Regulatory T cells promote microglia-mediated synapse engulfment and functional recovery via the OPN-CD74 axis after spinal cord injury in mice

Abstract

Spinal cord injury (SCI) is a debilitating neurological condition characterized by permanent sensory and motor dysfunction. While clearance of tissue debris represents a critical step in establishing a regenerative microenvironment after SCI, the underlying mechanisms remain incompletely understood. Regulatory T cells (Tregs) have emerged as critical immunomodulators in neurological diseases, with prior studies demonstrating their neuroprotective effects mediated through microglial regulation. As resident macrophages in the central nervous system (CNS), microglia play essential roles in debris clearance after SCI. Moreover, microglia-mediated synaptic elimination is crucial for maintaining tissue integrity and neural circuit function in neurological pathologies. However, it remains unclear whether and how Tregs influence microglial phagocytic activity, particularly synaptic engulfment post-SCI. In this study, we observed robust infiltration of Tregs into the injured spinal cords of both SCI patients and mouse models. Selective depletion of Tregs impaired the microglial phagocytosis of synaptic debris and reduced synapse density in mice post-SCI. Single-cell RNA sequencing and flow cytometry analyses revealed that microglial Cd74 expression was significantly upregulated following Tregs depletion. Remarkably, genetic ablation of Cd74 rescued the phagocytic deficits and mitigates reductions in synaptic density observed in Treg-deficient SCI mice. Osteopontin (OPN), a multifunctional cytokine implicated in regulating neuroinflammation, has previously been shown to mediate Treg-microglia interactions in stroke. Here, we demonstrated that Treg-derived OPN suppressed microglial CD74 expression, enhanced synaptic engulfment, and improved neurological outcomes after SCI. Collectively, our findings highlight a novel OPN-CD74 regulatory axis through which Tregs modulate microglial phagocytic function, offering new translational targets for SCI treatment.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12974-025-03661-7.

Introduction

Spinal cord injury (SCI), commonly caused by traumatic incidents such as motor vehicle accidents or falls from height, results in persistent functional impairments and lifelong disability in affected individuals [1, 2]. Owing to the complex pathophysiology of SCI, no effective therapeutic interventions currently exist for the management of this condition [3]. One of the major hurdles in SCI treatment is the formation of a hostile microenvironment that inhibits neural regeneration [4]. Following the primary mechanical insult, secondary injury cascades ensue, including edema, inflammatory activation, demyelination, and synapse loss [5–7]. These events result in secondary neuron death and the production of a large amount of cell debris in the microenvironment [8, 9]. Persistent deposition of cell debris, degenerated myelin, and synaptic remnants within the lesion site impedes neuron survival, axonal regeneration, and long-tern functional recovery after SCI [10–13]. Consequently, clearing cellular debris are essential steps toward creating a permissive environment for neural repair and functional recovery after SCI.

Recent advances in immunotherapy have opened new avenues for the treatment of neurological disorders [14]. In particular, regulatory T cells (Tregs) have gained recognition for their role in suppressing excessive immune responses and maintaining immune homeostasis in conditions such as multiple sclerosis and stroke [15, 16]. Tregs exert neuroprotective effects through the precise modulation of microglial activity, promoting reparative phenotypes and enhancing microglial phagocytosis to facilitate tissue regeneration [17–20]. As resident macrophages in the central nervous system (CNS), microglia are rapidly activated following SCI and play a pivotal role in debris clearance [21–24]. Increasing evidence suggests microglia-mediated synaptic elimination plays an essential role in maintaining tissue integrity and neural circuit function across various neurological conditions, including stroke, Parkinson’s disease (PD), and Alzheimer’s disease (AD) [25–29]. Impaired microglial phagocytosis of aberrant synapses may contribute to debris accumulation and impede neighboring neurons [30–33]. To date, the role of microglia in synaptic elimination following SCI remains undefined. Targeting microglia-dependent synapse engulfment could be crucial for mitigating synapse loss and may hold therapeutic potential for SCI. Previous studies have reported that Treg expansion via IL-10 administration suppresses microglial activation and improves functional outcomes in a mouse model of SCI [20]. However, whether or how Tregs influence microglial phagocytic activity, particularly in the context of synaptic engulfment following SCI, remains largely unknown.

Tregs exert their regulatory effects on microglia primarily through the release of immunomodulatory factors such as IL-4, IL-10, and transforming growth factor-β (TGF-β) [34, 35]. A recent report revealed that infiltrating Tregs in the ischemic brain regulate microglial phenotype switching by releasing osteopontin (OPN), which is a pleiotropic phosphorylated glycoprotein encoded by the Spp1 gene [36–38]. OPN has been shown to enhance microglial survival under lipopolysaccharide (LPS)-induced stress and to shift microglial polarization toward a neutral or anti-inflammatory phenotype [39]. Moreover, OPN promotes microglial synaptic engulfment in models of ischemic stroke and AD [40, 41]. Despite these findings, the role of Treg-derived OPN in modulating microglial function after SCI remains poorly understood.

In this study, we witnessed infiltrating Tregs in the injured spinal cords of both SCI patients and mouse models. Tregs deficiency compromises microglial synaptic phagocytosis and worsens neurological deficits after SCI. Transcriptomic analyses identified CD74 as a key regulator of microglial engulfment modulated by Tregs. Deletion of CD74 alleviated synaptic reduction caused by Tregs depletion and improved functional recovery. Furthermore, both in vitro and in vivo experiments demonstrated that OPN secreted by Tregs suppresses CD74 expression in microglia, thereby enhancing synaptic clearance. Treatment with OPN attenuated neuroinflammation, preserved synapses, and improved locomotor performance. Collectively, our findings elucidate novel Treg-microglia signaling axis involving OPN and CD74, offering mechanistic insight and potential therapeutic targets for SCI repair.

Materials and methods

Human samples

Paraffin-embedded spinal cord sections were obtained from the Department of Pathology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology. The cohort consisted of sections from three SCI patients and one control patient with lumbar disc herniation. Detailed clinical information for the patients is provided in Supplementary Table 1. The tissue samples were collected during surgical decompression procedures. The study was conducted in accordance with the Declaration of Helsinki and was approved by the Human Research Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology (Approval Number: TJ-IRB202506007). Written informed consent for the use of the spinal cord sections was obtained from all participants.

Animals

C57BL/6J, Foxp3-DTR, and Cd74^−/− mice were purchased from Shanghai Model Organisms Center, Inc. Both male and female mice aged 10–12 weeks were used for all in vivo experiments. All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of Tongji Medical College, Huazhong University of Science and Technology (Approval Number: TJH-202201003). Animals were housed in a controlled environment under a 12-hour light/dark cycle with free access to food and water. In all experiments, male and female mice were randomly assigned to each group. All researchers were blinded to experiment assignment, outcome assessment, and data analysis. Spinal cord contusion injury was induced as previously described [6]. Briefly, mice were anesthetized by inhalation of 2.5% isoflurane delivered via a precision vaporizer (RWD, China), and a laminectomy was performed to expose the thoracic spinal cord. The vertebral column was stabilized using bilateral steel clamps (RWD, China), and a contusion injury was delivered at the T10 level using a standardized impact force (5-gram weight from a height of 11 mm). A T10-level injury effectively models the paraplegic phenotype and, combined with the consistent anatomical structure of the thoracic spinal cord, facilitates the establishment of a reproducible standardized injury model [42–44]. Following surgery, the bladders of injured mice were manually voided twice daily until a reflex bladder was established. Additionally, the surgical wound and urethral orifice were regularly inspected and kept clean and dry with antiseptic care to prevent infection.

Immunohistochemistry

Human spinal cord sections were prepared as previously described [41]. Following antigen retrieval and blocking of nonspecific binding sites, sections were incubated with an anti-FOXP3 antibody (1:100, Abcam) for 1 h in a humidified chamber at room temperature. They were then washed with TBS-T and incubated with anti-IBA1 (1:300, Wako) or anti-CD4 (1:200, CST) for 1 h in a humidified chamber at room temperature, followed by incubation with species-appropriate fluorescent secondary antibodies (Alexa Fluor 488 for anti-Rabbit IBA1, corresponding to the green channel; Alexa Fluor 647 for anti-Mouse FOXP3, corresponding to the far-red channel; Jackson Laboratory) for 2 h.

For mouse spinal cord immunostaining, mice were anesthetized and transcardially perfused with cold saline followed by 4% paraformaldehyde. Coronal spinal cord Sect. (10 μm thick) were prepared using a freezing microtome (Thermo, America). Sections were blocked with 1% BSA in 0.3% Triton X-100 in PBS for 1 h and then incubated in a humidified chamber at 4 °C overnight with the following primary antibodies diluted 1:200: anti-FOXP3 (CST), anti-CD4 (1:200, CST), anti-IBA1 (Wako), Phalloidin (Invitrogen), anti-CD68 (Boster), anti-PSD95 (SYSY), anti-MBP (CST), PKH67 (Sigma), anti-C1q (Abcam), anti-VGLUT1 (SYSY), anti-CD74 (Bioss), anti-OPN (Abcam), anti-CD16 (Boster), and anti-HOMER1 (SYSY). After three 10-minute washes with PBS, sections were incubated with secondary antibodies (Jackson) for 1 h in a humidified chamber at room temperature. Specifically, the ROI was defined as the area extending 150–500 μm from the lesion core (Fig. S1C). For each staining, six sections per tissue sample were analyzed. Within the predefined ROI, images were captured using a confocal microscope (Olympus, Japan), and the number of positive cells was quantified. The final result for statistical analysis was expressed as (the number of positive cells/the total number of cells in the field) × 100%. Microglial morphology was assessed by quantitative Sholl analysis using the ImageJ software. For Luxol fast blue (LFB) staining, sections were immersed in 0.1% LFB solution (Sigma) at 60 °C for 20 min. Excess stain was removed by sequential immersion in 95% ethanol, distilled water, 0.05% lithium carbonate, and 70% ethanol until the white matter was clearly distinguished. Synaptic density in the injured area was examined by transmission electron microscopy (TEM, Zeiss, German), as previously described [12].

Treg depletion

To deplete Tregs, Foxp3-DTR mice received intraperitoneal injections of diphtheria toxin (DT, 40 µg/kg body weight; Listlabs) three days before surgery, followed by subsequent injections every three days to maintain depletion. In Cd74^−/− mice, Treg depletion was achieved by intraperitoneal injection of an anti-CD25 antibody (0.5 mg per mouse; BioXcell) seven days before SCI, with additional weekly injections until sacrifice.

Imaris 3D reconstruction

Three-dimensional reconstructions were generated using the Imaris software, as previously described [45]. Confocal Z-stack images were imported into Imaris, and the “Surface” module was used to reconstruct each fluorescent channel. A defined region of interest was selected, and absolute intensity values from each channel were applied for reconstruction. A uniform smoothing factor of 0.300 mm was used across all channels. A standardized intensity threshold was set to differentiate target signals from background, enabling the 3D rendering of the images.

Cell culture

Primary microglia were isolated from postnatal mice (postnatal days 1–3). After the removal of the meninges, the brains were dissected and dissociated. Cells were plated in T-flasks containing culture medium (DMEM supplemented with 10% FBS and 1% penicillin/streptomycin) and maintained at 37 °C in a humidified incubator with 5% CO₂ until confluence (12–14 days in vitro). Microglia were harvested by shaking the flasks at 180 rpm for 1 h.

For Cd74 silencing or overexpression, primary microglia were transfected with either negative control siRNA (NC), Cd74-targeting siRNA (Tsingke Biotech), or Cd74-targeting plasmid (Sangon Biotech) using Lipofectamine 3000 (Invitrogen). The siRNA sequence was GATGACCAACGCGACCTCATCTCTA. Cells were used for downstream experiments 24 h post-transfection. Transfection efficiency was confirmed by Western blotting.

Tregs were isolated from spleens using a mouse Treg isolation kit (Miltenyi Biotec) according to the manufacturer’s instructions. Isolated Tregs were stimulated with anti-CD3 (4 µg/mL; Thermo Fisher), anti-CD28 (5 µg/mL; Thermo Fisher), and IL-2 (100 ng/mL; BioLegend) for 3 days in Treg culture medium (RPMI 1640 with 10% FBS and 1% penicillin/streptomycin; Gibco). Stimulated Tregs were then co-cultured with primary microglia for 2 days in transwell inserts.

Oxygen–glucose deprivation/reperfusion (OGD/R)

Cells were washed twice with PBS and the medium was replaced with glucose-free DMEM. Cells were then exposed to ischemic conditions (1% O₂, 94% N₂, and 5% CO₂) at 37 °C for 3 h. For reperfusion, the medium was replaced with fresh microglia culture medium and the cells were incubated under normoxic conditions (95% air, 5% CO2) for an additional 3 h.

Phagocytosis assay

Myelin and synaptosomes were purified from wild-type C57BL/6J mouse brains, as previously described [41, 46]. Fluorescent labeling was performed using CFSE (MCE) following the manufacturer’s protocol. After OGD, primary microglia were incubated with CFSE-labeled microbeads (1 mg/mL; Invitrogen), myelin (5 µg/mL), or synaptosomes (5 µg/mL). Cells were then fixed with paraformaldehyde for immunofluorescence staining, and phagocytic activity was assessed by confocal microscopy.

Western blotting

Proteins were extracted from peri-injury spinal cord tissue using RIPA lysis buffer containing 1% PMSF (Roche), followed by sonication. Protein samples were separated by SDS-PAGE using a 12.5% Tris-glycine gel (Yeasen) and transferred to membranes. After blocking with 5% milk, membranes were incubated overnight at 4 °C with the following primary antibodies (1:1000): anti-C1q (Abcam), anti-C3 (Proteintech), anti-HOMER1 (SYSY), anti-PSD95 (SYSY), anti-SYN (SYSY), anti-OPN (Abcam), anti-CD74 (Bioss), anti-β-ACTIN (CST), anti-TUBULIN (Boster), and anti-GAPDH (Boster). HRP-conjugated secondary antibodies (1:1000; Jackson) were applied for 1 h at room temperature. Immunoreactive bands were visualized using an ECL reagent (Advansta), and protein expression levels were analyzed in a semi-quantitative manner.

qPCR

Total RNA from injured spinal cord tissue was extracted using the TRIzol reagent (Invitrogen). RNA from flow cytometry-sorted microglia was isolated using the RNeasy Mini Kit (QIAGEN), following the manufacturer’s instructions. cDNA was synthesized using the ReverTra Ace qPCR RT Kit (TOYOBO). Quantitative real-time PCR was performed using SYBR Green PCR Master Mix (Yeasen) on a BioRad CFX Connect system. Gene expression levels were normalized to Gapdh as the internal control.

Transcriptomic analysis

Microglia were isolated from spinal cord tissue as previously described [47]. Briefly, single-cell suspensions were incubated with an Fc receptor blocker (anti-CD16/32, 1:100; BD) for 15 min. Cells were then stained with anti-CD45-PerCP and anti-CD11b-APC (1:200; BD) for 30 min at 4 °C in the dark. CD45^intCD11b^high microglia were sorted using a FACSAria™ flow cytometer. For each sample, 200 cells were collected and total RNA was extracted using the RNeasy Micro Kit (QIAGEN, Germany). RNA sequencing was performed by the Beijing Genomics Institute (BGI) using the BGISEQ-500 platform. Pathway enrichment analyses, including Gene Set Enrichment Analysis (GSEA) and KEGG, were conducted using the Dr. Tom platform. Potential interactions between Tregs and microglia were predicted using the STRING database. Briefly, we extracted microglial differentially expressed genes (DEGs) enriched in the antigen processing and presentation pathway. Additionally, we extracted previously reported Treg-derived cytokines and immunomodulatory factors. Protein-protein interactions were calculated by using STRING database.

OPN treatment

OPN administration was performed as previously described [48]. Following cannula implantation into the left lateral ventricle, WT and Treg-depleted mice received intracerebroventricular injections of either recombinant OPN (3 µg per mouse) or PBS vehicle. The treatment regimen was initiated 1 h prior to SCI surgery and followed by weekly maintenance injections until sacrifice. An identical dosing schedule was applied across all groups to ensure comparability. For in vitro experiments, primary microglia were stimulated with OPN (10 µg/mL, R&D Systems) immediately following OGD.

Behavioral tests

Footprint analysis was performed 28 days after SCI according to established protocols [49]. The forelimbs and hindlimbs of mice were stained with blue and red ink, respectively, and stride lengths were measured. Basso Mouse Scale (BMS) scores were recorded at 1, 3, 7, 14, and 28 days post-injury, based on standard criteria [50]. General locomotor activity was assessed using the open field test. In each trial, mice were allowed to explore an open field chamber (50 × 50 × 40 cm³) for 5 min. A computer-assisted tracking system (VisuTrack) was used to continuously monitor and record animal behavior during the testing period.

Statistical analyses

Statistical analyses and figure generation were performed using GraphPad Prism 8. For in vivo experiments, five biological replicates were used, whereas in vitro experiments were performed in triplicate. Data are presented as mean ± SEM. Normality was assessed using the Kolmogorov-Smirnov test. Correlation and regression analyses were performed using either Pearson or Spearman correlation tests, as appropriate. For normally distributed data, comparisons between two groups were conducted using the student’s t-test, and comparisons among three or more groups were performed using one-way ANOVAs. For non-normally distributed data, the Mann–Whitney U test was applied. A P-value < 0.05 was considered statistically significant.

Results

Infiltrating Tregs regulate microglial activation in a distance-dependent manner after SCI

Previous studies have reported the infiltration of Tregs in stroke, traumatic brain injury, and other neurological disorders [51, 52]. To determine whether Tregs also infiltrate the spinal cord following SCI, we first performed double immunofluorescence staining for FOXP3 and CD4 on pathological sections from three SCI patients and one patient with lumbar disc herniation as an additional control (Supplementary Table 1). Representative images for CD4 and FOXP3 in each subject are shown in Fig. S1A. Treg cell infiltration into the injured tissue was clearly observed (Fig. 1A). Due to the sample size (n = 3 SCI, n = 1 control), only descriptive statistics were performed. Quantification of CD4⁺FOXP3⁺ Tregs demonstrated a clear increasing trend in infiltration within the SCI group compared to the control (Fig. S1B). Co-staining for FOXP3 and IBA1, combined with Sholl analysis, revealed that microglia in closer proximity to Tregs exhibited smaller soma areas and fewer branches, whereas microglia farther from Tregs displayed larger somas and more complex branching (Fig. 1B-C).

Fig. 1.

Infiltrating Tregs regulate microglial activation in a distance-dependent manner after SCI.A Representative images of CD4 and FOXP3 immunostaining in SCI patients and control. Scale bar = 50 μm. B Representative images of FOXP3 and IBA1 staining (scale bar = 20 μm) and Sholl analysis of Treg-distant/adjacent microglia (scale bar = 5 μm) around lesion in SCI patients. C Quantification of microglial soma area and count of microglial processes per cell in three SCI patients. Solid lines represent regression lines. P-values for Pearson’ s correlation analysis coefficient tests are shown. For correlation between the distance and microglial soma area, r = 0.3699; for correlation between the distance and microglial process number, r = 0.4059. D Experimental design for an animal experiment. n = 5 biological replicates per group. E Representative images of CD4 and FOXP3 staining (upper panel), FOXP3 and IBA1 staining (middle panel), and three-dimensional constructed images of IBA1⁺ microglia and FOXP3⁺ Treg (lower panel) in the peri-injury area following SCI in mice. Scale bar = 50 μm. For the 3D rendered view, scale bar = 20 μm. F-H Quantification of CD4⁺FOXP3⁺ cell percentage in the injured spinal cord (F), mean IBA1 intensity (G), and the colocalization coefficients between FOXP3⁺ cells and IBA1⁺ cells (H). n = 5 for each group, one-way ANOVA with post hoc Holm-Sidak test, *P < 0.05, **P < 0.01, and ***P < 0.001, ****P < 0.0001, data are presented as the mean ± SEM. I Quantification of the microglial soma area and the number of microglial processes per cell in mice post-SCI. Solid lines represent regression lines. P-values for Pearson’ s correlation analysis coefficient tests are shown. For correlation between the distance and microglial soma area, r = 0.8495; for correlation between the distance and microglial process number, r = 0.4813

To further explore this interaction between microglia and Tregs, we established a T10 contusive SCI model in wild-type (WT) mice (Fig. 1D). Our analysis focused on the peri-lesion area (specifically 150–500 μm from the lesion core, as outlined in Fig. S1C). Immunostaining revealed CD4⁺FOXP3⁺ Treg cell infiltration at multiple time points (3-, 7-, and 14-days post-injury), with peak infiltration observed on day 7 (Fig. 1E-F). Tregs were primarily localized to the peri-lesion area (within 500 μm of the lesion core, Fig. S1C). Furthermore, the mean fluorescence intensity of FOXP3 showed a decreasing gradient with increasing distance from the lesion core (Fig. S1D), supporting a distance-dependent distribution of infiltrating Tregs. Microglial activation also peaked on day 7 and then gradually declined (Fig. 1G). We next assessed the spatial relationship between Tregs and microglia at different time points post-injury. Microglia were frequently located in close proximity to Tregs beginning on day 3, with the proportion of Treg-associated microglia peaking on day 7 (Fig. 1H). Consistent with observations in human samples, microglia near Tregs at 28 days post-SCI in mice exhibited distinct morphological changes, including increased solidity and sphericity, along with reduced soma area and fewer branches (Fig. 1I). These findings suggest that Tregs regulate microglial activation in a distance-dependent manner following SCI.

Treg depletion impairs microglial clearance of synaptic debris and reduces synaptic density after SCI

To investigate the regulatory effect of Tregs on microglia, we established an in vitro Transwell co-culture system involving Tregs and microglia (Fig. 2A). Following OGD, fluorescently labeled microspheres, synaptosomes, or myelin were added to the microglial culture medium. After reoxygenation, microglia were stained with the cytoskeletal marker phalloidin, and their phagocytic activity was quantified by measuring the intracellular fluorescence intensity of the engulfed materials (Fig. 2B-C). Co-culture with Tregs had no effect on the microglial phagocytosis of microspheres. However, it significantly enhanced the proportion of microglia that engulfed synaptosomes, and also increased the fluorescence intensity of internalized myelin, indicating enhanced myelin phagocytosis. These results demonstrate that Tregs can modulate the phagocytic activity of primary microglia.

Fig. 2.

Treg depletion impairs microglial clearance of synaptic debris and reduces synaptic density after SCI.A Experimental design overview for the in vitro microglia-Treg coculture system, n = 5 biological replicates per group. B Representative images of microspheres, synaptosomes, and myelin engulfment in microglia following OGD/R. The white arrows indicate synaptosome⁺ microglia. Scale bar = 20 μm. For the enlarged view, scale bar = 5 μm. C Quantification of mean intensity of microspheres in microglia, synaptosome⁺ microglia and the mean intensity of myelin particles in microglia. n = 5 for each group, two-tailed Student’s t-test, **P < 0.01, and ***P < 0.001, data are presented as the mean ± SEM. D A schematic overview of the experimental design. DT was intraperitoneally administered to Foxp3-DTR mice for Treg depletion. n = 5 (E) Representative images of IBA1 and CD68 staining in mice at 7 days post-SCI. Scale bar = 20 μm. For the enlarged view, scale bar = 5 μm. F Three-dimensional rendered images of IBA1⁺ microglia and PSD95-labeled synaptic debris in the peri-injury area at 7 d and 28 d following SCI in mice. Scale bar = 5 μm. G Quantification of the colocalization coefficients between IBA1⁺ microglia and PSD95-labeled synaptic debris at 7- and 28- days following SCI in mice. n = 5 for each group, two-tailed Student’s t-test, ***P < 0.001, data are presented as the mean ± SEM. H-I Quantification of CD68⁺ microglia (H) and C1q⁺ microglia (I) in the injured spinal cord. n = 5 for each group, two-tailed Student’s t-test, ***P < 0.001, data are shown as the mean ± SEM. J Representative images of IBA1 and C1q staining in mice at 7 days post-SCI. Co-labelled cells are marked by white arrows. Scale bars: 20 μm (overview); 5 μm (enlarged view). K-L Representative images of western blotting results (K) and quantification of complement components and synaptic protein levels (L). n = 5 for each group, one-way ANOVA with post hoc Bonferroni test, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, data are presented as the mean ± SEM. M Representative immunofluorescence images of PSD95-VGLUT1 co-immunostaining around the lesion core at 7- and 28- days following SCI in mice. Colocalized puncta are marked by white circles. Scale bar = 5 μm. N Quantification of colocalized puncta in (M). n = 5 for each group, two-way repeated measures ANOVA with post hoc Bonferroni test, ***P < 0.001, ****P < 0.0001, data are presented as the mean ± SEM

To further assess this effect in vivo, we generated Treg-depleted mice by injecting DT into Foxp3-DTR transgenic mice (Fig. 2D). The proportion of phagocytic microglia was first assessed by CD68 immunostaining (Fig. 2E). Compared with wild-type controls, Treg-depleted mice showed no significant difference in the proportion of IBA1⁺ CD68⁺ phagocytic microglia (Fig. 2H). Synaptic debris, myelin debris, and cell debris within the peri-lesion area were labeled with PSD95, MBP, and PKH67, respectively. Three-dimensional reconstruction and quantitative analysis using Imaris software demonstrated that at 7 days post-injury, microglia in Treg-depleted mice showed significantly impaired phagocytosis of both synaptic debris (Fig. 2F–G) and myelin debris (Fig. S2A–B), while their ability to engulf cellular debris remained unaffected (Fig. S2A, C). Given that the impairment in synaptic clearance was more substantial than that in myelin phagocytosis, we focused our subsequent mechanistic investigation on how Tregs regulate microglia-mediated synaptic pruning. Notably, no significant difference in phagocytic activity was observed between groups at 28 days post-injury (Fig. 2G). These findings suggest that Treg depletion impairs microglial engulfment of synaptic debris during the subacute phase of SCI.

Under pathological conditions, synapse clearance by microglia is typically mediated by the classical complement cascade, with complement components C1q and C3 playing critical roles in tagging damaged synapses [53–55]. In the CNS, C1q is predominantly expressed by microglia [56]. To examine whether Tregs modulate complement-mediated synapse clearance, we quantified the number of C1q⁺ microglia at 7 days post-injury using immunostaining (Fig. 2I). Co-labeling for IBA1 and C1q revealed a significant reduction in C1q-positive microglia in Treg-depleted mice (Fig. 2J). Western blotting confirmed increased C1q expression in injured spinal cord tissue compared to sham-operated controls. However, in Treg-depleted mice, both C1q and C3 expression levels were markedly reduced compared to WT controls (Fig. 2K-L). These findings suggest that Treg depletion downregulates complement C3 levels and suppresses microglial C1q expression following SCI.

Prior studies have shown that dysregulated microglial synapse pruning contributes to synapse reduction and cognitive deficits in neurodegenerative diseases [32, 57]. Therefore, we examined whether Treg depletion affects synaptic density after SCI. Western blotting revealed that the expression of synaptic proteins (SYN, PSD95, and HOMER1) was significantly decreased at 7 days post-injury relative to sham controls, and was further reduced in the Treg-depleted group (Fig. 2K-L). Synapses were visualized as puncta co-labeled with the presynaptic marker VGLUT1 and the postsynaptic marker PSD95 (Fig. 2M). Quantification of co-localized puncta demonstrated a substantial reduction in synapse number at the lesion site in Treg-depleted mice compared to WT mice at both 7- and 28-days post-injury (Fig. 2N). Collectively, these results indicate that Treg depletion compromises microglial clearance of synaptic debris and reduced synapse density after SCI. This suggests a neuroprotective role for Tregs by regulating microglial-mediated clearance of synaptic debris.

Tregs modulate microglial phagocytosis by regulating CD74 expression

Having established a critical role for Tregs in regulating microglial function following SCI, we next employed transcriptomic sequencing to investigate the molecular mechanisms underlying this regulation. Microglia (CD11b⁺CD45^int) were isolated from injured tissue at 7 days post-injury (dpi) via fluorescence-activated cell sorting (Fig. 3A). Venn diagram was used to identify overlapping DEGs in microglia from Treg-depleted mice, WT mice, and sham-operated controls (Fig. 3B). KEGG pathway enrichment analysis revealed that microglia from Treg-depleted SCI mice exhibited a distinct transcriptomic profile, with DEGs significantly enriched in pathways related to phagosome formation, extracellular matrix (ECM)-receptor interaction, antigen processing and presentation, NF-κB signaling, and the complement and coagulation cascades (Fig. 3D). In addition, gene set enrichment analysis (GSEA) further confirmed enrichment in the antigen processing and presentation pathway in microglia from Treg-depleted mice compared to WT (Fig. 3E). A heatmap analysis identified Cd74 as one of the most significantly upregulated genes in microglia from the Treg-depleted group compared to WT (Fig. 3C). Notably, Cd74 was identified as a core gene enriched in the antigen processing and presentation pathway, exhibiting a fold change > 2 and a gene ranking score > 1.5.

Fig. 3.

Tregs modulate microglial phagocytosis by regulating CD74 expression.A Design for the RNA-seq study conducted in microglia at 7 days after SCI in mice, n = 3. B-C Venn diagram (B) and heat map (C) showing DEGs in microglia. D KEGG pathway enrichment analysis of the DEGs in Treg-depleted mice versus WT at 7 days post-SCI. E Gene set enrichment analyses (GSEA) of the microglial gene list in Treg-depleted mice compared to WT mice at 7 days post-SCI. F Identification of CD74 upregulation in microglia in Treg-depleted mice compared to WT. The white arrows indicate CD74⁺ microglia (top panel). Scale bar = 20 μm. For the enlarged view, scale bar = 5 μm. The percentage of CD74⁺ microglia (bottom left panel), and Cd74 mRNA expression in FACS-isolated microglia from injured spinal cord (bottom right panel) were quantified. n = 5 for each group, two-tailed Student’s t-test, ***P < 0.001, ****P < 0.0001, data are presented as the mean ± SEM. G Experimental design for siCd74 transfection in microglia in vitro. H Representative images of synaptosome and myelin engulfment in microglia following NC/siCd74 transfection and OGD/R. Scale bar = 20 μm. I Quantification of synaptosome⁺ microglia and mean intensity of myelin particles in microglia, respectively. n = 5 for each group, one-way ANOVA with post hoc Bonferroni test, **P < 0.01, ***P < 0.001, ****P < 0.0001, data are presented as the mean ± SEM

Cd74 is highly expressed in disease-associated microglia and plays a crucial role in inflammatory responses in neurological disorders [58]. To validate the transcriptomic data, we performed co-immunolabeling for IBA1 and CD74 and observed a significant increase in CD74⁺ microglia at the injury site in Treg-depleted mice relative to WT controls. Microglia were then isolated from injured tissue using flow cytometry, and Cd74 mRNA expression was assessed via qPCR. At 7 dpi, Cd74 expression was significantly elevated in microglia from Treg-depleted mice compared to WT controls (Fig. 3F). Previous studies have implicated CD74 in microglia-mediated synaptic plasticity during development [59]. However, its role in synaptic pruning following SCI remains unclear. To determine whether CD74 is involved in microglial synapse clearance, we assessed phagocytic capacity following CD74 knockdown or CD74 overexpression in vitro (Fig. 3G). Primary microglia were transfected with either siCd74 or plasmid and subsequently exposed to synaptosomes and myelin after OGD/R treatment. Western blot analysis confirmed efficient CD74 knockdown and overexpression at the protein level in primary microglial cultures (Fig. S3C-D). Our results demonstrated enhanced phagocytosis in siCd74-transfected microglia compared to controls, while overexpression of CD74 with a plasmid decreased phagocytotic capacity in primary microglia (Fig. 3H-I). These findings suggest that Tregs may regulate microglial synaptic engulfment in a CD74-dependent manner.

CD74-knockout improves microglial synaptic debris clearance and preserves synaptic density in Treg-depleted mice after SCI

To further investigate whether Tregs regulate microglial function through modulation of CD74 expression after SCI, we generated Cd74 knockout (Cd74^−/−) mice. Genotyping (Fig. S3E) and subsequent Western blot analysis (Fig. S3F–G) validated the successful generation and functional knockout of CD74 in our mouse model. Treg depletion was achieved via intraperitoneal injection of an anti-CD25 antibody dissolved in sterile PBS (Fig. 4A). We then assessed microglial synapse clearance using immunofluorescence and 3D reconstruction, with a particular focus on Treg-depleted mice (Fig. 4B). The phagocytic index of microglia was significantly increased in Treg-depleted Cd74^−/− mice compared to Treg-depleted WT mice. Importantly, we did not observe increased synaptic debris engulfment in Cd74^−/− mice relative to WT in the absence of Treg depletion (Fig. 4C). To evaluate whether CD74 is involved in Treg-mediated regulation of microglial C1q expression, we performed immunostaining for IBA1 and C1q (Fig. 4D). The proportion of C1q-positive microglia was elevated in Cd74^−/− mice compared to WT controls. This effect was even more pronounced in Treg-depleted Cd74^−/− mice, which exhibited a significantly higher percentage of C1q⁺ microglia than Treg-depleted WT mice (Fig. 4E). These data suggest that CD74 knockout restores microglial C1q expression and rescues the impaired synapse clearance process in microglia induced by Treg depletion.

Fig. 4.

CD74-knockout improves microglial synaptic debris clearance and preserves synaptic density in Treg-depleted mice after SCI.A A schematic diagram of the experimental design. Anti-CD25 was intraperitoneally administered to Cd74^−/− mice for Treg depletion. n = 5. B Three-dimensional constructed images of IBA1⁺ microglia and PSD95-labeled synaptic debris in the peri-injury area at 7 days following SCI in mice. Scale bar = 5 μm. C Quantification of the colocalization coefficients between IBA1⁺ microglia and PSD95-labeled synaptic debris at 7 days following SCI in mice subjected to the indicated treatments. n = 5 for each group, one-way ANOVA with post hoc Holm-Sidak test, **P < 0.01, data are presented as the mean ± SEM. D Representative images of IBA1 and C1q staining in mice at 7 days post-SCI. The white arrows indicate C1q⁺ microglia. Scale bars: 20 μm (overview); 5 μm (enlarged view). E Quantification of C1q⁺ microglia in the injured spinal cord with indicated treatments. n = 5 for each group, one-way ANOVA with post hoc Holm-Sidak test, *P < 0.05, data are presented as the mean ± SEM. F-I Representative Western blotting images and corresponding quantification of HOMER1, PSD95, SYN, and TUBULIN expression in mice at 28 days post-SCI. n = 5 for each group, one-way ANOVA with post hoc Holm-Sidak test, *P < 0.05, **P < 0.01, and ***P < 0.001, data are presented as the mean ± SEM. J Quantification of synapse density in the injured spinal cord with indicated treatments. n = 5 for each group, one-way ANOVA with post hoc Holm-Sidak test, **P < 0.01, and ****P < 0.0001, data are presented as the mean ± SEM. K Representative TEM images around the injury area at 28 days post-SCI. Synapses are marked by red arrows, scale bar = 1 μm (L) Representative images of the footprints of mice subjected to the indicated treatments at 28 days post-SCI. Double-headed arrows marked a representative step length, scale bar = 2 cm. M Quantification of stride length in (L). n = 5 for each group, one-way ANOVA with post hoc Holm-Sidak test, **P < 0.01, and ***P < 0.001, data are presented as the mean ± SEM

To determine the effect of CD74 deletion on synapse density following Treg depletion, we assessed the expression of synaptic proteins in the injured spinal cord at 28 dpi via Western blotting (Fig. 4F). CD74 knockout alone did not significantly affect synaptic protein levels compared to WT. However, in Treg-depleted Cd74^−/− mice, the expression of synaptic markers SYN and PSD95 was significantly higher than in Treg-depleted WT mice (Fig. 4G-I). Transmission electron microscopy further confirmed increased synaptic density at the lesion site in Treg-depleted Cd74^−/− mice compared to Treg-depleted WT controls (Fig. 4J-K). To assess functional outcomes, we performed footprint analysis to evaluate hindlimb motor function (Fig. 4L). Treg-depleted Cd74^−/− mice exhibited a significant improvement in stride length compared to their Treg-depleted WT counterparts (Fig. 4M). These findings demonstrate that CD74 knockout rescues microglial synaptic engulfment and preserves synaptic density in Treg-depleted mice, supporting a model in which Tregs regulate microglial phagocytosis of synaptic elements in a CD74-dependent manner.

OPN-CD74 axis mediates Treg-microglia interaction and enhances synaptic engulfment

Accumulating evidence suggests that Tregs modulate microglial phenotype and improve neurological outcomes via the secretion of immunomodulatory cytokines [34, 35]. To explore the factor through which Tregs exert regulatory effects on microglia, we used a STRING analysis to predict possible interactions between microglia and Tregs (Fig. 5A). Interactions between Cd74 and Spp1 through Cd44 or Itgb2 were strongly predicted. Spp1 encodes the pleiotropic cytokine OPN, which has been reported to play a crucial role in Treg cell-microglia interactions during stroke [36, 60]. We examined OPN expression in injured spinal cord tissue via Western blotting (Fig. 5B). OPN levels were significantly elevated after SCI compared to sham-operated controls. However, Treg depletion led to a marked reduction in OPN expression at 7 dpi, suggesting that Tregs are the primary source of OPN in this context (Fig. 5C). Immunofluorescence analysis corroborated these findings, showing reduced OPN fluorescence intensity in the Treg-depleted group compared to WT (Fig. 5D). Moreover, we confirmed the presence of OPN in infiltrating Tregs within the injured spinal cord (Fig. 5E). To further validate these results, we conducted in vitro experiments. Tregs subjected to OGD/R exhibited significantly elevated OPN expression, as confirmed by qPCR, ELISA, and immunofluorescence analyses, compared to control cells (Fig. 5F-G). Together, these results support a potential role for OPN in mediating Treg-microglia communication following SCI, potentially through the regulation of microglial CD74 expression and synaptic engulfment capacity.

Fig. 5.

OPN-CD74 axis mediates Treg-microglia interaction and enhances synaptic engulfment.A STRING analysis predicts Spp1 as a competitive ligand that mediates interactions between Tregs and microglia. B-C Representative images of Western blotting (B) and corresponding quantification (C) of OPN expression in mice. n = 5 for each group, one-way ANOVA with post hoc Bonferroni test, ****P < 0.0001, data are presented as the mean ± SEM. D Representative images of OPN staining in mice at 7 days following SCI, scale bar = 20 μm. E Representative immunofluorescence images of FOXP3 and OPN co-immunostaining in mice at 7 days following SCI, scale bar = 10 μm. F A schematic diagram of Tregs culture in vitro, n = 5 biological replicates per group. G Quantification of qPCR, ELISA, and immunostaining analyses of OPN expression in isolated Tregs in vitro. Scale bars: 20 μm (overview); 5 μm (enlarged view). n = 5 for each group, two-tailed Student’s t-test, ****P < 0.0001, data are presented as the mean ± SEM. H A schematic diagram of the experimental design for OPN treatment in vitro, n = 5 biological replicates per group. I-J Representative Western blotting images and corresponding quantification of CD74 expression in microglia in vitro. n = 5 for each group, one-way ANOVA with post hoc Bonferroni test, **P < 0.01, data are presented as the mean ± SEM. K-L Representative images (K) and quantifications (L) of CD74 and C1q intensity in microglia. Scale bars: 50 μm (top panel); 20 μm (bottom panel). n = 5 for each group, one-way ANOVA with post hoc Bonferroni test, ***P < 0.001, ****P < 0.0001, data are presented as the mean ± SEM. M A schematic diagram of siCd74 transfection and OPN treatment in vitro. N Representative images of synaptosome and myelin engulfment in primary microglia with indicated treatment. Scale bar = 20 μm. O Quantification of synaptosome⁺ microglia and the mean intensity of myelin particles in microglia, respectively. n = 5 for each group, one-way ANOVA with post hoc Holm-Sidak test, *P < 0.05, **P < 0.01, and ***P < 0.001, ****P < 0.0001, data are presented as the mean ± SEM

To investigate whether Tregs regulate microglial CD74 expression via the release of OPN, we stimulated microglia with recombinant OPN or co-cultured them with isolated Tregs (Fig. 5H). We observed that, similar to the Treg co-culture group, OPN treatment significantly decreased CD74 expression in microglia following OGD/R (Fig. 5I-L). Immunostaining further revealed that OPN treatment partially mimicked the effects of Treg co-culture, as indicated by a reduction in CD74 and C1q expression in microglia, albeit to a lesser extent (Fig. 5K-L). To further confirm that Tregs modulate microglial phagocytic function through the OPN-CD74 axis, we performed in vitro siCd74 or plasmid transfection experiments (Fig. 5M). Consistent with the effects seen in Treg co-cultured microglia, OPN-stimulated microglia demonstrated enhanced engulfment of synaptosomes and myelin compared to vehicle-treated controls. However, in OPN-treated microglia transfected with siCd74, there was no further increase in their phagocytic capacity toward synaptosomes or myelin. Furthermore, Cd74-targeting-plasmid transfection abrogated OPN-mediated augmentation of synaptosomes and myelin phagocytosis in primary microglia (Fig. 5N-O). These results collectively support the importance of the OPN-CD74 axis in Treg-microglia crosstalk and its role in enhancing microglial clearance of synaptosomes and myelin debris.

OPN maintains synaptic integrity and improves functional recovery after SCI

To further explore the role of OPN in Treg-mediated regulation of microglial activation, WT and Treg-depleted mice were treated with either recombinant OPN (3 µg per mouse) or PBS vehicle via intracerebroventricular injection. The treatment regimen was initiated 1 h prior to SCI surgery and followed by weekly maintenance injections until sacrifice. An identical dosing schedule was applied across all groups to ensure comparability (Fig. 6A). Immunostaining analysis revealed that OPN treatment significantly reduced the proportion of CD16⁺ neurotoxic microglia in the injured area compared to vehicle-treated controls in both WT and Treg-depleted mice, indicating that OPN suppresses proinflammatory microglial activation and ameliorates neuroinflammation (Fig. 6B-C). Furthermore, OPN administration decreased the number of CD74⁺ microglia at 7 days post-injury in both experimental groups (Fig. 6D–E). Furthermore, 3D reconstruction-based analysis of microglial phagocytosis revealed that OPN treatment robustly enhanced the engulfment of PSD95-labeled synaptic elements (Fig. 6F-G), indicating that OPN rescued the deficit in microglial synaptic clearance caused by Treg depletion. These findings collectively support the role of Treg-derived OPN as a key regulator of CD74 expression and phagocytic function in microglia after SCI.

Fig. 6.

OPN maintains synaptic integrity and improves functional recovery after SCI (A) Experimental design for intra-cerebroventricular OPN injection in WT and Treg-depleted mice, n = 5. B-C Representative images and quantification of CD16⁺ microglia in mice at 7 dpi. The white arrows indicate CD16⁺ microglia. Scale bars: 20 μm (overview); 5 μm (enlarged view). n = 5 for each group, one-way ANOVA with post hoc Holm-Sidak test, **P < 0.01, data are presented as the mean ± SEM. D-E Representative images and quantification of CD74⁺ microglia in mice at 7 dpi. The white arrows indicate CD74⁺ microglia. Scale bars: 20 μm (overview); 5 μm (enlarged view). n = 5 for each group, t one-way ANOVA with post hoc Holm-Sidak test, *P < 0.05, ****P < 0.0001, data are presented as the mean ± SEM. F Representative images and 3D reconstructed images of IBA1⁺ microglia and PSD95-labeled synaptic debris in the peri-injury area at 7 days following SCI in mice. The white arrows indicate PSD95⁺ microglia. Scale bars: 20 μm (overview); 5 μm (3D-rendered image). G Quantification of PSD95⁺ microglia in (F). n = 5 for each group, one-way ANOVA with post hoc Holm-Sidak test, *P < 0.05, ***P < 0.001, data are presented as the mean ± SEM. H Representative immunofluorescence images of PSD95-HOMER1 co-immunostaining around the lesion core at 7- and 28-days following SCI in mice. Colocalized puncta are marked by white circles. Scale bar = 10 μm. I Quantification of colocalized puncta in (H). n = 5 for each group, one-way ANOVA with post hoc Holm-Sidak test, ***P < 0.001, data are presented as the mean ± SEM. J Representative images of LFB staining at 28 days following SCI. Scale bar = 1 mm. K Quantification of demyelinated area in (J). n = 5 for each group, one-way ANOVA with post hoc Holm-Sidak test, **P < 0.01, data are presented as the mean ± SEM. L BMS-score of mice during a 28-day period post-SCI. n = 10 for each group, two-way repeated measures ANOVA with post hoc Holm-Sidak test, *P < 0.05, **P < 0.01 for WT + Vehicle vs. WT + OPN group; ^##P < 0.01 for Treg depletion + Vehicle vs. Treg depletion + OPN group; data are presented as the mean ± SEM. M Representative images of mouse traveling trajectories in the open field test. N Quantification of traveling distance, immobility time, and number of line crossings during open field test at 28 days following SCI in mice. n = 10 for each group, one-way ANOVA with post hoc Holm-Sidak test, *P < 0.05, **P < 0.01, ****P < 0.0001, data are presented as the mean ± SEM

We next investigated whether OPN-mediated enhancement of microglial synaptic engulfment influences synaptic integrity after SCI by quantifying synaptic density through dual staining for PSD95 and HOMER1 (Fig. 6H). A significant increase in PSD95-HOMER1 co-localized puncta at 28 days post-injury indicated that OPN treatment mitigates reductions in synaptic density in both WT and Treg-depleted mice (Fig. 6I). To evaluate the functional consequences of OPN treatment, we first examined demyelination at 28 days post-SCI using LFB staining (Fig. 6J). OPN administration significantly reduced the demyelinated area in both WT and Treg-depleted mice compared to controls (Fig. 6K). Locomotor function was assessed using the Basso Mouse Scale (BMS). Our results showed that OPN-treated WT mice exhibited improved motor performance beginning at 14 dpi, while OPN-treated Treg-depleted mice showed significant recovery by 28 dpi (Fig. 6L). Additionally, open-field tests at 28 dpi revealed that OPN-treated mice in both groups displayed increased travel distance, reduced immobility time, and more frequent line crossings compared to vehicle-treated controls (Fig. 6M–N), further supporting the beneficial role of OPN in functional recovery after SCI. Taken together, these results demonstrate that OPN administration counteracts the detrimental effects of Treg depletion by enhancing microglial synaptic engulfment, thereby maintains synaptic integrity and promotes neurological recovery following SCI.

Discussion

A growing body of evidence supports the immunosuppressive role of Tregs in modulating microglial inflammatory responses following SCI [20, 61, 62]. In addition to their immunomodulatory functions, microglia are also responsible for the phagocytic clearance of synaptic debris, a process critical for maintaining tissue integrity in various neurological disorders [29]. However, the mechanisms by which Tregs influence microglial synapse phagocytosis after SCI remain largely unexplored. In this study, we found that Tregs enhance the microglial engulfment of synapses through the OPN-CD74 axis, thereby maintaining synaptic integrity and promoting functional recovery following SCI.

Due to blood-brain barrier disruption, Treg infiltration has been observed in several neurological conditions, including stroke, traumatic brain injury, and multiple sclerosis [36, 63–65]. Consistent with these findings, we detected infiltrating Tregs at the injury site in human SCI patients. However, the extreme scarcity and limited availability of human SCI tissue restricted our cohort to only three eligible cases, representing a notable limitation of this study. To enhance the reliability of our observations, we introduced lumbar disc herniation patient samples as an additional control. Quantification of CD4⁺FOXP3⁺ Tregs revealed a consistent increasing trend in infiltration within the SCI group compared to control group, providing supportive biological evidence for Treg involvement in SCI. Given the sample size limitation, we focused on descriptive and correlative analyses to generate hypotheses for future investigation. Specifically, morphological analysis of microglia adjacent to Tregs across all three patients along with Pearson correlation analysis indicated that microglia located closer to Tregs tended to exhibit smaller soma areas and fewer branches, whereas those situated farther away displayed larger somas and more complex branching. These observations, while preliminary, suggest a spatial relationship worthy of further study. We believe these initial observations will help lay the groundwork for further investigation in larger cohorts.

To further explore the role of Tregs and their interaction with microglia following SCI, we established a T10 contusive SCI model in WT mice. Our data showed that both the number of infiltrating Tregs and the number of Treg-interacting microglia peaked at 7 dpi in mice. Concurrently, we found that the mean fluorescence intensity of the microglial marker IBA1 also reached its maximum at 7 dpi. This temporal pattern is in agreement with previous work using a spinal cord crush model, where the microglial population similarly peaked by day 7 [66]. These observations suggest that Treg-mediated immunoregulation of microglia is particularly prominent during the subacute phase of SCI. Notably, we also identified the presence of infiltrating Treg in SCI patients several years post-injury, implying a potential long-term role for Treg in the chronic progression of SCI. Further studies are warranted to elucidate the specific functions of Tregs during the chronic phase.

Our previous work demonstrated that Treg depletion in mice led to microglia with enlarged somata and increased branching complexity after SCI [6]. In the present study, we further observed that microglia in proximity to Tregs displayed significantly reduced soma area and fewer branches, both in human SCI patients and in mouse models. Given that microglial morphology is closely associated with functional states, including phagocytic activity [67], we hypothesized that Tregs might also regulate microglial phagocytosis in the context of SCI. Supporting this hypothesis, our in vitro experiments showed that microglia co-cultured with Tregs exhibited enhanced engulfment of both myelin and synaptosomes compared to controls. Additionally, in vivo Treg deficiency resulted in impaired synaptic clearance by microglia following SCI. These findings are consistent with recent studies reporting reduced microglial phagocytic capacity for neuronal debris in Treg-deficient mice after traumatic brain injury, further emphasizing the regulatory role of Tregs in microglial-mediated synaptic clearance [51]. Interestingly, we did not observe significant differences in the phagocytosis of fluorescent microspheres by Treg-co-cultured microglia in vitro. Furthermore, the proportion of CD68⁺ phagocytic microglia remained unchanged in Treg-depleted mice post-injury. These findings contrast with previous reports in stroke models, where Treg-conditioned medium enhanced microsphere uptake by BV2 microglia following OGD/R [60]. Such discrepancies may be attributable to differences in experimental conditions, including disease models, cell types used, and the duration of reoxygenation. Further investigations are needed to determine whether, and by what mechanisms, Tregs modulate microglial phagocytosis of cellular debris and myelin in the context of SCI.

Microglial synaptic pruning is primarily mediated by the classical complement cascade, particularly through components C1q and C3, which facilitate phagocytosis by binding to superfluous or damaged synaptic elements [68, 69]. In this study, we observed downregulation of C1q and C3 expression in Treg-depleted mice following SCI, suggesting that Treg depletion disrupts complement activation and consequently impairs synapse clearance. In addition, our data indicate that the deficit in microglial synapse engulfment resulting from Treg deletion is implicated in the observed synapse loss. This is demonstrated by a concomitant decrease in synaptic protein expression and a reduction in synaptic density within the lesion area. By contrast, previous studies have demonstrated that excessive microglial phagocytosis of synapses contributes to synaptic loss, impaired neural circuitry, and worsened neurological deficits in neurodegenerative diseases and ischemic stroke [70–73]. A recent review introduced a new framework distinguishing two modes of synapse elimination: culling, wherein microglia actively detach and phagocytose functional synapses, and scavenging, wherein microglia remove synaptic debris shed by neurons [30]. Inhibiting microglial culling has been shown to preserve neural network integrity [33], whereas blocking scavenging may be detrimental, as it prevents the clearance of synaptic debris [31]. Based on this conceptual framework, we hypothesize that microglial synapse elimination following SCI predominantly occurs through scavenging. Impaired microglial scavenging may result in the accumulation of damaged synaptic material in the injury microenvironment, thereby hindering tissue regeneration and worsening the long-term outcome of SCI. Further investigation is needed to distinguish between culling and scavenging in the progression of SCI.

Transcriptomic analysis revealed that microglia in Treg-depleted mice underwent functional reprogramming, with DEGs enriched in antigen processing and presentation pathways. Among these, Cd74 emerged as the most upregulated gene in microglia following Treg depletion and was identified as a central component of the antigen presentation pathway. We propose that Cd74 may represent a potential target for Treg-mediated regulation of microglial function. Cd74 is known to be highly expressed in disease-associated microglia under various pathological conditions, including stroke, amyotrophic lateral sclerosis, AD, MS, and aging [74–82]. As a cell surface receptor associated with MHC II, CD74 plays a critical role in signal transduction and the initiation of inflammatory responses [83]. In our study, microglial CD74 expression was upregulated in SCI mice compared to sham-operated controls, consistent with previous reports. Treg depletion further increased microglial CD74 levels post-SCI, suggesting enhanced microglial activation and neuroinflammatory responses. Previous studies have shown that reduced CD74 expression leads to synaptic deficits during development due to aberrant complement activation [84]. Moreover, CD74 knockdown has been shown to increase microglial synaptic pruning activity in microglia-neuron coculture models in vitro [84]. Consistently, our in vitro experiments demonstrated that silencing CD74 in primary microglia restored their phagocytic capacity for synaptosomes following OGD/R, while overexpression of CD74 further impaired microglial synaptic clearance. These findings highlight the role of CD74 in regulating microglial synaptic engulfment. Based on this evidence, we hypothesize that Tregs modulate microglial synapse clearance in a CD74-dependent manner.

To further evaluate the role of CD74, we investigated synapse engulfment following SCI in Cd74^−/− mice. Our results showed that CD74 knockout reversed the downregulation of microglial C1q expression induced by Treg depletion and restored microglial phagocytic activity toward synaptic debris after SCI. Previous studies have reported that CD74 ablation attenuates phosphorylation of NF-κB p65, promoting M2 microglial polarization and enhancing remyelination in MS and traumatic brain injury models [85–88]. Moreover, C1q expression has been shown to be regulated by NF-κB activation [89, 90]. We propose that Tregs modulate CD74-mediated nuclear translocation of NF-κB subunits, thereby upregulating C1q expression in microglia. Further research is needed to clarify the role of NF-κB in CD74-dependent microglial synaptic engulfment following SCI.

Additionally, we observed no significant differences in microglial synaptic phagocytosis or synapse density between Cd74^−/− and WT mice. This discrepancy may stem from the use of a global CD74 knockout model, which limits interpretation of microglia-specific functions and constitutes a key limitation of our study. Although CD74 is predominantly expressed in microglia within the central nervous system, it is also expressed by peripheral immune cells [91]. CD74 serves as a coordinator in antigen-presenting cells and modulates immune responses through interactions with other immune populations [92]. For instance, CD74⁺ tumor-associated macrophages can promote CD8⁺ T cell responses and correlate with improved prognosis [93], and CD74⁺ B cells have been shown to activate transcription factors in naïve T cells, amplifying immune reactivity in the tumor microenvironment [94]. While our in vitro Cd74 knockdown and overexpression experiments in primary microglia support a role for the OPN-CD74 axis in Treg-microglia communication, systemic CD74 deletion may affect neuroprotective immune responses after SCI via peripheral mechanisms. Furthermore, our in vitro results also show that CD74 inhibition also enhances the phagocytosis of myelin debris in primary microglia, suggesting an extended regulatory role for CD74 in microglial phagocytosis. Future studies utilizing microglia-specific CD74 conditional knockout models will be essential to precisely delineate the cell-autonomous functions of CD74 in microglial phenotypic and functional regulation.

STRING network analysis identified OPN as a promising candidate for Treg-mediated regulation of microglial function. OPN is an early marker of inflammation and tissue damage [95]. Accumulating evidence supports that Treg-derived OPN confers neuroprotection in murine models of stroke [36, 60]. In our study, we detected OPN expression in infiltrating Tregs within the injured spinal cord, and Treg depletion significantly reduced OPN levels post-SCI. Moreover, exogenous OPN treatment enhanced microglial phagocytosis to a degree comparable to that observed in Treg-microglia cocultures. Importantly, the phagocytic capacity of microglia toward synaptosomes and myelin was not further increased by OPN treatment following CD74 knockdown, suggesting that OPN promotes microglial synaptic engulfment in a CD74-dependent manner. CD74 often forms a complex with the transmembrane glycoprotein CD44, which is involved in cellular processes such as apoptosis, adhesion, migration, and immune signaling [88, 96]. Notably, the interaction between OPN and CD44 in immune cells has been well documented [97]. Based on these findings, we hypothesize that OPN regulates microglial CD74 expression through binding to CD44, thereby modulating the complement cascade and facilitating microglial synapse clearance. Further investigations are required to elucidate the specific mechanisms by which OPN controls microglial CD74 expression.

In addition to Tregs, microglia can also produce OPN during central nervous system development and disease [37]. A recent study reported that microglial OPN expression was reduced in Treg-depleted mice compared to WT following stroke, suggesting that Treg-derived OPN may act as a trigger for microglial OPN production [36]. Increased OPN levels in the injured microenvironment have been shown to support white matter repair and improve functional recovery after ischemic injury [60]. In line with these findings, we demonstrated that OPN treatment suppressed neurotoxic microglial activation, enhanced microglial synaptic engulfment, and promoted motor function recovery post-SCI. These results suggest that OPN may serve as a potential therapeutic agent for SCI.

Conclusions

In summary, our findings reveal that infiltrating Tregs regulate microglial synapse phagocytosis through the OPN-CD74 axis following SCI. Treg-derived OPN downregulates microglial CD74 expression, thereby enhancing synaptic engulfment and maintaining synaptic integrity, ultimately contributing to improved functional outcomes. Targeting the OPN-CD74 pathway may represent a promising and druggable strategy for the treatment of SCI.

Abbreviations

AD

Alzheimer’s disease

BMS

Basso Mouse Scale

CFSE

Carboxyfluorescein succinimidyl amino ester

dpi

Days post-injury

DEGs

Differential expressed genes

ECM

Extracellular matrix

ELISA

Enzyme-linked immunosorbent assay

FACS

Fluorescence-activated cell sorting

FBS

Fetal bovine serum

GSEA

Gene set enrichment analysis

KO

Knockout

LFB

Luxol fast blue

LPS

Lipopolysaccharide

OGD/R

Oxygen–glucose deprivation/reperfusion

PBS

Phosphate-buffered saline

OPN

Osteopontin

PD

Parkinson’s disease

qPCR

Quantitative polymerase chain reaction

ROI

Region of interest

SCI

Spinal cord injury

TEM

Transmission electron microscope

Tregs

Regulatory T cells

WT

Wild-type

Authors’ contributions

X.L., MH.W., and ZY.Y. designed the study. R.L and H.Y. performed animal experiments. XT.L., and ZY.W. performed *in vitro* experiments. H.Y., Y.X., Y.L., and ZY.W. collected data and/or performed analyses. R.L., Y.X., and Y.L. wrote the manuscript. R.L., Y.X., H.H., ZY.Y., WS.Q., MH.W., and X.L. reviewed the manuscript and provided critical discussion.

Funding

This study was supported by the National Nature Science Foundation of China (82171385 to X. Luo), Integrated Chinese and Western Medicine Project for Chronic Disease Management (CXZH2024014 to X. Luo), The High-Quality Clinical Research Fund of Tongji Hospital (2024TJCR013 to X. Luo)), Interdisciplinarity research program of Huazhong University of Science and Technology (2023JCYJ030 to X. Luo), Medical innovation and transformation incubation project of Tongji Hospital (2022CXZH010 X. Luo).

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The clinical samples included in this study were approved by the Human Research Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology (Approval Number: TJ-IRB202506007). The animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of Tongji Medical College, Huazhong University of Science and Technology (Approval Number: TJH-202201003).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.