 vs. the Control+ Control-GFP lentivirus group; ##P < 0.01, ###P < 0.001 and ####P < 0.0001 vs. the CUMS+ Control-GFP lentivirus group (Two-way ANOVA with Bonferroni’s post hoc test), n = 10 per group. Abbreviation: ACT, active

Restoration of DYM expression mitigates neuroinflammation in CUMS-induced stress

After demonstrating that Golgi protein DYM plays a key role in mouse depressive-like behaviors, we attempted to validate the in vivo function of DYM by microinjecting DYM lentivirus into the hippocampus of C57BL/6J mice. The mice were monitored to determine the role of DYM in depression pathogenesis, as illustrated in Fig. 3A. After microinjection of lentivirus, we then examined the efficacy of DYM-GFP lentivirus transduction in vivo. As shown in Fig. 3B-C, increased protein expression level of DYM was observed in DYM-injected mice compared with Control-injected mice. Electron microscopy of the hippocampus revealed that CUMS treatment caused structural loosening of the Golgi apparatus, and DYM overexpression improved this structural change (Fig. 3D-E). QPCR analysis demonstrated that the overexpression of DYM markedly counteracted the CUMS-induced reduction in mRNA expression levels of both DYM and GRASP65 in the hippocampal region of the brain (Fig. 3F-G). To determine the role of DYM in microglial activation in the hippocampus, immunofluorescence staining was used to detect the expression of Iba1 and DYM. CUMS treatment significantly increased the expression level of Iba1, while overexpression of DYM significantly reduced the expression level of Iba1 (Fig. 3H-J). One week after lentiviral administration, mice were subjected to CUMS for four weeks, and the expression of NLRP3, GRASP65, and DYM in hippocampal tissue was detected. As shown in Fig. 3K-N, compared with the control group, CUMS treatment induced upregulation of NLRP3 expression levels in the hippocampal brain region, while the expression levels of Golgi apparatus proteins DYM and GRASP65 were downregulated. However, overexpression of DYM significantly inhibited the upregulation of NLRP3 in the hippocampal brain region, as well as the downregulation of the expression levels of Golgi apparatus proteins DYM and GRASP65. In addition, ELISA results showed that compared with the Con group, the levels of IL-6, IL-1β, MCP-1, and TNF-α in the hippocampus and plasma of the CUMS group were significantly increased. However, restoring DYM expression could inhibit the elevation of cytokine (IL-6, IL-1β, MCP-1, and TNF-α) levels induced by CUMS in mouse hippocampal cells (Fig. 3O-R). The same results for cytokine levels in the plasma of the CUMS model further confirmed the same results in the mouse hippocampus (Fig. 3S-V).

Fig. 3.

Restoration of DYM expression attenuates neuroinflammation in CUMS mice. A Schematic diagram of the in vivo validation experimental. B-C The protein expression levels of DYM in the hippocampus, Student’s t-test, t = 16.99. Data are presented as mean ± SD; Data are presented as mean ± SD, ****p < 0.0001vs Control group. D Observation of Golgi apparatus in hippocampus tissue using electron microscopy in CUMS mice, Nucleus (blue arrow), mitochondria (red arrow), Golgi apparatus (green arrow). Scale bar, 1 μm (n = 3 per group). E The percentage of golgi with fragmented Golgi. Interaction, F (1, 8) = 18.75, P = 0.0025; CUMS, F (1, 8) = 154.1, P < 0.0001; DYM, F (1, 8) = 24.08, P = 0.0012. Data are presented as mean ± SD; ***P < 0.001 vs. the Control + PBS group; ##P < 0.01 vs. the CUMS + PBS group (Two-way ANOVA with Bonferroni’s post hoc test), n = 3 per group. F-G The mRNA levels of DYM and GRASP65 in the hippocampus. F The mRNA levels of DYM. Interaction, F (1, 20) = 77.13, P < 0.0001; CUMS, F (1, 20) = 135.2, P < 0.0001; DYM, F (1, 20) = 475.2, P < 0.0001. n = 6 per group. G The mRNA levels of GRASP65. Interaction, F (1, 20) = 18.70, P = 0.0003; CUMS, F (1, 20) = 34.74, P < 0.0001; DYM, F (1, 20) = 196.7, P < 0.0001. n = 6 per group. Data are presented as mean ± SD; ****P < 0.0001 vs. the Control+ Control-GFP lentivirus group; ####P < 0.0001 vs. the CUMS+ Control-GFP lentivirus group (two-way ANOVA with Bonferroni’s post hoc test), n = 6 per group. H Immunofluorescence co-staining and quantitative analysis of Iba1 (microglial marker, red) and DYM (green) in the hippocampal dentate gyrus. Scale bar, 50 μm, n = 3 per group. I-J Quantification of mean fluorescence intensity of Iba1. Interaction, F (1, 8) = 16.34, P = 0.0037; CUMS, F (1, 8) = 134.0, P < 0.0001; DYM, F (1, 8) = 21.01, P = 0.0018. (I) The intensity of DYM. Interaction, F (1, 8) = 8.086, P = 0.0217; CUMS, F (1, 8) = 74.94, P < 0.0001; DYM, F (1, 8) = 180.4, P < 0.0001. J The intensity of DYM. Data are presented as mean ± SD; *P < 0.05 and ****P < 0.0001 vs. the Control+ Control-GFP lentivirus group; ##P < 0.01 and ###P < 0.001 vs. the CUMS+ Control-GFP lentivirus group (two-way ANOVA with Bonferroni’s post hoc test), n = 3 per group. K-N Western blot analysis of NLRP3, GRASP65, and DYM protein expression, n = 6 per group. L The expression of NLRP3, Interaction, F (1, 20) = 73.49, P < 0.0001; CUMS, F (1, 20) = 760.1, P < 0.0001; DYM, F (1, 20) = 85.16, P < 0.0001. M The expression of DYM, Interaction, F (1, 20) = 0.5556, P = 0.4647; CUMS, F (1, 20) = 497.1, P < 0.0001; DYM, F (1, 20) = 163.0, P < 0.0001. N The expression of GRASP65, Interaction, F (1, 20) = 7.805, P = 0.0112; CUMS, F (1, 20) = 216.5, P < 0.0001; DYM, F (1, 20) = 63.43, P < 0.0001. Data are presented as mean ± SD; ****P < 0.0001 vs. the Control+ Control-GFP lentivirus group; ####P < 0.0001 vs. the CUMS+ Control-GFP lentivirus group (two-way ANOVA with Bonferroni’s post hoc test), n = 6 per group. O–R Cytokine levels (IL-6, IL-1β, MCP-1, TNF-α) in hippocampal tissue. O IL-6. Interaction, F (1, 20) = 35.59, P < 0.0001; CUMS, F (1, 20) = 189.6, P < 0.0001; DYM, F (1, 20) = 36.08, P < 0.0001. P IL-1β. Interaction, F (1, 20) = 35.09, P < 0.0001; CUMS, F (1, 20) = 489.4, P < 0.0001; DYM, F (1, 20) = 41.91, P < 0.0001. Q MCP-1. Interaction, F (1, 20) = 103.9, P < 0.0001; CUMS, F (1, 20) = 1205, P < 0.0001; DYM, F (1, 20) = 112.5, P < 0.0001. R TNF-α. Interaction, F (1, 20) = 41.05, P < 0.0001; CUMS, F (1, 20) = 188.1, P < 0.0001; DYM, F (1, 20) = 42.01, P < 0.0001. n = 6 per group. S–V Cytokine levels (IL-6, IL-1β, MCP-1, TNF-α) in plasma. S IL-6. Interaction, F (1, 20) = 178.9, P < 0.0001; CUMS, F (1, 20) = 1414, P < 0.0001; DYM, F (1, 20) = 206.0, P < 0.0001. T IL-1β. Interaction, F (1, 20) = 81.08, P < 0.0001; CUMS, F (1, 20) = 764.6, P < 0.0001; DYM, F (1, 20) = 96.03, P < 0.0001. U MCP-1. Interaction, F (1, 20) = 77.74, P < 0.0001; CUMS, F (1, 20) = 565.0, P < 0.0001; DYM, F (1, 20) = 88.89, P < 0.0001. V TNF-α. Interaction, F (1, 20) = 43.66, P < 0.0001; CUMS, F (1, 20) = 432.2, P < 0.0001; DYM, F (1, 20) = 46.54, P < 0.0001. Data are presented as mean ± SD; ****P < 0.0001 vs. the Control+ Control-GFP lentivirus group; ####P < 0.0001 vs. the CUMS+ Control-GFP lentivirus group (two-way ANOVA with Bonferroni’s post hoc test), n = 6 per group. Abbreviation: ACT, active

Metabolomics reveals that restoration of DYM expression modulates the hippocampal metabolic profile in CUMS mice

To better understand how CUMS and DYM restoration affect the metabolic profile, we performed untargeted metabolomics using UPLC-MS. First, we investigated the global metabolic differences in the negative and positive modes by using PCA (Fig. 4A-H). Biplot of individuals and variables in hippocampal metabolites under negative and positive ion mode (Supplementary Fig. 2A–H). The corresponding biplot delineated the principal compounds contributing to the differentiation among groups. The x-axis and y-axis represented the first and second principal components (PC1 and PC2), respectively. In both positive and negative ion modes, lysophosphatidylcholines, lysophosphatidylethanolamines, and lysophosphatidylserines were predominantly located in the lower-left quadrant across the four groups (control group, Con + DYM group, CUMS-treated group, and CUMS + DYM group) and specifically between the control and CUMS groups characterized by negative values on both PC1 and PC2 (Supplementary Fig. 2A, E and 2 C, G). This distribution suggests pathological lipid metabolic disturbances in the hippocampal region of the brain in the CUMS model. The results revealed that the different groups of samples were well-distinguished. The supervised method PLS-DA was applied to evaluate the ability of a sample set to differentiate between groups in both positive and negative modes (Fig. 4Q-X). Distinct separation was observed among the various treatment groups, with some overlap between the CUMS and CUMS + DYM groups, while the CUMS group remained distinct from the Con group. Our data revealed the restoration of DYM expression may be able to reduce the effects of CUMS-induced hippocampus metabolic abnormalities.

Fig. 4.

Multivariate statistical analysis of the effects of DYM on hippocampal metabolites in CUMS mice. The 2D score plot of PCA plots in hippocampal metabolites under negative ion mode (A–D). The 3D score plot of PCA plots in hippocampal metabolites under negative ion mode (E-H). The 2D score plot of PCA plots in hippocampal metabolites under positive ion mode (I–L). The 3D score plot of PCA plots in hippocampal metabolites under positive ion mode (M–P). The 2D PLS-DA scattered plots of metabolomic profiles in hippocampus under negative ion mode (Q–T). The 2D PLS-DA scattered plots of metabolomic profiles in hippocampus under positive ion mode (U–X). “C” denotes the Control group, “D” signifies the Con + DYM group, “M” refers to the CUMS treatment group, and “MD” indicates the CUMS + DYM group. n = 5 per group

The effect of DYM on hippocampus metabolites

In hippocampal samples, a total of 430 and 367 metabolites were detected in positive and negative ion modes, respectively. Based on untargeted metabolomics profiling of the hippocampus, the detected metabolites were classified into ten chemical categories according to the Human Metabolome Database 4.0. In the negative ion mode, lipids and lipid-like molecules predominated (130 metabolites, 42.62%), followed by organic acids and derivatives (72, 23.61%), nucleosides, nucleotides, and analogues (32, 10.49%), organoheterocyclic compounds (24, 7.87%), organic oxygen compounds (22, 7.21%), benzenoids (12, 3.93%), phenylpropanoids and polyketides (11, 3.61%), and organic nitrogen compounds (2, 0.66%) (Supplementary Fig. 1A–D). In the positive ion mode, the major classes were lipids and lipid-like molecules (113, 40.65%), organic acids and derivatives (72, 25.90%), organoheterocyclic compounds (34, 12.23%), organic nitrogen compounds (16, 5.76%), nucleosides, nucleotides, and analogues (16, 5.76%), benzenoids (12, 4.32%), phenylpropanoids and polyketides (8, 2.88%), organic oxygen compounds (4, 1.44%), alkaloids and derivatives (2, 0.72%), and hydrocarbons (1, 0.36%) (Supplementary Fig. 1E–H). Orthogonal partial least-squares discriminant analysis (OPLS-DA) revealed maximal metabolic discrimination between the Control and CUMS groups as well as between the CUMS and CUMS + DYM groups (Supplementary Fig. 3A–H). To screen candidate biomarkers, receiver operating characteristic (ROC) curve analyses were conducted, and the top 15 most important metabolites in each ion mode were selected for model construction (Supplementary Fig. 4A–H). Ultimately, 20 potential metabolites were identified (Fig. 5A–T). Within the CUMS depression model, lysophosphatidylethanolamine LPE 18:1, LPE 18:2, stearic acid, stearoyl ethanolamide, oleamide, arachidonoyl amide, stearamide, p-mentha-1,3,8-triene, glutathione (oxidized), and 12-oxo leukotriene B4 exhibited significantly elevated levels, whereas LPC 20:4-SN1, adenosine, methandrostenolone, indole-3-acetic acid, and sphinganine were markedly decreased. Overexpression of DYM significantly inhibited the CUMS-induced upregulation of LPE 18:1, LPE 20:2, stearic acid, stearoyl ethanolamide, oleamide, and arachidonoyl amide.

Fig. 5.

Differential metabolite levels in the hippocampus across experimental groups. Candidate metabolic biomarkers of screened out by statistical analysis (A-T). A LPE 18:1. Interaction, F (1, 16) = 10.13, P = 0.0058; CUMS, F (1, 16) = 3.222, P = 0.0916; DYM, F (1, 16) = 0.8984, P = 0.3573. B LPE 18:2. Interaction, F (1, 16) = 8.139, P = 0.0115; CUMS, F (1, 16) = 2.292, P = 0.1495; DYM, F (1, 16) = 0.2584, P = 0.6182. C LPE 20:2. Interaction, F (1, 16) = 5.367, P = 0.0341; CUMS, F (1, 16) = 4.214, P = 0.0568; DYM, F (1, 16) = 2.231, P = 0.1547. D LPC 20:4-SN1. Interaction, F (1, 16) = 1.066, P = 0.3172; CUMS, F (1, 16) = 1.550, P = 0.2310; DYM, F (1, 16) = 2.886, P = 0.1087. E Stearic acid. Interaction, F (1, 16) = 13.62, P = 0.0020; CUMS, F (1, 16) = 8.047, P = 0.0119; DYM, F (1, 16) = 0.9635, P = 0.3409. F Stearoyl Ethanolamide. Interaction, F (1, 16) = 6.373, P = 0.0225; CUMS, F (1, 16) = 5.632, P = 0.0305; DYM, F (1, 16) = 3.177, P = 0.0937. G Oleamide. Interaction, F (1, 16) = 15.20, P = 0.0013; CUMS, F (1, 16) = 15.31, P = 0.0012; DYM, F (1, 16) = 1.188, P = 0.2919. H Arachidonoyl amide. Interaction, F (1, 16) = 10.91, P = 0.0045; CUMS, F (1, 16) = 9.542, P = 0.0070; DYM, F (1, 16) = 4.977, P = 0.0404. I Adenosine. Interaction, F (1, 16) = 0.2045, P = 0.6572; CUMS, F (1, 16) = 41.99, P < 0.0001; DYM, F (1, 16) = 5.072, P = 0.0387. J Indole-3-lactic acid. Interaction, F (1, 16) = 9.438, P = 0.0073; CUMS, F (1, 16) = 37.18, P < 0.0001; DYM, F (1, 16) = 0.9508, P = 0.3440. K p-Mentha-1,3,8-triene. Interaction, F (1, 16) = 5.703, P = 0.0296; CUMS, F (1, 16) = 3.620, P = 0.0752; DYM, F (1, 16) = 0.2987, P = 0.5922. L Methandrostenolone. Interaction, F (1, 16) = 1.910, P = 0.1859; CUMS, F (1, 16) = 2.476, P = 0.1351; DYM, F (1, 16) = 16.42, P = 0.0009. M Stearamide. Interaction, F (1, 16) = 4.586, P = 0.0480; CUMS, F (1, 16) = 6.673, P = 0.0200; DYM, F (1, 16) = 2.493, P = 0.1339. N Resolvin E1. Interaction, F (1, 16) = 1.519, P = 0.2355; CUMS, F (1, 16) = 1.657, P = 0.2163; DYM, F (1, 16) = 1.707, P = 0.2098. O Glutathione (oxidized). Interaction, F (1, 16) = 0.5746, P = 0.4595; CUMS, F (1, 16) = 3.249, P = 0.0903; DYM, F (1, 16) = 0.8317, P = 0.3753. P L-Glutathione oxidized. Interaction, F (1, 16) = 0.3932, P = 0.5395; CUMS, F (1, 16) = 2.749, P = 0.1168; DYM, F (1, 16) = 0.8924, P = 0.3589. Q 12-Oxo leukotriene B4. Interaction, F (1, 16) = 4.450, P = 0.0510; CUMS, F (1, 16) = 2.741, P = 0.1173; DYM, F (1, 16) = 1.792, P = 0.1993. R Adenosine 5’-Diphosphate. Interaction, F (1, 16) = 2.457, P = 0.1365; CUMS, F (1, 16) = 2.021, P = 0.1744; DYM, F (1, 16) = 1.854, P = 0.1922. S Sphinganine. Interaction, F (1, 16) = 0.7204, P = 0.4085; CUMS, F (1, 16) = 3.164, P = 0.0943; DYM, F (1, 16) = 0.1586, P = 0.6957. T (±)8(9)-DiHET. Interaction, F (1, 16) = 0.1531, P = 0.7007; CUMS, F (1, 16) = 2.177, P = 0.1595; DYM, F (1, 16) = 0.5697, P = 0.4613. U Class of differential metabolites between groups in the hippocampus under negative ion mode. V Class of differential metabolites between groups in the hippocampus under positive ion mode. W The metabolic pathway analysis of hippocampus metabolites under negative ion mode. X The metabolic pathway analysis of hippocampus metabolites under positive ion mode. Y Heatmap of relative content of differential metabolites among groups in the hippocampus under negative and positive ion mode. Data are presented as mean ± SD; *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001 vs. the Control+ Control-GFP lentivirus group; #P < 0.05 and ##P < 0.01 vs. the CUMS+ Control-GFP lentivirus group (two-way ANOVA with Bonferroni’s post hoc test), n = 5 per group. Abbreviation: ACT, active

Restoration of DYM expression inhibits microglia activation via Golgi stress in vitro

Given that microglia play an important role in neuroinflammation, the present study investigated the effect and potential mechanism of DYM action in microglia. DYM overexpression lentiviruses were transduced into microglia and primary microglia, and DYM protein expression levels were detected (Fig. 6A). Compared with the control group, LPS treatment induced upregulation of NLRP3 expression levels and downregulation of Golgi apparatus proteins DYM and GRASP65 in microglia. However, overexpression of DYM significantly inhibited the upregulation of NLRP3 and the downregulation of Golgi apparatus proteins DYM and GRASP65 expression levels in Microglia and primary microglia (Fig. 6B). To investigate the effect and potential mechanism of the action of DYM in microglia, we further examined the effect of DYM. DYM significantly inhibited the increased level of cytokines (IL-6, IL-1β, MCP-1 and TNF-α) induced by LPS in microglia cells (Fig. 6C-F).

Fig. 6.

DYM inhibits microglial activation via Golgi stress pathways in vitro. A Western Blot validation of DYM-overexpressing virus transduction efficiency in primary microglia. Microglia were transduced with the Con/DYM-GFP lentivirus for 24 h. Data are presented as mean ± SD, Student’s t-test, t = 25.11, ****p < 0.0001. n = 3 per group. B Western Blot and quantitative analysis of NLRP3, GRASP65, and DYM protein expression in primary microglia after LPS stimulation. NLRP3. Interaction, F (1, 8) = 71.89, P < 0.0001; CUMS, F (1, 8) = 820.6, P < 0.0001; DYM, F (1, 8) = 123.5, P < 0.0001. DYM. Interaction, F (1, 8) = 36.56, P = 0.0003; CUMS, F (1, 8) = 1096, P < 0.0001; DYM, F (1, 8) = 483.0, P < 0.0001. GRASP65. Interaction, F (1, 8) = 12.71, P = 0.0073; CUMS, F (1, 8) = 678.4, P < 0.0001; DYM, F (1, 8) = 179.3, P < 0.0001. After transduction with the Con/DYM-GFP lentivirus, primary mouse microglial cells were treated with LPS (100 ng/ml) for 24 h, and cytokine levels (IL-6, IL-1β, MCP-1, TNF-α) were determined in the cell culture supernatants. C The level of IL-6. Interaction, F (1, 20) = 15.91, P = 0.0007; CUMS, F (1, 20) = 76.49, P < 0.0001; DYM, F (1, 20) = 17.89, P = 0.0004. D The level of IL-1β. Interaction, F (1, 20) = 7.575, P = 0.0123; CUMS, F (1, 20) = 160.4, P < 0.0001; DYM, F (1, 20) = 16.30, P = 0.0006. E The level of MCP-1. Interaction, F (1, 20) = 35.03, P < 0.0001; CUMS, F (1, 20) = 649.2, P < 0.0001; DYM, F (1, 20) = 30.55, P < 0.0001. F The level of TNF-α. Interaction, F (1, 20) = 19.47, P = 0.0003; CUMS, F (1, 20) = 205.1, P < 0.0001; DYM, F (1, 20) = 26.03, P < 0.0001. Data are presented as mean ± SD of 3 independent experiments. **P < 0.01 and ****P < 0.0001 vs. the Control group; ##P < 0.01, ###P < 0.001 and ####P < 0.0001 vs. the LPS+ Control-GFP lentivirus group. Abbreviation: ACT, active

Discussion

This study elucidates the critical role of DYM in a chronic unpredictable mild stress-induced mouse model of depression. We demonstrate for the first time that DYM expression is significantly reduced in the hippocampus of CUMS-exposed mice, and its restoration ameliorates depressive-like behaviors, mitigates Golgi apparatus dysfunction, modulates hippocampal metabolic disturbances, and suppresses neuroinflammation (Kang et al. 2024; Liu et al. 2024). These findings position DYM as a potential key player in the pathophysiology of depression, particularly through its involvement in regulating Golgi stress and microglial activation.

Research on the role of the Golgi apparatus in depression is still in its early stages. However, its structural dynamics, interactions with other organelles, and transport abnormalities may contribute to neurological disease progression by affecting neurotransmitter release, synaptic remodeling, and inflammatory responses (He and Zhang 2024; Liu et al. 2024; Ulecia-Moron et al. 2025; Xu et al. 2024b). Investigating the specific molecular mechanisms of the Golgi apparatus in depression and its feasibility as a therapeutic target is of great significance.

Golgi apparatus dysfunction represents an important pathological substrate for depression. In chronic unpredictable mild stress (CUMS)-induced depressive models, mice exhibited abnormalities in Golgi-related structures in the hippocampus, including endoplasmic reticulum expansion and Golgi hypertrophy, accompanied by downregulation of key Golgi structural proteins such as DYM, GM130, and GRASP65. This phenomenon is similar to the morphological changes observed in neurodegenerative diseases (e.g., Alzheimer’s disease), indicating that impaired Golgi structural integrity is a common pathological feature of neuropsychiatric disorders (Artan et al. 2022; Kang et al. 2024; Lei et al. 2024; Liu et al. 2021; Meng et al. 2023; Song et al. 2024). Notably, the Golgi apparatus not only serves as a hub for protein processing and sorting but also its structural stability directly affects the efficiency of inter-organellar membrane transport (Liu et al. 2021; Meng et al. 2023). Additionally, it has been reported that the density of glial cells, including astrocytes and oligodendrocytes, is significantly reduced in the prefrontal and limbic brain regions of patients with MDD, which may be intertwined with this Golgi pathology (Lei et al. 2024). Astrocytes are critical for the glutamate/GABA cycle, and oligodendrocytes are essential for white matter integrity; both largely depend on intact Golgi function to transport proteins and maintain membranes (Li et al. 2022b; Liu et al. 2021; Meng et al. 2023; Xie et al. 2022). The mammalian Golgi apparatus plays a crucial role in protein transport, processing, and localization. Under stress, it triggers a Golgi stress response. Its structure can be dynamically regulated to adapt to specific cellular needs. Research has shown that under Golgi stress conditions, Golgi structure is disrupted and GM130 expression is downregulated (Oberholtzer et al. 2024). Our data suggest that DYM deficiency potentially induced by chronic stress may disrupt this neuroglial homeostasis. Partial reversal of Golgi structural alterations upon DYM overexpression (Fig. 5C) highlights the role of DYM in maintaining Golgi integrity, which may occur through interactions with cytoskeletal components or other structural proteins (Li et al., 2022a; Xu et al. 2025). Disruption of Golgi structure leads to fragmentation of its ribbon-like form and disassembly of the trans-Golgi network (TGN) into dispersed vesicles (dTGN). These vesicles facilitate NLRP3 clustering, which triggers polymerization of the adaptor protein ASC and activates downstream signaling pathways (Gong et al. 2018). In the current study, CUMS induced a significant pro-inflammatory state, characterized by upregulated NLRP3 expression in the hippocampus, accompanied by an increase in Iba1-positive cells (a marker of microglial activation) in the hippocampus and elevated levels of IL-6, IL-1β, MCP-1, and TNF-α in both the hippocampus and plasma. Overexpression of DYM significantly suppressed NLRP3 upregulation and the cytokine storm. As a platform for innate immune signaling, the Golgi apparatus undergoes dispersal, which can recruit NLRP3 and promote its oligomerization (Leishman et al. 2024). DYM inhibits microglia-mediated neuroinflammation by blocking the aggregation of NLRP3 on dispersed Golgi apparatus through the maintenance of Golgi structural integrity (Williams and Peden 2024). In vitro experiments in primary microglial cells confirmed that DYM can inhibit LPS-induced NLRP3 inflammasome pathway activation and the release of related cytokines (Dusabimana et al. 2023). This anti-inflammatory effect may involve multiple mechanisms. The role of DYM in Golgi homeostasis may be crucial, as Golgi stress has increasingly been recognized as a trigger for inflammasome activation and promotion of inflammatory responses (Dusabimana et al. 2023; Hu et al. 2023).

Recent metabolomic analyses of the hippocampal region have demonstrated that lysophosphatidylcholines, lysophosphatidylethanolamines, and lysophosphatidylserines predominantly localize in the lower left quadrant, characterized by negative values on both PC1 and PC2 axes, across four experimental groups: the control group, Con + DYM group, CUMS-treated group, and CUMS + DYM group. This pattern is observed under both positive and negative ion modes and is consistent between the control and CUMS groups. These findings suggest the presence of pathological lipid metabolic disturbances in the hippocampus within the CUMS model. The Golgi apparatus is integral to lipid metabolism, encompassing lipid modification, sorting, and transport (Fasimoye et al. 2023). This study has identified structural and functional disruptions of the Golgi apparatus in the hippocampal region of CUMS mice, which are ameliorated by the overexpression of DYM. These observations imply that the Golgi protein DYM may play a significant role in the pathogenesis of depression through its regulatory effects on lipid metabolism.

A KEGG enrichment analysis was performed on the top 15 differential metabolites, revealing significant involvement in metabolic pathways including glycerophospholipid metabolism (LPEs), linoleic acid metabolism (12-oxo-leukotriene B4), and tryptophan metabolism (indole-3-lactic acid). These results align with the established functions of the Golgi apparatus. Phospholipids, such as glycerophospholipids, are produced through complex enzymatic processes in the endoplasmic reticulum, mitochondria, and Golgi apparatus (Morita and Ikeda 2022). Glycerophospholipid metabolism involves membrane remodeling in the Golgi apparatus. Lysophosphatidylethanolamines (LPEs), important byproducts of glycerophospholipids, undergo metabolic changes that are crucial for regulating cellular membranes(Yuan et al. 2025). The metabolism of linoleic and α-linolenic acids is closely tied to glycerophospholipid pathways, and disruptions in glycerophospholipid metabolism often lead to significant changes in arachidonic acid metabolism (Gao et al. 2025a; Zhou et al. 2023). Glycerophospholipids, notably phosphatidylcholine and phosphatidylethanolamine, function as principal membrane reservoirs for arachidonic acid. Enzymes from the phospholipase A2 (PLA2) family, including PLA2G4A and LPCAT3, are localized within cellular organelles such as the Golgi apparatus. These enzymes play a crucial role in the liberation and remodeling of arachidonic acid within phospholipids, thereby facilitating the production of downstream eicosanoids, including leukotrienes(Kondreddy et al. 2025; Murakami et al. 2022; Zhang et al. 2021; Zhou et al. 2023). Furthermore, tryptophan can be metabolized by gut microbiota or mammalian cells to yield various indole metabolites, such as indole-3-lactic acid. These indole metabolites have been demonstrated to modulate immune cell function and influence cellular metabolism through the aryl hydrocarbon receptor (AhR) pathway(Zhang et al. 2026). Lipid exchange and membrane contact sites are present between lysosomes and the Golgi apparatus, exemplified by lysosomes receiving hydrolases from the Golgi (Mahanty et al. 2024; Nuchel et al. 2026; Pearson et al. 2024). Indole metabolites may indirectly contribute to Golgi-mediated lipid metabolic regulation by altering the membrane lipid environment or modulating inter-organellar signaling.

Current research on small molecule metabolites associated with disorders involving Golgi apparatus dysfunction and other organ impairments remains notably scarce. Existing studies have elucidated a significant association between Golgi apparatus dysfunction and metabolic disturbances, particularly in lipid metabolism. The Golgi apparatus is integral to lipid metabolism, encompassing lipid modification, sorting, and transport(Fasimoye et al. 2023). Clinical evidence suggests that serum levels of lysophosphatidylcholine (LPC) are markedly elevated in patients with Parkinson’s disease (PD) and correlate with disease progression(Paul et al. 2021). LPC functions not only as a biomarker for the disease but also actively participates in neurotoxic processes. Research has demonstrated that LPC can upregulate Cleaved-Caspase3 by activating the ERK signaling pathway downstream of the orphan receptor GPR35, thereby inhibiting the expression of the Golgi structural protein GRASP65 and resulting in the disruption of Golgi polarization(Jacquemyn et al. 2026; Paul et al. 2021). Our current study further corroborates this finding by demonstrating elevated levels of LPC 16:0 and LPC 17:0 in the chronic unpredictable mild stress (CUMS) model. Indole-3-lactic acid (IAA), an indole compound derived from the metabolism of tryptophan by gut microbiota, exhibits a range of biological functions(Nie et al. 2023; Wu et al. 2026). Recent studies have identified a significant association between Golgi apparatus PAPS-dependent sulfation and IAA(Xu et al. 2024c; Zhang et al. 2025). Specifically, IAA facilitates the synthesis and transport of PAPS, which are essential for mucin sulfation within the Golgi apparatus, through the AhR-Papss2-Slc35b3 signaling pathway(Cui et al. 2026). This process indirectly supports the proper modification and secretion of mucins by the Golgi apparatus, ultimately enhancing intestinal barrier integrity(Cui et al. 2026; Li et al. 2024). In our study, we observed that the overexpression of DYM mitigates the reduction of Indole-3-lactic acid levels induced by CUMS.

Our untargeted metabolomics analysis revealed that chronic unpredictable mild stress (CUMS) induces significant hippocampal metabolic disturbances, which can be partially reversed by DYM overexpression. Identified metabolites, including elevated lysophospholipids (LPE 18:1, LPE 18:2), stearic acid derivatives, and altered neuroactive lipids (such as oleamide and arachidonyl amide), suggest dysregulation in lipid metabolism and bioenergetics (He et al. 2023; Li et al., 2023b; Lin et al. 2025; Xu et al. 2024a). These lipids function not only as structural components but also as signaling molecules influencing synaptic plasticity, neuroinflammation, and emotional behavior (Choe and Park 2026; Ge et al. 2025; Liu et al. 2026). The reversal of these metabolic changes by DYM expression restoration indicates its involvement in maintaining metabolic homeostasis. These metabolic alterations may underlie the widespread cerebral structural and functional disturbances observed in patients with major depressive disorder (MDD) (Chen et al. 2025; Choe and Park 2026; Ge et al. 2025). Metabolic disturbances associated with DYM deficiency may lead to dysfunction in neural networks by affecting neuronal and glial membrane integrity, receptor trafficking, and energy supply (Ge et al. 2025; He et al. 2023).

Elevated LPE 18:1 is linked to inflammation, as seen in COVID-19, where it correlates with inflammatory lipid changes and may worsen disease severity (Wei et al. 2023). LPEs, as inflammatory molecules, can disrupt cell membrane fluidity, affected macrophage membrane structure and signaling pathways like TNFR1-mediated p38-MK2, thus promoting inflammation (Sit et al. 2024). In Golgi dysfunction, lipid disorders like LPE 18:1 accumulation can activate the NLRP3 inflammasome by disturbing lipid raft environments. Phosphatidylinositol 4-phosphate (PI4P) on the trans-Golgi network is crucial for NLRP3 localization. Electrostatic interactions between NLRP3 and the Golgi membrane aid its recruitment to trans-Golgi vesicles, facilitating inflammasome assembly(Leishman et al. 2024; Williams and Peden 2024). This mechanism connects Golgi axis dysfunction to neuroinflammation by activating NLRP3, which releases pro-inflammatory factors like IL-1β and IL-18, contributing to neurodegenerative diseases (Chen et al. 2023a; Possemato et al. 2023).

Based on our findings and existing literature, we propose an integrative model for DYM in depression pathogenesis. Chronic stress induces downregulation of hippocampal DYM, leading to impaired Golgi apparatus function, which disrupts protein processing and trafficking (Supplementary Fig. 4). Concurrently, DYM deficiency promotes a pro-inflammatory microglial state, likely via facilitation of NLRP3 inflammasome assembly and activation—mechanisms potentially involving disrupted lipid raft dynamics or organelle stress (Anand 2025; Arumugam et al. 2022; Dusabimana et al. 2023; Jin et al. 2022; Leishman et al. 2024; Liu et al. 2022; Zhang et al. 2023; Zhu et al. 2024). These neuroinflammatory and metabolic perturbations collectively impair neural circuit function in emotion-related brain networks, resulting in depressive-like behaviors. Notably, DYM exhibits anti-inflammatory properties; its capacity to suppress microglial activation and NLRP3 inflammasome activity positions it as a promising endogenous regulatory mechanism (Li et al. 2023a; Liu et al. 2022; Xia et al. 2023; Zhu et al. 2024). Our data indicated that targeting DYM or its downstream pathways may represent a novel therapeutic strategy for depression, addressing both neuroinflammatory and metabolic components of the disorder. As a Golgi apparatus-specific protein, overexpression of DYM can simultaneously ameliorate structural abnormalities, metabolic disturbances, and neuroinflammation, making it a potential target for multi-dimensional intervention in depression (Fig. 7). Notably, Golgi apparatus-related proteins (e.g., TM9SF3) serve as prognostic markers in cancer (Yang et al. 2025a; Yang et al. 2025b), and the protective role of DYM in depression models suggests its clinical translation value. Furthermore, activation of the innate immune molecule STING in the Golgi apparatus depends on the polymerization of sulfated glycosaminoglycans (sGAGs) (Fang et al. 2021), and whether DYM affects such metabolites remains to be explored, providing a new insight for metabol-immune cross-research. DYM improves depression-like behaviors induced by chronic unpredictable mild stress (CUMS) by maintaining Golgi apparatus structural stability, regulating lipid metabolic homeostasis, and inhibiting NLRP3 inflammasome activation. This study not only deepens the pathological significance of Golgi apparatus dysfunction in depression but also provides experimental evidence for anti-depressive strategies targeting organelle homeostasis.

Fig. 7.

Schematic diagram of the molecular mechanism involving Golgi protein DYM dysfunction in depression. Chronic stress downregulates hippocampal DYM expression, leading to Golgi apparatus structural damage and functional disruption. Golgi dysfunction disrupts metabolic homeostasis on one hand, and promotes the assembly and activation of the NLRP3 inflammasome in hippocampal microglia on the other, triggering neuroinflammation. Neuroinflammation and metabolic disturbances collectively impair neural circuit function, ultimately leading to depressive-like behaviors. DYM overexpression can break this vicious cycle by maintaining Golgi integrity, exerting antidepressant effects

Limitations

This study highlights DYM’s role but has limitations. The molecular mechanisms by which DYM affects Golgi structure and the NLRP3 inflammasome need clarification, including whether DYM directly interacts with inflammasome components or influences them via membrane trafficking. The cell-type specificity of DYM’s effects, beyond microglial activation, should be studied, as it may also impact astrocytes and neurons. Additionally, understanding the upstream regulators of DYM expression, such as the impact of chronic stress and potential involvement of glucocorticoid signaling or epigenetic mechanisms, is important. Translational research should examine if DYM levels change in MDD patients and their correlation with symptoms or treatment outcomes.

Conclusion

This study highlights Dymeclin as a key link between chronic stress, Golgi apparatus stability, and neuroinflammation in depression. Chronic stress reduces hippocampal DYM, causing Golgi fragmentation, NLRP3 inflammasome activation, and metabolic issues, leading to depressive behaviors. Restoring DYM improves Golgi integrity, reduces NLRP3 activation, and normalizes inflammation and metabolism. These findings suggest that maintaining Golgi homeostasis is crucial in depression’s stress-neuroimmune-metabolic pathway. Targeting the DYM/Golgi/NLRP3 axis could be a promising antidepressant strategy, with potential for developing drugs that stabilize Golgi structure. Drugs targeting the structural stability of the Golgi apparatus may serve as potential candidates for novel antidepressants in the future.

Authors’ contributions

Longfei Du contributed to conceptualization, writing-original draft preparation and Data Analysis. Zifan Ding, Lei Sun and Ye Qin Contributed to writing-reviewing and editing. Shukui Wang, Yuqin Pan and Huiling Sun contributed to writing (both original draft and review and editing) and supervision.

Funding

This study was supported by grants from the China Postdoctoral Science Foundation (No.2023M731747), Jiangsu Provincial Medical Key Discipline Cultivation Unit (JSDW202239), the National Natural Science Foundation of China (No. 82001425) and Double-Innovation Doctor of Jiangsu Province (No.2020–30199).

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

All authors approved the final version of the manuscript and agreed to be accountable for all aspects of the work.

Competing interests

The authors declare no competing interests.