Human bone marrow-derived mesenchymal stem cells alleviate fluoxetine-resistant depression-like behaviors in mice by multi-target effects in the hippocampus

Abstract

BACKGROUND

Treatment-resistant depression (TRD) remains a major clinical challenge due to the limited efficacy of conventional first-line antidepressants. Mesenchymal stem cells (MSCs), which possess potent immunomodulatory and neurotrophic properties, have emerged as promising therapeutic candidates for neuropsychiatric disorders. Human bone marrow-derived MSCs (hBMSCs) are one of the most widely used sources. However, their therapeutic potential and their underlying mechanisms in the context of TRD have yet to be fully elucidated.

AIM

To investigate the antidepressant effects of hBMSCs in a TRD-like mouse model and to explore the potential molecular mechanisms involved.

METHODS

A mouse model of fluoxetine-resistant depression (FRD), defined as an early stage of TRD, was established by combining chronic unpredictable restraint stress with corticosterone administration and validated by the lack of behavioral response to fluoxetine (FLX). hBMSCs (2 × 10⁶ cells) were administered via the tail vein twice weekly for 4 weeks. A battery of behavioral tests was conducted to evaluate depressive- and anxiety-like behaviors. Immunofluorescence staining and western blotting were performed to assess microglial activation, astrocytic integrity, neuronal survival, and hippocampal neurogenesis.

RESULTS

hBMSC treatment significantly ameliorated depressive- and anxiety-like behaviors in FRD model mice, whereas FLX failed to produce comparable behavioral improvements. hBMSCs markedly attenuated microglial activation and reduced neuroinflammatory responses in the hippocampus. In addition, hBMSCs restored astrocytic structure, protected neuronal integrity, and promoted adult hippocampal neurogenesis. These neuroprotective effects were associated with the brain-derived neurotrophic factor/tropomyosin receptor kinase B/cAMP response element-binding signaling pathway. hBMSCs labeled with 1,1’-dioctadecyl-3,3,3’,3’-tetramethylindocarbocyanine perchlorate were not detected in brain tissue, suggesting that their therapeutic effects may primarily rely on indirect or paracrine mechanisms.

CONCLUSION

hBMSCs exert robust antidepressant-like effects in an FRD mouse model by reducing neuroinflammation, restoring glial function, enhancing neurogenesis, and activating neurotrophic signaling, supporting hBMSCs as a promising therapy for FRD.

Key Words: Major depressive disorder; Treatment-resistant depression; Chronic unpredictable restraint stress; Corticosterone; Fluoxetine; Human bone marrow-derived mesenchymal stem cells; Neuroinflammation; Neurogenesis; Brain-derived neurotrophic factor/tropomyosin receptor kinase B/cAMP response element-binding signaling

Core Tip: We established a mouse model of fluoxetine-resistant depression (FRD) by combining chronic unpredictable restraint stress with corticosterone administration. Systemic administration of human bone marrow-derived mesenchymal stem cells (hBMSCs) exerted robust antidepressant-like effects through multiple mechanisms, including suppression of hippocampal neuroinflammation, restoration of astrocytic integrity, protection of neuronal survival, and promotion of adult neurogenesis. These effects were associated with reactivation of the brain-derived neurotrophic factor/tropomyosin receptor kinase B/cAMP response element-binding signaling pathway. hBMSCs did not engraft in the brain, suggesting paracrine-mediated actions. Our findings support the use of hBMSCs as a promising multi-target therapeutic strategy for FRD, an early stage of treatment-resistant depression.

INTRODUCTION

Major depressive disorder (MDD) is a heterogeneous mental disorder that is characterized by persistent low mood, loss of interest, and negative cognition. Recent epidemiological studies estimate that ~320 million people have MDD worldwide[1]. In recent years, the incidence of MDD among adolescents has increased annually. Approximately 20% of adolescents globally are diagnosed with MDD[2]. It is a leading cause of self-harm and suicide. It is also the second leading cause of death in young people and a global public health concern[2,3]. About half of patients do not respond adequately to first-line antidepressants developed based on the monoamine hypothesis, such as selective serotonin reuptake inhibitors. Most of those who initially respond are prone to relapse[4]. About one-third of MDD patients meet the diagnostic criteria for treatment-resistant depression (TRD)[5]. These limitations underscore the urgent need to develop novel, more effective therapeutic interventions for MDD, particularly for TRD.

The neuroinflammation hypothesis posits that chronic, low-grade systemic inflammation can trigger neuroinflammatory responses. These responses, in turn, contribute to the pathogenesis of MDD[6]. Accumulating preclinical evidence shows that hyperactivation of microglia, resident macrophages in the brain, induces neuroinflammation by releasing proinflammatory cytokines [e.g., interleukin (IL)-1β, IL-6, and tumor necrosis factor-α (TNF-α)][7-9]. This process compromises neuronal function and synaptic plasticity. For example, mice with high IL-6 expression were susceptible to subthreshold repeated social defeat stress, while IL-6 knockout mice were resilient[10]. Minocycline, a microglial inhibitor, significantly attenuated chronic unpredictable mild stress (CUMS)-induced depression-like behavior. It achieved this by shifting microglia from the M1 to the M2 phenotype[11]. Similarly, clinical studies have consistently shown that peripheral blood inflammatory factors (e.g., IL-6, TNF-α, and C-reactive protein) are elevated in patients with MDD[12]. Anti-inflammatory drugs (e.g., nonsteroidal anti-inflammatory drugs and cytokine inhibitors) may exert beneficial antidepressant effects[13]. However, findings remain inconsistent across studies and may only explain the pathophysiology in a subset of patients.

Another prominent theory - the neurogenesis hypothesis - posits that impaired neurogenesis in the subgranular zone of the hippocampal dentate gyrus (DG) is a key pathological basis for onset of MDD[14]. Many gain-of-function and loss-of-function studies have provided robust evidence for this hypothesis. For instance, various antidepressant treatments, including pharmacological therapies and behavioral interventions, have been shown to significantly stimulate adult hippocampal neurogenesis[15,16]. Conversely, when neurogenesis is blocked by genetic approaches or chemogenetic silencing, the antidepressant effects of these interventions are abolished[17], establishing the essential role of neurogenesis in antidepressant efficacy. Brain-derived neurotrophic factor (BDNF), a key regulator that promotes adult hippocampal neurogenesis, is significantly downregulated in the hippocampus of depressed patients[18] and animal models of depression[19], while exogenous BDNF can directly produce antidepressant-like effects[20]. BDNF primarily exerts its effects by binding to its high-affinity receptor, tropomyosin receptor kinase B (TrkB), on the surface of neural stem/progenitor cells, activating downstream signaling cascades, including the mitogen-activated protein kinase/extracellular signal-regulated kinase and phosphoinosytol-3 phosphate/protein kinase B pathways[21]. This subsequently regulates gene expression, promoting the transition of cells from a quiescent state into the cell cycle and initiating their division and proliferation. Therefore, the BDNF/TrkB signaling pathway is necessary for maintaining basal hippocampal neurogenesis and serves as the molecular foundation for the therapeutic efficacy of antidepressants.

Mesenchymal stem cells (MSCs) have emerged as a promising therapeutic candidate for MDD due to their potent immunomodulatory and tissue-repair capabilities[22]. A substantial body of preclinical research has elucidated the complex mechanisms through which MSCs exert antidepressant effects via multitarget pathways. First, MSCs provide neurotrophic support by secreting neurotrophic factors, including BDNF, glial cell line-derived neurotrophic factor, and nerve growth factor, promoting neuroregeneration, enhancing neuronal survival, and facilitating functional recovery[23]. Second, MSCs exert immunomodulatory effects by inhibiting the release of proinflammatory cytokines (e.g., TNF-α and IL-1β) while upregulating anti-inflammatory cytokines [e.g., IL-10 and transforming growth factor (TGF)-β], thereby attenuating central neuroinflammation and improving neuronal function[24]. Third, MSCs indirectly modulate neurotransmitter systems by influencing the metabolism and release of monoamines such as dopamine and serotonin, alleviating depressive-like behavior[25]. For instance, Gallagher et al[26] demonstrated that intravenous administration of human MSCs suppressed stress-induced peripheral inflammation, which, in turn, indirectly reduced central neuroinflammation and alleviated depressive-like behavior in mice. Huang et al[27] demonstrated in a mouse model of depression that intravenously administered MSCs activate BDNF/TrkB signaling in lung tissue, which transmits signals along the pulmonary vagus neural pathway, ultimately modulating the function of serotonergic neurons in the raphe nucleus to produce anxiolytic and antidepressant-like effects. However, findings across preclinical studies remain mixed due to variations in MSC sources (e.g., adipose tissue, bone marrow, umbilical cord, and dental tissue), differences in depression models, and heterogeneity in dose and administration routes[28].

In this study, we established and validated a novel fluoxetine-resistant depression (FRD) mouse model using an 8-week chronic unpredictable restraint stress (CURS) procedure combined with 4 weeks of intraperitoneal corticosterone (CORT) injections, and confirmed the treatment-resistant phenotype by nonresponse to fluoxetine (FLX). Utilizing this model, we investigated the therapeutic effects of intravenously administered human bone marrow-derived MSCs (hBMSCs) by evaluating anxiety- and depression-like behaviors, hippocampal neuroinflammation, neurogenesis, and the BDNF/TrkB/cAMP response element-binding (CREB) signaling pathway. Our findings establish hBMSCs as a promising multi-target therapeutic strategy for FRD, potentially addressing the critical limitations of current first-line antidepressants.

MATERIALS AND METHODS

Cell culture

hBMSCs were obtained from Zhong Qiao Xin Zhou Biotechnology Co. Ltd. (PRI-H-00137, Shanghai, China). Cells were cultured in a specific basal medium (ZQ-1320, Zhong Qiao Xin Zhou, Shanghai, China) supplemented with 10% (v/v) fetal bovine serum and 1% (v/v) penicillin-streptomycin. Cultures were maintained at 37 °C in a humidified incubator with 95% humidity and 5% (v/v) CO₂. For all experiments, cells were used when they reached 80%-90% confluency.

Experimental animals

Experimental animals were purchased from Jicui Yaokang Biotechnology Co. Ltd. (Chengdu, Sichuan Province, China). Male C57BL/6J specific pathogen-free-grade mice, aged 8-10 weeks and weighing 20-24 g, were used. The animals were housed in a controlled environment at 22-25 °C, 50%-60% relative humidity, and a 12-hour light/dark cycle. The experimental design and all animal procedures were approved by the Zigong Mental Health Center Animal Ethics Committee (Approval No. ZPRC-20250223). All experiments were conducted in compliance with the Guide for the Care and Use of Laboratory Animals (8^th ed, NRC 2011). The study was reported in accordance with the ARRIVE guidelines 2.0 (Animal Research: Reporting of In Vivo Experiments).

Characterization of hBMSCs

hBMSCs were first characterized by assessing their differentiation potential. Cells were cultured in specific media to induce osteogenic and adipogenic differentiation. For osteogenic induction, upon reaching 80% confluence, cells were cultured in medium containing dexamethasone (100 nM), β-glycerophosphate (10 mmol/L), and ascorbic acid (50 μM) for 21 days, with medium changes every 2 days. For adipogenic induction, after reaching full confluence, cells were subjected to cycles of Medium A for 3 days followed by Medium B for 1 day over 21 days. Medium A contained insulin (10 μg/mL), dexamethasone (1 μM), 3-isobutyl-1-methylxanthine (0.5 mmol/L), and indomethacin (200 μM), and Medium B contained insulin (10 μg/mL) only. Following differentiation, osteogenesis and adipogenesis were confirmed by Alizarin Red S and Oil Red O staining, respectively. The immunophenotype of the hBMSCs was further characterized by flow cytometry. Cells were stained with fluorochrome-conjugated antibodies against the positive markers CD73, CD90, and CD105, and the negative marker CD34. Appropriate isotype-matched controls were used to define the gating parameters.

In vivo distribution of hBMSCs

To observe the distribution of hBMSCs in vivo, the cells were labeled with 1,1’-dioctadecyl-3,3,3’,3’-tetramethylindocarbocyanine perchlorate (Dil, MKL-D775877, Macklin, Shanghai, China). Cells were incubated with a Dil working solution (5 μmol/L) for 30 minutes, followed by three washes with phosphate-buffered saline (PBS) to completely remove unbound dye. Following trypsin digestion and resuspension, 2 × 10⁶ cells were injected via the tail vein. At 24 hours post-injection, the mice were killed, and tissues were harvested. The tissues were processed into cryosections, and the localization of Dil-labeled cells within the various tissues was observed under a fluorescence microscope.

Establishment of FRD mouse model and treatment protocol

The FRD model was established using an adapted CURS[29] protocol over an 8-week period (Supplementary Table 1), with intraperitoneal administration of CORT (40 mg/kg; 515287, J&K Scientific, Shanghai, China) during the final 4 weeks. FLX was obtained from Aladdin Biotechnology Co. Ltd. (Shanghai, China). A total of 48 mice were randomly divided into four groups of 12 based on a computer-generated random number: Control, stress, stress + FLX, and stress + hBMSCs. Mice in the stress + hBMSCs group received tail vein injections of 2 × 10⁶ hBMSCs twice weekly for 4 weeks[27,30]. Mice in the stress + FLX group received daily oral gavage of FLX (10 mg/kg) for 4 weeks. Mice in the control and stress groups received daily oral gavage administration of an equal volume of sterile saline. No animal was excluded from the analysis. The data analyst was blinded to group assignment.

Behavioral tests

All behavioral tests were conducted according to established protocols. Mice were acclimated to the behavioral testing room for 1 hour prior to each test. The apparatus was thoroughly cleaned with 75% ethanol after each trial to eliminate olfactory cues.

Open field test: The apparatus consisted of an open-top, square arena (40 cm × 40 cm × 40 cm) with the floor divided into a central and peripheral zone. At the start of the trial, each mouse was placed in the center of the arena and allowed to explore freely for 10 minutes. The number of entries into the central zone and the total time spent in the central zone were recorded.

Elevated plus maze test: The maze comprised two open arms, two closed arms, and a central connecting platform. Each mouse was placed on the central platform facing an open arm, and the number of entries into the open arms and the time spent in the open arms were recorded over a 5-minute period.

Forced swim test: A transparent cylindrical container (20 cm diameter, 30 cm height) was filled with fresh water (23-25 °C) to a depth of 15 cm. Each mouse was placed into the water for a 6-minute test, and the total immobility time during the last 4 minutes was recorded.

Tail suspension test: The distal 2 cm of each mouse’s tail was fixed to a horizontal bar with adhesive tape, suspending the animal head-down. The test lasted 6 minutes, and the immobility time during the final 4 minutes was recorded.

Sucrose preference test: Mice were first habituated to the testing conditions during a 48-hour adaptation period, during which they had free access to 1% sucrose solution and plain drinking water. The positions of the two bottles were switched halfway through the adaptation period to prevent place preference. For the formal test, mice were individually housed and, following 24 hours of water deprivation, were presented with two pre-weighed bottles: One containing 1% sucrose solution and the other containing plain water. The test lasted 24 hours, with the bottle positions switched halfway through the test period. At the end of the test, the bottles were reweighed to measure consumption. The sucrose preference percentage was calculated as: [Sucrose consumption/(sucrose consumption + water consumption)] × 100%.

Hematoxylin and eosin staining

Fixed tissues from the heart, liver, spleen, lung, and kidney were paraffin-embedded and sectioned at 5 μm thickness. The sections were deparaffinized in xylene and rehydrated through a graded ethanol series. They were stained with hematoxylin for ~5 minutes, rinsed under running water, differentiated in acid alcohol, and stained blue in 0.1% ammonia water. The sections were dehydrated sequentially in 85% and 95% ethanol, counterstained with eosin for ~5 minutes, dehydrated in absolute ethanol, cleared in xylene, and mounted with neutral resin. Stained sections were examined, and images were captured using an inverted microscope.

Immunofluorescence staining

After completion of the behavioral tests, mice were deeply anesthetized via intraperitoneal injection of 20% urethane solution (1.0 g/kg). Once the absence of the pedal withdrawal reflex was confirmed, animals were killed by cervical dislocation. Transcardial perfusion was performed, followed by immediate extraction of the whole brain. The brain tissue was fixed in 4% paraformaldehyde for 48 hours, dehydrated through a graded series of sucrose solutions (10%, 20%, and 30%, each for 24 hours), and sectioned coronally at 20 μm thickness using a cryostat. Tissue sections were permeabilized with 0.5% Triton X-100 for 20 minutes and blocked with 5% bovine serum albumin at room temperature for 40 minutes. Sections were incubated overnight at 4 °C with primary antibodies (Supplementary Table 2), washed with PBS, and incubated with corresponding species-specific fluorescent secondary antibodies for 1.5 hours at room temperature in the dark. After further washes with PBS, nuclei were counterstained with 4’,6-diamidino-2-phenylindole for 15 minutes. All samples were mounted with an antifade mounting medium. Images were acquired using a Nikon C2 confocal microscope system, and fluorescence signals were quantified using Fiji software (v2.3.0).

Western blotting

Tissue samples were homogenized in RIPA lysis buffer supplemented with protease and phosphatase inhibitors. Protein concentrations were determined using the BCA method. Equal amounts of protein (20 μg per lane) were separated by sodium-dodecyl sulfate gel electrophoresis and transferred onto polyvinylidene difluoride membranes using a wet transfer system. The membranes were blocked with 5% skimmed milk in Tris-buffered saline/Tween 20 (TBST) at room temperature for 1.5 hours, followed by incubation with primary antibodies (Supplementary Table 2) overnight at 4 °C. After washing with TBST, membranes were incubated with horseradish-peroxidase-conjugated secondary antibodies (diluted 1:5000) for 2 hours at room temperature. Following further washes with TBST, protein bands were visualized using an ECL chemiluminescent substrate, and signals were captured using a Vilber FUSION Fx6 imaging system (Paris, French). Densitometric analysis was performed using ImageJ (v1.53k), with GAPDH serving as the loading control for normalization.

Statistical analysis

Statistical analyses were performed using SPSS software (version 20.0). Data are presented as mean ± SEM. Graphs were generated using GraphPad Prism (version 8.02). Normality of the data distribution was assessed using the Kolmogorov-Smirnov test, and homogeneity of variance was evaluated using Levene’s test. For comparisons among multiple groups, one-way analysis of variance was applied, followed by Tukey’s post hoc test for pairwise comparisons. Comparisons between two groups were conducted using a two-tailed unpaired Student’s t test, with Welch’s correction applied when variances were unequal. When applicable, Bonferroni correction was used to adjust for multiple comparisons to control the type I error rate. Statistical significance was defined as ^aP < 0.05, ^bP < 0.01, and ^cP < 0.001; NS indicates no statistically significant difference (P ≥ 0.05).

RESULTS

Characterization and in vivo distribution of hBMSCs

The hBMSCs displayed typical MSC morphology. Under an inverted microscope, cells adhered to plastic and exhibited a uniform, spindle-shaped or fibroblast-like appearance with distinct nuclei and abundant cytoplasm. Following passaging, cells proliferated actively and formed whirlpool-like or radial colonies, consistent with classical MSCs morphology (Figure 1A).

Figure 1 Characterization of human bone-marrow-derived mesenchymal stem cells. A: Representative bright-field image showing the morphology of human bone-marrow-derived mesenchymal stem cells (hBMSCs); B: Alizarin Red S staining; C: Oil Red O staining; D: Surface marker profiling of hBMSCs by flow cytometry. Mesenchymal stem cells showed higher levels of CD73, CD90, and CD105, and lower levels of CD34; E: Representative images of the DiI-labeled hBMSCs; F: The fluorescence imaging of different organs. hBMSCs: Human bone-marrow-derived mesenchymal stem cells.

To verify multilineage differentiation potential, hBMSCs were induced toward osteogenic and adipogenic lineages. After 21 days of osteogenic induction, Alizarin Red S staining revealed abundant red mineralized nodules, indicating calcium deposition (Figure 1B). Following 21 days of adipogenic induction, Oil Red O staining demonstrated numerous intracellular lipid droplets, confirming adipogenic differentiation (Figure 1C). Additionally, flow cytometry showed high expression of MSCs-positive markers CD73, CD90, and CD105 (> 95%) and minimal expression of the hematopoietic marker CD34 (< 2%) (Figure 1D). This immunophenotypic profile (CD34-/CD73+/CD90+/CD105+) conforms to the criteria for human MSCs defined by the International Society for Cell and Gene Therapy.

To evaluate in vivo distribution, hBMSCs were labeled with the fluorescent dye Dil, exhibiting stable orange fluorescence under confocal microscopy (Figure 1E). At 24 hours after tail vein injection, fluorescence imaging of cryosectioned organs revealed an organ-specific distribution pattern. A large number of Dil-positive cells were retained in the lung, primarily within pulmonary capillaries and alveolar septa, indicating significant pulmonary entrapment. Scattered Dil-positive cells were also observed in the heart, liver, and spleen, but were not detected in the brain (Figure 1F).

hBMSCs, but not FLX, ameliorate anxiety- and depression-like behaviors in chronically stressed mice

The experimental design and treatment schedule are illustrated in Figure 2A. In the open field test (OFT), heatmap analysis revealed that mice in the stress group predominantly remained in the peripheral zone and rarely entered the center, in contrast to the increased central exploration observed in the control, stress + FLX, and stress + hBMSCs groups (Figure 2B). No significant differences in locomotor speed were observed among groups (Figure 2C), excluding potential motor confounding effects. However, compared with the control group, mice in the stress group exhibited a significant reduction in the frequency of center entries (P < 0.01) (Figure 2D) and the time spent in the center (P < 0.001) (Figure 2E), whereas treatment with either FLX or hBMSCs markedly reversed these deficits. In the elevated plus maze test, the stressed mice primarily remained in the closed arms with minimal open-arm exploration (Figure 2F). Compared with the control group, significant reductions in both open-arm entries (P < 0.01) (Figure 2G) and time spent in open arms (P < 0.05) (Figure 2H) were observed in the stress group, deficits that were significantly improved by hBMSCs, but not FLX, treatment.

Figure 2 Human bone-marrow-derived mesenchymal stem cells ameliorated anxiety- and depression-like behaviors in stressed mice. A: Animal experiment flowchart; B: Representative heatmaps of open field test (OFT); C: Speed in the OFT; D: Frequency of entries into the central zone; E: Residence time in the central area; F: Representative heatmaps of elevated plus maze test; G: Frequency of entries into the open arm; H: Residence time in the open arm; I: Immobility time in the tail suspension test; J: Immobility time in the forced swim test; K: The percentage of sucrose preference. ^aP < 0.05, ^bP < 0.01, ^cP < 0.001. Data were analyzed using one-way analysis of variance followed by Tukey’s multiple-comparisons test (mean ± SEM, n = 12 per group). FLX: Fluoxetine; hBMSCs: Human bone-marrow-derived mesenchymal stem cells; OFT: Open field test; EPM: Elevated plus maze; TST: Tail suspension test; FST: Forced swim test; SPT: Sucrose preference test.

In the forced swim test and tail suspension test, mice in the stress group exhibited significantly prolonged immobility times compared with the control group, indicating depression-like behavior (P < 0.01 or 0.001) (Figure 2I and J). This increase in immobility was markedly reduced by treatment with hBMSCs, but not by FLX. Stressed mice showed significantly reduced sucrose preference in the sucrose preference test compared to controls, reflecting anhedonia. This deficit was significantly ameliorated by hBMSCs treatment, whereas FLX administration had no effect (P < 0.01) (Figure 2K). The OFT and histochemical analyses in healthy mice revealed no significant differences following hBMSC administration (Supplementary Figure 1). These findings indicate that the stress paradigm successfully established an FRD mouse model characterized by nonresponse to FLX, whereas injection of hBMSCs reversed the anxiety- and depression-like behaviors in stressed mice.

hBMSCs promote a shift from a proinflammatory to anti-inflammatory microenvironment, alleviating hippocampal neuroinflammation

To assess hippocampal neuroinflammation, we evaluated microglial activation using immunofluorescence. Ionized calcium-binding adapter molecule (Iba)1, a pan-microglial marker, was significantly upregulated in the CA1, CA3, and DG regions of stressed mice, as demonstrated by an increased number of Iba1+ cells (Figure 3A). Quantitative analysis confirmed significant elevations in Iba1 intensity across all three subregions in the stress group compared to the control group, which were markedly reversed by FLX or hBMSCs (P < 0.001) (Figure 3B-D). Similarly, CD68, a marker of microglial phagocytic activity, was markedly increased in several hippocampal subregions following stress (Figure 3E). Quantitative analysis confirmed significantly elevated fluorescence intensity in the stress group compared to controls, which was reversed by treatment with FLX or hBMSCs (P < 0.05, 0.01 or 0.001) (Figure 3F-H). In the CA1 region, hBMSC treatment resulted in a significantly greater reduction in CD68 fluorescence intensity compared to FLX treatment (Figure 3F).

Figure 3 Human bone-marrow-derived mesenchymal stem cells suppressed microglial overactivation. A: Representative immunofluorescence images of ionized calcium-binding adapter molecule (Iba)1 in the different subregions of the hippocampus; B-D: Quantification of relative fluorescence intensity of Iba1 in CA1 (B), CA3 (C), and dentate gyrus (DG) (D) regions; E: Representative immunofluorescence images of CD68 in the different subregions of the hippocampus: F-H: Quantification of relative fluorescence intensity of CD68 in CA1 (F), CA3 (G), and DG (H) regions. ^aP < 0.05, ^bP < 0.01, ^cP < 0.001. Data were analyzed using one-way analysis of variance followed by Tukey’s multiple-comparisons test (mean ± SEM, n = 5 per group). FLX: Fluoxetine; hBMSCs: Human bone-marrow-derived mesenchymal stem cells; DG: Dentate gyrus.

Western blotting validated these findings (Figure 4A). Chronic stress significantly increased protein expression of Iba1 and CD68, whereas hBMSCs treatment markedly reduced their levels to an extent comparable to, or even exceeding, that with FLX treatment (P < 0.001) (Figure 4B and C). In contrast, chronic stress decreased expression of the anti-inflammatory microglial marker arginase 1 (Arg1), which was significantly restored by FLX or hBMSCs (P < 0.01) (Figure 4D). Compared with the stress group, FLX and hBMSCs significantly reduced the proinflammatory cytokine TNF-α (P < 0.001) (Figure 4E) and markedly increased the anti-inflammatory cytokines TGF-β (P < 0.001) (Figure 4F) and IL-10 (P < 0.01 or 0.001) (Figure 4G). These results demonstrate that injection of hBMSCs attenuates chronic stress-induced hippocampal neuroinflammation through suppression of microglial activation and phagocytic phenotype, accompanied by promotion of an anti-inflammatory microenvironment. These effects were more pronounced than with FLX treatment.

Figure 4 Human bone-marrow-derived mesenchymal stem cells alleviated hippocampal neuroinflammation in stressed mice. A: Western blot plot of protein in the different groups; B-G: Relative expression levels of CD68 (B), ionized calcium-binding adapter molecule 1 (C), arginase 1 (D), tumor necrosis factor-α (E), transforming growth factor-β (F), interleukin-10 (G), and normalized to GAPDH. ^aP < 0.05, ^bP < 0.01, ^cP < 0.001. Data were analyzed using one-way analysis of variance followed by Tukey’s multiple-comparisons test (mean ± SEM, n = 4 per group). TGF: Transforming growth factor; Arg1: Arginase 1; TNF: Tumor necrosis factor; IL: Interleukin; Iba1: Ionized calcium-binding adapter molecule 1.

hBMSCs protect hippocampal astrocytes from stress-induced damage and reactivity

To evaluate the effects of hBMSCs on hippocampal astrocytes, glial fibrillary acidic protein (GFAP), a marker of astrocytic structural integrity and homeostatic function, and S100-β, a stress-associated astrocytic protein, were examined by immunofluorescence staining. Chronic stress significantly reduced GFAP fluorescence intensity and the number of GFAP+ cells in the CA1, CA3, and DG regions of the hippocampus; reductions that were completely reversed by FLX or hBMSCs treatment (P < 0.01 or 0.001) (Figure 5A-D). hBMSCs restored GFAP levels to those of controls, whereas FLX treatment did not achieve full recovery. In contrast, S100-β expression in the stress group was markedly increased across these subregions, exhibiting widespread and enhanced fluorescence signals; an effect that was suppressed by hBMSCs but not by FLX treatment (P < 0.01 or 0.001) (Figure 5E-H).

Figure 5 Human bone marrow-derived mesenchymal stem cells reduced the damage of hippocampal astrocytes in stressed mice. A: Representative immunofluorescence images of glial fibrillary acidic protein (GFAP) in the different subregions of the hippocampus; B-D: Quantification of relative fluorescence intensity of GFAP in CA1 (B), CA3 (C), and dentate gyrus (DG) (D) regions; E: Representative immunofluorescence images of S100-β in the different subregions of the hippocampus; F-H: Quantification of relative fluorescence intensity of S100-β in CA1 (F), CA3 (G), and DG (H) regions. ^aP < 0.05, ^bP < 0.01, ^cP < 0.001. Data were analyzed using one-way analysis of variance followed by Tukey’s multiple-comparisons test (mean ± SEM, n = 5 per group). GFAP: Glial fibrillary acidic protein; FLX: Fluoxetine; hBMSCs: Human bone-marrow-derived mesenchymal stem cells; DG: Dentate gyrus.

Western blotting revealed that chronic stress markedly decreased GFAP protein levels (P < 0.001) (Figure 6A and B) and increased S100-β expression (P < 0.001) (Figure 6A and C), which were reversed by both treatments. Although minor discrepancies were observed, the overall trends were consistent with the immunofluorescence findings. These results indicate that injection of hBMSCs alleviates chronic stress-induced astrocytic damage and attenuates astrocyte stress responses in the hippocampus.

Figure 6 Human bone marrow-derived mesenchymal stem cells improved hippocampal function in stressed mice. A: Western blot plot of protein in the different groups; B-F: Relative expression levels of glial fibrillary acidic protein (B), S100-β (C), NeuN (D), MAP2 (E), and DCX (F) normalized to GAPDH. ^aP < 0.05, ^bP < 0.01, ^cP < 0.001. Data were analyzed using one-way analysis of variance followed by Tukey’s multiple-comparisons test (mean ± SEM, n = 4 per group). GFAP: Glial fibrillary acidic protein; FLX: Fluoxetine; hBMSCs: Human bone-marrow-derived mesenchymal stem cells.

hBMSCs exert neuroprotective effects and promote hippocampal neurogenesis

To evaluate the effects of hBMSCs on hippocampal neurons, we examined the expression of NeuN (a mature neuron marker) and MAP2 (a dendritic structural protein) by immunofluorescence. Compared to the control group, chronically stressed mice showed significantly reduced density of NeuN+ neurons in the CA1 and DG regions, but not in CA3. These reductions were ameliorated by FLX or hBMSCs. In the DG, hBMSCs restored NeuN levels to those of controls, whereas FLX treatment failed to reverse the reduction (P < 0.001) (Figure 7A-D). Similarly, compared to the control group, MAP2+ fluorescence signals were markedly diminished and sparser in stressed mice, with disrupted dendritic network continuity. These alterations were reversed by treatment with FLX or hBMSCs (P < 0.01 or 0.001) (Figure 7E-H). Western blotting confirmed that hBMSCs reversed chronic stress-induced reductions in NeuN and MAP2 protein levels (P < 0.01 or 0.001) (Figure 6D and E). These results indicate that chronic stress induced neuronal loss and dendritic structural damage, and that hBMSCs significantly restored NeuN and MAP2 signal intensity, accompanied by increased dendritic network complexity.

Figure 7 Human bone-marrow-derived mesenchymal stem cells exerted neuroprotective effects on the hippocampus of stressed mice. A: Representative immunofluorescence images of NeuN in the different subregions of the hippocampus; B-D: Quantification of relative fluorescence intensity of NeuN in CA1 (B), CA3 (C), and dentate gyrus (DG) (D) regions; E: Representative immunofluorescence images of MAP2 in the different subregions of the hippocampus; F-H: Quantification of relative fluorescence intensity of MAP2 in CA1 (F), CA3 (G), and DG (H) regions. ^aP < 0.05, ^bP < 0.01, ^cP < 0.001. Data were analyzed using one-way analysis of variance followed by Tukey’s multiple-comparisons test (mean ± SEM, n = 5 per group). FLX: Fluoxetine; hBMSCs: Human bone-marrow-derived mesenchymal stem cells; DG: Dentate gyrus.

We next assessed hippocampal neurogenesis by examining DCX (an immature neuron marker) and BrdU (a proliferation marker). DCX+ cells were defined as newly generated neurons, representing neuronal lineage cells at an early differentiation stage. Immunofluorescence revealed that the chronic stress-induced reduction in DCX+ cells in the DG region was reversed by FLX or hBMSCs (P < 0.001) (Figure 8A and B). Similarly, the number of BrdU+ proliferating cells was markedly reduced in stressed mice and significantly increased following treatment with either FLX or hBMSCs (P < 0.001) (Figure 8C and D). Western blotting confirmed that hBMSCs markedly increased DCX protein expression (P < 0.01) (Figure 6F). These results demonstrate that hBMSCs ameliorated chronic stress-induced suppression of hippocampal neurogenesis by promoting neural precursor cell proliferation and their subsequent differentiation into neurons. These results indicate that chronic stress suppressed the proliferative activity of neural precursor cells and differentiation/maturation of newly generated neurons. Following hBMSCs intervention, the number of DCX+ immature neurons and NeuN+ mature neurons, as well as dendritic morphology, were significantly restored.

Figure 8 Human bone marrow-derived mesenchymal stem cells promoted the proliferation of newborn hippocampal neurons in stressed mice. A: Representative immunofluorescence images of DCX in the dentate gyrus (DG) region of the hippocampus; B: Quantification of DCX in the DG region; C: Representative immunofluorescence images of BrdU in the DG region of the hippocampus; D: Quantification of BrdU in the DG region. ^aP < 0.05, ^bP < 0.01, ^cP < 0.001. Data were analyzed using one-way analysis of variance followed by Tukey’s multiple-comparisons test (mean ± SEM, n = 5 per group for B, n = 3 per group for D). FLX: Fluoxetine; hBMSCs: Human bone-marrow-derived mesenchymal stem cells.

hBMSCs reversed the stress-induced downregulation of tyrosine hydroxylase expression in the hippocampus

Monoaminergic neurons play a key role in the pathophysiology of depression and the therapeutic effects of antidepressants[31]. Therefore, we examined the impact of hBMSCs on hippocampal monoaminergic neurons by assessing tyrosine hydroxylase (TH) expression. Immunofluorescence staining revealed that the fluorescence intensity of TH+ neurons in the CA1, CA3, and DG subregions was significantly reduced in the stress group compared to controls; these alterations were reversed by treatment with FLX or hBMSCs (P < 0.01 or 0.001) (Figure 9A-D). Consistent results were obtained by western blotting (P < 0.01 or 0.001) (Figure 9E and F). These findings imply that hBMSCs may ameliorate chronic stress-induced damage to hippocampal monoaminergic neurons.

Figure 9 Neuroprotective effects of human bone marrow-derived mesenchymal stem cells were associated with the brain-derived neurotrophic factor/tropomyosin receptor kinase B/cAMP response element-binding protein signaling pathway. A: Representative immunofluorescence images of tyrosine hydroxylase (TH) in the different subregions of the hippocampus; B-D: Quantification of relative fluorescence intensity of TH in CA1 (B), CA3 (C), and dentate gyrus (DG) (D) regions (n = 5 per group); E: Western blot plot of protein in the different groups; F-H: Relative expression levels of TH (F), tropomyosin receptor kinase B (G), and brain-derived neurotrophic factor (H) normalized to GAPDH; I: Relative levels phospho-cAMP response element-binding protein (CREB)/CREB. (n = 4 per group). ^aP < 0.05, ^bP < 0.01, ^cP < 0.001. Data were analyzed using one-way analysis of variance followed by Tukey’s multiple-comparisons test (mean ± SEM). FLX: Fluoxetine; hBMSCs: Human bone-marrow-derived mesenchymal stem cells; TH: Tyrosine hydroxylase; DG: Dentate gyrus; TrkB: Tropomyosin receptor kinase B; p-CREB: Phospho-cAMP response element-binding protein; CREB: CAMP response element-binding protein; BDNF: Brain-derived neurotrophic factor.

Neuroprotective effects of hBMSCs were associated with the BDNF/TrkB/CREB signaling pathway

Given that the BDNF/TrkB/CREB signaling pathway plays a critical role in antidepressant response[32], anti-inflammation[33], neurogenesis[21], and synaptic transmission[34], we further investigated its expression in the hippocampus. Western blotting demonstrated that expression of BDNF, TrkB, and phospho-CREB protein was significantly downregulated in the hippocampus of stressed mice compared to controls. hBMSCs markedly upregulated expression of these proteins to an extent comparable to FLX (P < 0.01 or 0.001) (Figure 9E and G-I). These results indicate that the antidepressant, anti-inflammatory, and neurogenic effects of hBMSCs may involve the BDNF/TrkB/CREB signaling pathway.

hBMSCs exert protective effects against chronic stress-induced inflammatory damage in the liver and kidney

We next evaluated the pathological status of major organs, including the heart, liver, spleen, lung, and kidney (Supplementary Figure 2). Histopathological analysis revealed that stressed mice exhibited pronounced inflammatory changes specifically in the liver and kidney. In the liver, marked inflammatory cell infiltration was observed in the portal areas and around the central veins, accompanied by hepatocyte edema and disorganized cellular architecture. In the kidneys, diffuse inflammatory cell infiltration in the renal interstitium and swelling of renal tubular epithelial cells were evident. Notably, these pathological alterations appeared to be partially alleviated following treatment with FLX or hBMSCs, as suggested by reduced inflammatory cell infiltration and some restoration of tissue architecture. No significant pathological abnormalities were observed in the heart, spleen, or lung across all groups.

DISCUSSION

We demonstrated that systemic administration of hBMSCs exerted robust antidepressant-like effects in a mouse model of FRD induced by combined CURS and CORT exposure. These behavioral improvements were accompanied by coordinated neurobiological changes in the hippocampus, including attenuation of microglial activation, restoration of astrocytic integrity, protection against neuronal damage, and enhancement of adult hippocampal neurogenesis. These effects were associated with the BDNF/TrkB/CREB signaling pathway, suggesting that hBMSCs exert antidepressant effects through a multimodal mechanism involving neuroimmune modulation and neuroplasticity restoration. Our findings provide experimental evidence supporting the therapeutic potential of MSC-based interventions for FRD. A schematic summary of the proposed mechanism is presented in Figure 10.

Figure 10  Proposed mechanisms underlying the antidepressant effects of human bone marrow-derived mesenchymal stem cells in fluoxetine-resistant depression. The figure was created with BioRender.com. FRD: Fluoxetine-resistant depression; CURS: Chronic unpredictable restraint stress; CORT: Corticosterone; SPT: Sucrose preference test; FST: Forced swim test; TST: Tail suspension test; BDNF: Brain-derived neurotrophic factor; TrkB: Tropomyosin receptor kinase B; TH: Tyrosine hydroxylase; GFAP: Glial fibrillary acidic protein; IL-10: Interleukin-10; Arg1: Arginase 1; TGF: Transforming growth factor; hBMSCs: Human bone-marrow-derived mesenchymal stem cells; Iba1: Ionized calcium-binding adapter molecule 1; TNF: Tumor necrosis factor.

Animal models of depression are typically established through various approaches, including physical[35], chemical[36], biological[37], and genetic[38] manipulations. Among these, chronic restraint stress (CRS) is a widely used strategy to induce MDD in rodents[39]. Unpredictability is considered a key factor in the pathogenesis of depression[40]; accordingly, incorporating an unpredictable element into CRS gives rise to the CURS model. For instance, Mograbi et al[29] successfully induced depression-like behavior in rats using a 3-week CURS protocol. Chronic exposure to elevated glucocorticoids has been shown to impair hippocampal neurogenesis, promote neuroinflammation, and reduce responsiveness to conventional monoaminergic antidepressants[41]. CORT administration is therefore another common strategy for inducing depression-like phenotypes in rodents[42]. In the present study, the combination of prolonged CURS and exogenous CORT exposure was designed to mimic both chronic environmental stress and sustained glucocorticoid elevation. The lack of behavioral responsiveness to FLX further supports the drug-resistant phenotype of this model. Similar approaches have been used in previous studies to reproduce key features of TRD in rodents[43-46].

The dose and administration route of hBMSCs in this study were optimized based on previous literature on stem cell therapy for depression. For example, weekly intravenous injection of 10⁶ adipose-derived stem cells for 3 weeks alleviated depressive-like behavior in mice exposed to chronic mild stress[47]. Another study demonstrated that 2 × 10⁶ human umbilical-cord-derived MSCs produced superior therapeutic effects compared with the 10⁶ dose in a CUMS mouse model[30]. Given that our depression model mice experienced prolonged physical and pharmacological stress, we selected a higher dose and longer treatment duration in this study.

Increasing evidence suggests that neuroinflammation plays a pivotal role in the pathophysiology of depression. Activated microglia release proinflammatory cytokines such as IL-1β, IL-6, and TNF-α, which can disrupt synaptic plasticity, inhibit neurogenesis, and contribute to depressive behaviors[9]. In our study, chronic stress induced pronounced microglial activation in the hippocampus, consistent with previous findings in stress-related depression models[48]. Treatment with hBMSCs significantly attenuated microglial activation and shifted the cytokine profile toward an anti-inflammatory state. These findings align with previous reports demonstrating that MSCs exert antidepressant effects through immunomodulation. For instance, in a 7-week CUMS-induced mouse depression model, stereotaxic injection of 10⁵ BMSCs into the hippocampus exerted antidepressant efficacy surpassing even that of FLX, and inhibited neuroinflammation by promoting the conversion of microglia into an Arg1^high anti-inflammatory state; an effect attributed to secretion of TGF-β1[24]. Similarly, in another study using a Sprague-Dawley rat model of depression established by 2 weeks of CUMS combined with lipopolysaccharide administration, intravenous injection of 5 × 10⁶ human umbilical cord-derived MSCs over 4 weeks reversed depression-like behavior, suppressed proinflammatory cytokine expression, and ameliorated hippocampal morphological abnormalities[49]. These findings suggest that suppression of neuroinflammation represents a key mechanism underlying the antidepressant effects of MSCs.

Hippocampal neurogenesis is another well-documented mechanism implicated in both depression pathophysiology and antidepressant efficacy. Accumulating evidence suggests that promotion of neurogenesis represents a key mechanism underlying the therapeutic effects of MSCs[50]. For instance, Tfilin et al[51] demonstrated that intracerebroventricular injection of 10⁵ BMSCs in Flinders sensitive line rats - a genetic model of depression - produced antidepressant-like effects and promoted neurogenesis in the ipsilateral hippocampus. Similarly, implantation of 3 × 10⁵ encapsulated MSCs into the lateral ventricle of Wistar Kyoto rats counteracted depression-like behavior and enhanced endogenous neurogenesis[52]. Our findings extend these observations by demonstrating that intravenous administration of hBMSCs exerted antidepressant efficacy in an FRD mouse model induced by combined physical and chemical stressors, further expanding the therapeutic applicability of MSCs.

Astrocytes are increasingly recognized as critical regulators of synaptic homeostasis and neuronal survival[53]. Stress-induced astrocytic dysfunction can disrupt metabolic support to neurons and impair synaptic plasticity[54]. In the present study, chronic stress markedly reduced GFAP expression and astrocytic density in the hippocampus, indicating astrocytic impairment. These findings are consistent with a previous report by Aten et al[55], showing that 6-week CUMS exposure disrupted astrocyte network structure and function in a mouse model of depression. hBMSC treatment restored astrocytic integrity, which may contribute to improved neuronal survival and enhanced neurogenesis. Indeed, we observed increased numbers of DCX+ immature neurons and NeuN+ mature neurons following hBMSC administration. These findings suggest that restoration of astrocyte-neuron interactions represents an important mechanism through which MSCs promote hippocampal plasticity and alleviate depressive-like behavior.

Dil-labeled hBMSCs were not detected in the brain following systemic administration, suggesting that the antidepressant effects observed in this study were unlikely to have depended on direct engraftment within the central nervous system. Due to their large size, intravenously administered MSCs are thought to become entrapped in the lungs, where they exert paracrine effects via secreted factors before being cleared[27]. Our findings therefore support the notion that MSCs exert therapeutic effects primarily through indirect, peripheral mechanisms[56]. Accumulating evidence indicates that MSCs can modulate systemic immune responses, release bioactive molecules and extracellular vesicles (EVs), and influence peripheral-central immune communication[57]. MSC-derived EVs have garnered particular interest, as they can carry a diverse cargo of microRNAs, proteins, and lipids that mimic the immunomodulatory and neurotrophic functions of their parent cells[58]. We did not directly isolate or characterize hBMSC-derived EVs in this study. Future studies examining the role of MSC-derived EVs are essential to delineate the specific molecular mediators underlying their therapeutic effects. We observed that hBMSCs also exerted anti-inflammatory effects in the hippocampus and peripheral tissues, particularly in the liver and kidneys. These observations suggest that peripheral immunomodulation and MSC-derived paracrine signaling represent key mechanisms mediating the neuroprotective and antidepressant effects observed in this study.

Our study had several limitations. First, the FRD model was established based on nonresponse to FLX, which can only reflect the initial stage of TRD. The clinical definition of TRD typically requires nonresponse to at least two different classes of antidepressants[59]. This discrepancy limits the construct validity of our model. Second, while we assessed multiple molecular changes (e.g., neuroinflammation, neurogenesis, and neurotransmitter synthesis), we did not examine synaptic plasticity; a key functional endpoint that integrates these pathophysiological alterations[60]. Third, although previous studies have suggested that the antidepressant effects of MSCs are largely mediated by their secreted bioactive molecules (e.g., microRNAs, proteins, and other paracrine factors)[61], we did not identify the specific molecules responsible for these effects in our model, leaving the precise upstream molecular mechanisms unknown. Fourth, because of sex differences in stress responses[62], this study was performed solely in male mice. Hence, the current findings are limited to males, and future validation in females is required to comprehensively evaluate the therapeutic efficacy of hBMSCs in both sexes. Finally, although we observed that the BDNF/TrkB/CREB signaling pathway was associated with the multifaceted molecular effects of hBMSCs, we did not perform loss- or gain-of-function experiments to manipulate this pathway; therefore, we cannot definitively establish a causal relationship. Further studies using genetic modulation or pharmacological inhibition of this pathway in our model are warranted to elucidate its causal role.

CONCLUSION

This study demonstrates that intravenous injection of hBMSCs exerts robust antidepressant effects in an FRD mouse model. The therapeutic efficacy of hBMSCs is associated with multifaceted mechanisms, including suppression of neuroinflammation, restoration of astrocytic and neuronal function, and promotion of hippocampal neurogenesis, with involvement of the BDNF/TrkB/CREB signaling pathway. These findings highlight hBMSCs as a promising multi-target therapeutic candidate for FRD, as an early stage of TRD, offering a potential avenue to overcome the limitations of conventional antidepressants.

ACKNOWLEDGEMENTS

We thank Yi-Lin Wang, Xiao-Yan Chen, Bing-Long Wang, and Jie Huang for their helpful comments and discussions on the progress of the study, and Zhen Hu for her statistical review.