Signatures in the Protein Content of Human and Murine Blood Serum Exosomes, in the Context of Major Depressive Disorder, Are Associated with Cytokine Activity

Based on the fact that hormones influence neurotransmission, neuro-inflammatory responses, and HPA axis regulation [ 12 , 13 ], contributing to stress responses and vulnerability to depressive disorders in women [ 14 ], here, we focused on women. Thus, twenty women aged 22–36 years were enrolled in this study after providing written informed consent. Participants were recruited from the outpatient services of the Ramón de la Fuente Muñiz National Institute of Psychiatry (INPRFM). Twelve subjects were diagnosed with major depressive disorder (MDD) according to DSM-5 criteria by a certified psychiatrist and had a score greater than 13 on the 17-item Hamilton Depression Rating Scale (HDRS). Eight healthy control women (CTRL) were recruited from the general population and matched for age. Control participants were screened to exclude any current or past psychiatric disorders. All participants (diagnosed and healthy controls) were not receiving pharmacological treatment (including antidepressants, anti-inflammatory drugs, and antibiotics) at the time of enrollment. Subjects were excluded if they reported the use of hormonal or dietary supplements, contraceptives, or alternative treatments. Other exclusion criteria included the presence of infectious, inflammatory, metabolic, or autoimmune diseases, substance use (including alcohol, tobacco, and illicit drugs), and any history of psychotic, manic, or hypomanic episodes. Participants with known hormonal or gynecological disorders, including polycystic ovary syndrome or endometriosis, were excluded. No participants were pregnant during the study. Clinical, biochemical, and demographic variables were recorded ( Table 1 ). The human study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the Instituto Nacional de Psiquiatría Ramón de la Fuente Muñiz (INPRFM; approval number CEI/C/002/2018, 15 January 2018) and by the Comisión Nacional de Bioética (CONBIOÉTICA, approval number CONBIOÉTICA-09-CEI-010-20170316, 15 January 2018) ( Figure 1 A). To align our study performed in women to the general aim, we included female mice to be exposed to the CUMS. This model reflects clear effects of chronic stress on depression-like behavior and neurochemical modifications, as we mentioned in the last part of the Section 1 . A total of twenty Balb/C female mice, 8–12 weeks old and weighing 20 ± 2 g, were used in this study and divided into two groups: ten subjects under the CUMS model and ten healthy control (CTRL) subjects. The animals were facilitated by the INPRFM, and they were held in standard laboratory plexiglass cages (28 × 17 × 15 cm) with a 12/12 h light/dark cycle (light on at 07:00 h and lights off at 19:00 h) and access to food and water ad libitum. All the handling procedures complied with the Official Mexican Standard for Animal Care, as well as international guidelines and protocols [ 54 , 55 , 56 , 57 ]. The animal study protocol was approved by the Institutional Animal Care and Use Committee of the INPRFM (approval number CICUAL/05/2022, 12 June 2024) ( Figure 1 A). To induce depressive-like behaviors, the mice were subjected to the CUMS protocol. This protocol consisted of the application of one to three stressors of mild-to-moderate intensity during each day for a period of four to eight weeks ( Figure 2 A). The stressors were applied until the generation of observable and characteristic depressive-like behaviors in the mice (anhedonia, weight change, alterations in emotional self-care, psychomotor retardation, and hopelessness). The stressors included exposure to a cold room (4 °C), cage in horizontal rotation (40 rpm), inclined cage (45°), wet cage, empty cage, wet bedding (cage), overcrowding, light off (extension of the dark phase), light on (extension of the light phase), strobe light, predator odor, water deprivation, food deprivation, movement restriction, and reversal of light/dark cycle ( Figure 2 A). Also, in the CUMS-exposed group, the rodents underwent the forced swim test to confirm depressive like behavior [ 58 , 59 , 60 ]. According to Willner (2005) and Nollet (2021) [ 18 , 19 ], the sequence and timing of stressor application correspond to a logical and unpredictable order that does not harm the physical integrity of the animal according to the Grimace Scale [ 55 , 56 , 57 ]. The estrous cycle stage was determined by vaginal smear cytology according to Caligioni (2009) [ 61 ] at the end of the CUMS protocol and recorded for subsequent analysis. Also, corticosterone and 17β Estradiol were quantified by ELISA (Enzo Life Sciences, Farmingdale, NY, USA, ADI-900-097, ADI-900-008) according to the manufacturer’s instructions. The 96-well plates were read in a Promega ELISA reader (Glomax Discover, Madison, WI, USA Ref#GM3000) [ 61 , 62 , 63 ]. Coat condition is described as one of the most replicable indexes or scales in the monitoring of self-care behavioral alterations induced by the CUMS model, validated as a useful parameter for studying depressive-like behaviors [ 65 , 66 , 67 ]. The coat state (CS) analysis consists of assigning a score from 0 to 1 to the coat appearance, where a value of 0 corresponds to a good coat condition (smooth, shiny and unaltered coat), a value of 0.5 corresponds to an intermediate condition (slightly fluffy coat), and a value of 1 corresponds to a very bad condition (fluffy and unkempt, with patches on the coat, almost all over the body) [ 18 , 19 ]. The recording of the score was performed in five areas of the mice’s body, namely, the head, neck, torso, abdomen, and tail, three times per week during the whole protocol, and the data obtained were analyzed to identify the appearance and evolution of depressive-like behaviors in the female Balb/C mice. Measurement of mouse body weight has also been reported as an important index of self-care behavioral alterations induced by the CUMS model. In this study, the weight of each mouse was recorded three times per week throughout the protocol, and statistical comparisons were made between experimental groups. The forced swim test (FST) was used to assess depressive-like behavior in the mice [ 58 , 59 , 68 ] exposed to the CUMS protocol, but not in the control group, to avoid confounding effects of the acute stress produced by the FST. The behavioral outcomes of the CUMS-exposed group were compared against previous data published by our group [ 64 ]. The test consisted of a single session performed at the end of the CUMS protocol. The mice were individually placed in a 12 × 30 cm cylinder filled to half its capacity with water maintained at room temperature (25 °C ± 2), resulting in a water depth of 15 cm, and allowed to swim for up to 5 min (300 s). During this period, the sessions were video-recorded and analysed using AnyMaze behavioural tracking software version 7.01 (Stoelting Co., Wood Dale, IL, USA) [ 64 ]. Immobility-related parameters, including the total immobility time and latency to immobility, were quantified. Immobility was defined as the absence of any movement for at least 2000 ms. Human and murine blood and serum samples were processed according to the recommendations for collection, handling, storage, isolation, separation, characterization, and reproducibility described in the technical guidelines of the International Society for Extracellular Vesicles (ISEV), including MISEV2023 [ 69 ] and MiBlood-EV [ 70 ], among others [ 71 , 72 , 73 ]. In addition, the guidelines established in the Mexican Official Standards NOM-087 and NOM-166 were followed [ 74 , 75 ]. Fresh blood samples from the human participants were collected at the Clinical Laboratory of the INPRFM between 7:00 and 9:00 a.m. after at least 8 h of fasting. Blood was obtained from the antecubital vein by venipuncture using a BD Vacutainer ® system (21-gauge Eclipse needle and 5 mL SST tube for serum; BD Vacutainer, Becton Dickinson, Franklin Lakes, NJ, USA). Serum was separated by centrifugation at 1500× g for 15 min at room temperature (RT) using a Smart R17 centrifuge (Hanil Science Industrial, Gimpo, Republic of Korea) ( Figure 1 B). The mice were euthanized by decapitation, and fresh blood samples were collected in 1.5 mL microtubes. Brain tissue was also collected for subsequent histological analyses. Mouse serum was obtained after centrifugation at 6000× g for 30 min at RT using a MIKRO 200R centrifuge (Hettich Zentrifugen, Tuttlingen, Germany) ( Figure 1 B). All aliquots obtained from collected serum samples were processed within 2 h of blood collection. Serum samples were clarified to remove debris by filtration through a 0.22 μm nitrocellulose membrane followed by centrifugation at 2000× g for 30 min, prior to EVs isolation. Throughout the procedures, the samples were maintained under cold-chain conditions (4 °C ± 2). Unnecessary freeze–thaw cycles were avoided, and only three were allowed per sample. Finally, two to three microtubes containing aliquots of 0.1–0.5 mL were frozen and stored at −80 °C until further analysis, including the isolation and characterization of EVs ( Figure 1 B). Extracellular vesicles were isolated using the Total Exosome Isolation reagent (from serum) (Invitrogen™, Thermo Fisher Scientific, Waltham, MA, USA, #4478360) according to the manufacturer’s protocol. Briefly, 500 μL of clarified serum obtained from human and mouse samples was mixed with 100 μL of the isolation reagent and incubated for 30 min at 8 °C. Exosomes were then precipitated by centrifugation at 10,000× g for 10 min using a Smart R17 centrifuge (Hanil Science Industrial, Gimpo, Republic of Korea). After removal of the supernatant, the pellet was washed twice with 1× PBS. This precipitation-based method allows the enrichment of exosome-associated EVs from serum samples. Finally, the EV pellets were resuspended in the appropriate buffer depending on the downstream application: 1× PBS for EV quantification, characterization, and morphological analyses (TEM), or 1× RIPA buffer for vesicle lysis and access to their molecular cargo ( Figure 1 B). The total protein concentration of exosome-enriched EVs isolated from human and murine serum samples was determined using the Bradford assay (Abcam, Cambridge, UK, # AB119216 ). All samples were homogenized and lysed using 1× RIPA buffer supplemented with protease (cOmplete, Roche, Basel, Switzerland, #11697498001) and phosphatase inhibitors cocktail (PhosStop, Roche, #04906845001), and treated with three 15 s cycles of ultrasound (VibraCell™ SONICS, Newtown, CT, USA) at 4–8 °C. A standard curve was generated using bovine serum albumin (BSA; Research Organics, Cleveland, OH, USA, #1332A), and the total protein content in 1 μL of each sample was determined by measuring absorbance. Protein concentrations were calculated by linear regression analysis based on the BSA standard curve ( Figure 1 C). In addition, the relative abundance of CD63-positive EVs was assessed using an Exo ELISA-Ultra CD63 assay (System Biosciences, Palo Alto, CA, USA, EXEL-ULTRA-CD63-1). Briefly, the equivalent of 100 μg of exosome-enriched EV-isolated protein from non-lysed exosome-enriched EVs preparations resuspended in PBS was incubated in a 96-well microplate with capture and binding buffer according to the manufacturer’s protocol. Subsequently, the wells were incubated with a primary antibody against the EV surface marker CD63 (a tetraspanin protein). Signal detection was performed using a colorimetric substrate, and absorbance was measured at 450 nm using a Glomax Discover multimode plate reader (Promega, Madison, WI, USA, Ref#GM3000). Quantification was performed using a CD63 standard curve followed by linear regression analysis of absorbance values for each sample ( Figure 1 C). Finally, the particle-to-protein ratio—defined as the ratio of total protein concentration to the number of particles—was calculated for the experimental groups of both species (CTRL and MDD; CTRL and CUMS). Analysis of EV-associated transmembrane markers was performed using an Exo-Check Antibody Array (Neuro) Standard Kit (System Biosciences, EXORAY500A-8, Palo Alto, CA, USA), following the manufacturer’s instructions. Briefly, 200 μg of exosome-enriched EVs resuspended in 1× PBS were lysed using 10× lysis buffer to obtain a final concentration of 1×. The lysed exosome-enriched EV samples from one human or one mouse were then incubated with the antibody array membrane containing immobilized antibodies against common EV markers overnight at 4–8 °C. Chemiluminescent HRP substrate (Immobilon ® Western, Cat. No. WBKLS0100, Millipore™, Burlington, MA, USA) was added, and signals were detected using a ChemiDoc Touch Imaging System (Bio-Rad, Hercules, CA, USA) ( Figure 1 C and Figure 3 D,J). The presence of the EV-associated tetraspanin CD63 was confirmed by slot blot analysis. Briefly, 1, 5, 10, and 15 μg of total protein from non-lysed exosome-enriched EVs derived from human (CTRL, MDD) and murine (CTRL, CUMS) blood serum samples were loaded onto a nitrocellulose membrane (45 μm size pore, Millipore™ Cat. No. HATF00010, Dublin, Ireland) using a vacuum-driven slot blot manifold (Hybri-Slot™ Manifold, Cat. No. 1052M, Life Technologies, Inc., Waltham, MA, USA). Phosphate-buffered saline (PBS 1×) was used as a negative control, and primary antibody CD63 was used as a positive control. After the sample application, the membrane was air-dried at room temperature and subsequently blocked with a solution containing 3% non-fat dry milk, 0.1% bovine serum albumin (BSA), and 0.1% Tween-20 in Tris-buffered saline (TBS) for 30 min under agitation. The membrane was then washed three times for 10 min with TBS-T (0.05% Tween-20) under agitation. Then, the membrane was incubated overnight at 4–8 °C under agitation with a primary antibody against CD63 (1:2000 dilution in TBS-T 0.05%; Ab216130 , Abcam, Cambridge, UK). After washing, the membrane was incubated with an HRP-conjugated anti-rabbit secondary antibody (1:10,000 dilution in TBS-T 0.05%) for 1 h at room temperature under agitation. After the incubation, the membrane was washed three times with TBS-T. Chemiluminescent detection was performed using an HRP substrate (Immobilon ® Western, Cat. No. WBKLS0100, Millipore™, Burlington, MA, USA). Signals were acquired with a ChemiDoc Touch Imaging System (Bio-Rad), and densitometry was done with Imagelab software Version 6.1.0 build 7 (Bio-Rad, Hercules, CA, USA) ( Figure 1 C and Figure 3 E,J). The morphology and size of isolated exosome-enriched EVs were evaluated by transmission electron microscopy (TEM). Briefly, a 200-mesh copper grid coated with carbon film (Electron Microscopy Sciences, Morgantown, PA, USA) was placed on a drop of exosome-enriched EV suspension resuspended in 1× PBS and incubated for adsorption. Excess liquid was removed using Whatman filter paper, and the grids were fixed with 2.5% glutaraldehyde. The grids were then washed twice with filtered deionized water (0.22 μm) and negatively stained with uranyl acetate. Excess stain was removed using filter paper, and the grids were allowed to air-dry. The samples were examined at accelerating voltages of 60–100 kV using a JEOL JEM-1010 transmission electron microscope (JEOL, Tokyo, Japan) equipped with an ATM digital camera (JEOL, Tokyo, Japan) ( Figure 1 C). A semi-quantitative analysis of cytokine and chemokine expression was performed using lysed exosome-enriched EV samples isolated from human and mouse serum samples obtained from the experimental groups ( Figure 4 A). All the human samples were analyzed ( Figure 4 C), but only eight mouse samples (n = 4 CTRL and n = 4 CUMS) were analyzed due to the quantity of protein available ( Figure 4 D). Cytokine and chemokine profiles were evaluated using Human Cytokine Antibody Array C1000 (AAH-CYT-1000-8) and Mouse Cytokine Antibody Array C1000 (AAM-CYT-1000-8) kits (RayBiotech ® , C-Series, Norcross, GA, USA), allowing the simultaneous detection of 120 and 96 analytes, respectively. Both assays were performed in a two-membrane format (C6 and C7 for human samples; C3 and C4 for mouse samples) according to the manufacturer’s instructions. The arrays are based on sandwich immunoassay chemistry with chemiluminescent detection and exhibit detection sensitivities in the pg/mL range ( Figure 4 A). For each membrane, 250 µg of total EV protein was used for incubation. Chemiluminescent detection was performed using an HRP-based substrate (Immobilon ® Western, Cat. No. WBKLS0100, Millipore™, Burlington, MA, USA), and membrane signals were acquired using a ChemiDoc™ Touch Imaging System (Bio-Rad Laboratories, Hercules, CA, USA). Signal intensities were quantified by densitometric analysis using Image Lab 6.0.1 software (Bio-Rad Laboratories, Hercules, CA, USA). Spot intensities were measured as integrated optical density (OD) using identical circular regions of interest for all spots, and duplicate spots corresponding to each cytokine were averaged. Background correction and signal normalization were performed using the manufacturer’s analysis templates (RayBiotech ® , C-Series Cytokine Array Analysis Tool, Norcross, GA, USA). Normalized signal intensities were calculated relative to the mean signal of the internal positive and negative control spots present on each membrane. These normalized values were subsequently used to calculate relative cytokine expression levels between the experimental groups. For the statistical analysis of the human and mouse cytokine arrays, unpaired Student’s t -tests were performed to evaluate the relative expression levels of each cytokine and chemokine (120 and 96 analytes for the human and mouse arrays, respectively). In addition, a two-way ANOVA (with Bonferroni’s test for multiple comparisons) was conducted to analyze the relative expression of the 52 cytokines and chemokines shared between both membranes, considering species and experimental condition as factors ( Figure 4 B). Bioinformatic analyses were performed based on the statistical results obtained from the comparisons of relative cytokine and chemokine expression levels between the experimental groups. Only proteins showing statistically significant differences between groups and species were included in the analyses ( Figure 4 B). Protein annotation and functional information were obtained using the UniProt database (Universal Protein Resource, version release 2024_01. Developed and maintained by the UniProt Consortium, including the European Bioinformatics Institute (EMBL-EBI, Hinxton, Cambridge, UK), the Swiss Institute of Bioinformatics (SIB, Geneva, Switzerland), and the Protein Information Resource (PIR, Washington, DC, USA); Access on 24 November 2024, in https://www.uniprot.org/ ). Protein–protein interaction (PPI) networks were analyzed using the STRING database platform (Functional Protein Association Networks, STRING Consortium, version 11.5. Developed and maintained by the STRING Consortium, including the Swiss Institute of Bioinformatics (SIB, Geneva, Switzerland), the Novo Nordisk Foundation Center for Protein Research (CPR, University of Copenhagen, Copenhagen, Denmark), and the European Molecular Biology Laboratory (EMBL, Heidelberg, Germany); Access on 24 November 2024, in https://string-db.org/ ) to identify functional interaction networks between proteins and their associated genes. Network construction was performed using a minimum required interaction score of 0.700 (high confidence) [ 76 , 77 ]. Functional enrichment analyses were subsequently carried out to evaluate associated biological processes, cellular components, and molecular functions using Gene Ontology (GO) annotation through the WebGestalt platform (WEB-based Gene Set Analysis Toolkit, Version 2024, Houston, TX, USA; Access on 24 November 2024, in http://www.webgestalt.org/ ). These analyses were complemented by pathway enrichment analysis using the KEGG (Kyoto Encyclopedia of Genes and Genomes) database [ 78 , 79 ]. In summary, an over-representation analysis (ORA) was performed separately for each organism ( Homo sapiens and Mus musculus ). The parameters for the enrichment analysis included a minimum of 5 IDs and a maximum of 2000. Multiple testing correction was applied using the Benjamini–Hochberg false discovery rate (FDR) method, and pathways with an FDR-adjusted p -value < 0.05 were considered statistically significant (q-value). All the bioinformatic analyses were performed in November 2024. Statistical analyses were performed using Microsoft Excel version 2603 (Microsoft Corporation, Redmond, WA, USA), SigmaPlot version 15.0 (Systat Software, Inc., San Jose, CA, USA) and GraphPad Prism version 8.4.2 (GraphPad Software, Inc., Boston, MA, USA). Data are presented as mean ± standard error of the mean (SEM). The appropriate statistical test was selected for each determination. Data distribution was assessed for normality using the Shapiro–Wilk test. Comparisons between two groups were performed using an unpaired two-tailed Student’s t -test for normally distributed data or the Mann–Whitney U test for non-normally distributed data. Multiple comparison corrections were applied using the Benjamini–Hochberg procedure. Comparisons involving two or more groups and factors were analyzed using two-way analysis of variance (ANOVA), followed by Bonferroni’s post hoc test for multiple comparisons. Differences were considered statistically significant at p < 0.05.

The present study has several methodological limitations that should be considered when interpreting the results. First, the EV isolation method based on precipitation could be optimized by employing size exclusion chromatography (SEC), which is currently regarded as a more robust approach that improves the purity and yield of small vesicles, such as exosomes, compared to traditional methods. In addition, the quantification of particle number and size should be complemented with orthogonal techniques, such as nanoparticle tracking analysis (NTA) or tunable resistive pulse sensing (TRPS), in order to more accurately validate the EV population under study and to comply with recommended characterization criteria. Regarding molecular and cell line characterization, the analysis of transmembrane markers could be strengthened by incorporating additional techniques, such as Western blotting, including both EV-enriched and EV-depleted (negative) markers, and confirming possible neural and glial origins in human and murine blood serum samples. However, all this must be confirmed by Western blot, mass spectrometry, or flow cytometry, in accordance with the recommendations established by the International Society for Extracellular Vesicles in the MISEV2023 guidelines [ 69 ]. This would enable a more comprehensive validation of the identity and purity of the isolated EVs. Another important limitation lies in the scope of the proteomic analysis, which was restricted to the identification of 120 and 96 proteins in human and murine samples, respectively, using antibody arrays targeting cytokines and chemokines. Although this approach enabled the identification of differentially expressed proteins of interest, it does not capture the full proteomic content of EVs. In this context, future studies should incorporate untargeted proteomic strategies, such as liquid chromatography–mass spectrometry (LC–MS), to achieve a more comprehensive characterization of EV cargo in the context of MDD and the CUMS model. These limitations may have influenced the sensitivity and specificity of EV detection and downstream analyses. Despite these limitations, this study represents an initial integrative approach that highlights the potential relevance of EV cargo in the neurobiology of MDD and stress-related responses. Furthermore, it provides a foundation for exploring EVs as a potential source of biomarkers in psychiatric disorders and for advancing our understanding of conserved molecular mechanisms across human and animal models.

Major depressive disorder (MDD) is one of the most prevalent and disabling psychiatric disorders worldwide [ 1 , 2 , 3 ]. MDD is characterized by persistent negative mood and thoughts accompanied by cognitive impairment, excessive fatigue, and alterations in appetite, libido, and sleep, as described in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) [ 2 , 4 ]. Approximately 5–10% of the global adult population experiences MDD [ 3 , 5 ]. Despite its high prevalence, depression is a complex and multifactorial disorder involving diverse biological mechanisms, including neurotransmitter dysregulation, structural brain alterations, changes in the hypothalamic–pituitary–adrenal (HPA) axis, neuro-inflammation, and molecular signaling abnormalities [ 6 , 7 , 8 , 9 ]. Another important aspect in the study of major depressive disorder is the marked sex-related difference in its prevalence and clinical presentation [ 10 , 11 ]. Epidemiological studies consistently report that women are affected by MDD approximately twice as frequently as men, particularly during the reproductive years [ 2 , 3 , 5 ]. This increased susceptibility has been associated with hormonal fluctuations in estrogen and progesterone, which can influence neurotransmission, neuro-inflammatory responses, and HPA axis regulation [ 12 , 13 ]. These biological factors may contribute to differences in stress responses and vulnerability to depressive disorders in women [ 14 ]. For this reason, focusing on women may provide clinically relevant information associated with depression. Consequently, the pathophysiology of MDD remains incompletely understood and requires integrative approaches that consider sex-related biological, cellular, molecular, and psychological mechanisms contributing to its development. In this context, animal models of stress have become essential tools for investigating the neurobiological mechanisms underlying depressive disorders [ 15 , 16 , 17 ]. Among these models, the chronic unpredictable mild stress (CUMS) protocol is widely used to induce depressive-like behaviors in rodents, including anhedonia, weight changes, impaired self-care, and behavioral despair [ 18 , 19 , 20 ]. These models facilitate the identification of neurobiological substrates associated with the disorder and enable the development of improved pharmacological and therapeutic strategies for the diagnosis, monitoring, and treatment of depression. In recent years, extracellular vesicles (EVs), particularly exosomes, have emerged as a promising approach for investigating cellular and molecular mechanisms underlying multiple pathological conditions [ 21 , 22 , 23 ]. EVs are nanosized lipid bilayer vesicles, typically ranging from approximately 30 to 200 nm in diameter, that mediate intercellular communication through the transfer of proteins, lipids, and nucleic acids [ 24 ]. Importantly, the molecular cargo of EVs reflects the physiological or pathological state of their cells of origin, suggesting that they may serve as informative indicators of disease processes [ 25 ]. In the central nervous system (CNS), EVs have been implicated in several critical physiological processes, including synaptic function and plasticity, neurogenesis, neuro-inflammation, and neural regeneration [ 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 ]. Furthermore, EVs can cross the blood–brain barrier (BBB), suggesting that EVs present in peripheral biofluids such as blood or cerebrospinal fluid (CSF) may reflect bidirectional communication between the CNS and peripheral tissues [ 22 , 27 , 34 , 35 ]. This property has generated increasing interest in EVs as a potential source of biomarkers for neurological and neuropsychiatric disorders [ 36 , 37 ], including depression. Indeed, several studies have reported alterations in EV characteristics in individuals with MDD. Analyses of morphological and physicochemical properties of plasma- or serum-derived EVs have revealed differences between patients with MDD and healthy controls [ 38 , 39 ]. In addition, multiple studies have focused on identifying differential EV cargo, particularly microRNAs (miRNAs), associated with depression. These include miR-139-5p [ 40 ], miR-207 [ 41 ], miR-26a [ 42 , 43 ], miR-494, miR-30c, miR-93, and miR-101 [ 43 ], as well as let-7e, miR-21-5p, miR-223, miR-145, miR-146a, and miR-155 [ 44 , 45 ], among others [ 46 , 47 , 48 , 49 , 50 , 51 ]. Several of these miRNAs have been investigated in vitro in neuronal, astrocytic, or microglial cultures [ 41 , 44 , 49 , 52 , 53 ], as well as in vivo in animal models [ 40 , 41 , 52 , 53 ], where they appear to regulate signaling pathways such as NF-κB, MAPK, Wnt, and mTOR, which are implicated in depressive-like behaviors [ 41 , 49 , 50 , 51 ]. Despite these advances, relatively few studies have investigated the protein cargo of EVs in the context of MDD, and comparative analyses between EVs derived from human patients and those obtained from experimental animal models remain limited. Understanding the similarities and differences between human and murine EV profiles may provide important insights into the translational relevance of rodent models for depression research. Therefore, in the present study, we compared the protein content of serum-derived EVs obtained from female patients diagnosed with MDD and from female mice subjected to CUMS, a well-established model of depressive-like behavior. By identifying common and differential EV protein signatures between humans and mice under healthy and depressive-like conditions, this study aims to evaluate EVs as a potential source of clinically relevant biochemical information and to explore their possible value as biomarker candidates for the diagnosis and monitoring of depression.

Table 1 shows the results of the Student’s unpaired two-tailed t -test for the 20 clinical, biochemical and hormonal parameters analyzed. For the menstrual cycle parameters, the Mann–Whitney U test was used for day of the cycle, and Fisher’s exact test was used for phase of the cycle. All healthy controls and depressed female participants were of the same age range and, according to their clinical histories and the inclusion and exclusion criteria, did not manifest any other medical or psychiatric disorder during their participation in the study. There was a significant difference between the healthy control group (2.41 ± 2.15) and participants with MDD (23.92 ± 4.68) on the 17-HDRS scale ( p = 0.0001), indicating that participants with MDD had experienced symptoms for more than 2 weeks, as defined by the DSM-5. In the hormonal evaluations, only FSH levels differed significantly ( p = 0.0328) between healthy control subjects (3.96 ± 1.76) and participants with MDD (5.66 ± 1.78). On the day of the menstrual cycle when blood was collected, there was a significant difference between the healthy control group (19.43 ± 3.631) and the participants with MDD (9.556 ± 2.561). Interestingly, although there were no significant differences in the menstrual cycle phase between the groups ( p = 0.0623), there was a possible trend toward a predominance of the luteal phase in the CTRL group (n = 5; 62.50%) and of the follicular phase in the MDD group (n = 10; 83.33%). Table 2 shows the results of the Student’s unpaired two-tailed t -test for the quantification of corticosterone and 17-β Estradiol. For the estrous cycle parameters, the Mann–Whitney U test was used for day of the cycle, and Fisher’s exact test was used for phase of the cycle. Here, we did not find significant differences in corticosterone or 17-β Estradiol. The results suggest a slight increase in corticosterone in the CUMS-exposed mice compared with the control group. Contrary results were found after the quantification of 17-β Estradiol, with a slight decrease in the CUMS-exposed mice. However, in both cases we did not find a statistically significant difference. The weekly analysis of CS in the CTRL and CUMS groups is shown in Figure 2 B. A two-way ANOVA of repeated measures was performed to compare the effect of stress exposure on coat condition (CS) and the time (weeks) at which changes in this behavior were observed. For the time factor, it was identified that the period between weeks 4 and 5 in the group exposed to the CUMS protocol was when a generalized poor condition and significant deterioration of the coat began to be observed (F(2.358, 28.29) = 52.70; p < 0.0001), with an increase in the mean CS score, characteristic of the manifestation of depressive-like behavior, compared to the CTRL group, which was not exposed to the CUMS protocol ( Figure 2 B). Regarding the stress exposure factor, a significant difference was observed between the CTRL and CUMS groups in the CS (F(1, 12) = 469.3; p < 0.0001) ( Figure 2 B). Finally, a significant interaction between the two factors (stress exposure and time) was identified (F(7, 84) = 65.57; p < 0.0001). On the other hand, in the weekly body weight records for each subject between groups ( Figure 2 C), no significant differences were observed between subjects exposed to the CUMS protocol and healthy CTRL subjects (F(1, 39) = 1.022; p = 0.3182). Taken together, these results suggest that mice under chronic stress showed impaired grooming, indicative of impaired self-care, without differences in body weight. As part of the behavioral characterization, at the end of the CUMS protocol, the rodents were exposed to acute stress induced by FS. Thus, the outcomes of those rodents were compared with data previously generated [ 68 ]. The mice subjected to the CUMS protocol exhibited a significant increase in the immobility time (31.30%) compared to the previously reported control group (t = 2.213, df = 15; p = 0.0428) [ 68 ] ( Figure 2 D). Similarly, latency to the first immobility episode was significantly reduced in the CUMS-treated mice (71.42%) compared to the previously reported control group (t = 6.862, df = 15; p < 0.0001) [ 68 ] ( Figure 2 E). Taken together, these findings are consistent with a depressive-like state in mice and in agreement with the CS measurements for the CUMS protocol ( Figure 2 B). Also, we analyzed the presence of exosome-enriched EV markers in one human and one mouse sample, respectively. With this approximation, we wanted to validate that the precipitation-based method enriches exosome-associated EVs from serum samples. From this approximation, we visually identified the presence of tetraspanin 29 (CD9), tetraspanin 28 (CD81), and the tumor susceptibility gene 101 (TSG101) in the human and murine samples; these are commonly used as exosome markers for their characterization ( Figure 3 A,G). Furthermore, in the exosome protein marker array, calnexin (CANX) and the intracellular adhesion molecule 1 (ICAM1) were barely visible or undetected. The low abundance or absence of CANX, a control for cellular contamination, suggests that EVs were correctly and effectively isolated using this precipitation-based method; the absence of ICAM1 may suggest low enrichment of EVs with this general EV marker. In addition, we identified the presence of proteins related, but not exclusively, to neural lineage, such as L1 transmembrane protein (L1CAM), neural cell adhesion molecule 1 (NCAM1), enolase 2 (ENO2), and total tau protein (MAPT), and proteins possibly only linked to oligodendrocyte lineage, such as proteolipid protein 1 (PLP), for human exosome-enriched EVs. In the murine exosome-enriched EV samples, we detected low immunoreactivity of NCAM, ENO2, MAPT, and glutamate receptor 1 (GRIA1). However, tetraspanin 30 (CD63) was not correctly detected. Thus, we performed slot blots with exosome-enriched EVs from human CTRL and MDD samples ( Figure 3 B) and murine CTRL and CUMS samples ( Figure 3 H, Supplementary Figure S1 ). We identified CD63 in both species, suggesting that the precipitated samples contain this exosomal marker. Moreover, TEM analysis revealed the presence of small EVs, ranging from 30 to 180 nm, in both the human and murine samples ( Figure 3 C,I). In the negative staining method, these EVs corresponded to the size described for small EV-like exosomes. The exosome-enriched EVs of both species were similar. Taken together, these results suggest that the isolated EVs are exosome-enriched EVs. Quantification of total protein from exosome-enriched EVs isolated from both human groups (CTRL and MDD) and murine blood serum (CTRL and CUMS) was performed for each experimental group ( Figure 3 D,J). We found a 14.62% significant decrease in protein content in the MDD group compared to the CTRL group (3.782 μg/μL ± 0.06120 vs. 3.229 μg/μL ± 0.02994; t = 8.975, df = 18, p < 0.0001, respectively; Figure 3 D). In the murine exosome-enriched EVs samples, a 10.40% significant decrease in protein content was also found for the group of mice exposed to the CUMS protocol compared to the CTRL mice (2.712 μg/μL ± 0.04521 vs. 3.027 μg/μL ± 0.06536; t = 3.962, df = 18, p = 0.0009, respectively; Figure 3 J). In addition, the relative abundance of CD63-positive EVs was determined by ELISA from both human groups (CTRL and MDD) and murine blood serum (CTRL and CUMS) samples of isolated exosome-enriched EVs ( Figure 3 E,K). For the human samples, a 44.30% significant increase in the relative abundance of CD63-positive EVs (particles/mL) was identified in the MDD group compared to the CTRL (t = 2.085, df = 22, p = 0.0489, mean ± SEM, 1.07 × 10 10 particles/mL ± 1.39 × 10 9 vs. 7.47 × 10 9 particles/mL ± 7.55 × 10 8 ; respectively; Figure 3 E), and in the murine samples, no significant differences were found between the CTRL and CUMS groups (t = 0.5693, df = 14, p = 0.5693; mean ± SEM, 1.75 × 10 10 particles/mL ± 3.52 × 10 9 vs. 1.45 × 10 10 particles/mL ± 3.75 × 10 9 ; respectively; Figure 3 K). Finally, analysis of the relative abundance of the CD63-positive EVs/total protein ratio showed significant differences between the CTRL and MDD groups (t = 3.066, df = 18, p = 0.0067; mean ± SEM, 1,701,481 particles/protein ± 224,742 vs. 3,325,148 particles/protein ± 402,686; respectively; Figure 3 F), but not in the murine samples (t = 0.1206, df = 14, p = 0.9057; mean ± SEM, 5,776,901 particles/protein ± 1,224,055 vs. 5,539,874 particles/protein ± 1,537,386; respectively; Figure 3 L). In the case of the CTRL and MDD groups, an inverse result was seen when we analyzed the total protein/relative abundance of CD63-positive EVs (t = 3.139; df = 18, p = 0.0057; mean ± SEM, 7.115 × 10 −7 protein/particles ± 1.428 × 10 −7 vs. 3.348 × 10 −7 protein/particles ± 2.785 × 10 −8 , respectively; Supplementary Figure S2 ). Optical density analysis of 120 human and 96 murine proteins ( Figure 4 A,C,D) revealed differential expression patterns between the experimental groups. In the human samples, 36 proteins showed statistically significant differences between the CTRL and MDD groups ( Table 3 , Figure 4 B). In the murine dataset, 19 proteins were significantly differentially expressed between the CTRL and CUMS groups ( Table 4 , Figure 4 B). To complement the findings derived from the sets of significantly altered proteins across the human and murine datasets, volcano plots were generated as a graphical representation integrating fold change (MDD/CTRL and CUMS/CTRL) and statistical significance (−log10 p -value) to identify differentially expressed proteins across experimental conditions ( Figure 5 ). This approach enables rapid visualization of proteins exhibiting both biologically relevant changes and statistical significance, thereby facilitating the identification of potential molecular candidates ( Table 3 , Table 4 and Table 5 ). Nineteen proteins were identified as significantly downregulated in the human dataset, including IGF-BP-6, BDNF, BMP-6, EGF-R, CK-beta-8-1, sgp130, BLC, BMP4, FAS/TNFRSF6, ENA-78, CNTF, fractalkine, GDNF, sTNFRII, ICAM-1, Flt-3 ligand, eotaxin, Axl, and BTC, while two proteins (MIP-1α and G-CSF) were significantly upregulated ( Figure 5 A). In the murine dataset ( Figure 5 B), seventeen proteins were found to be significantly downregulated, including SCF, CRG-2, VEGF, CD30L, BLC, GM-CSF, fractalkine, leptin receptor, IL-13Rβ, IL-2, IL-9, IL-4, CTACK, IGF-BP-5, MIG, eotaxin, and leptin, whereas only one protein, osteopontin, was significantly upregulated. All significantly altered proteins identified in both datasets were part of the original sets of differentially expressed proteins determined for each species ( Table 3 and Table 4 ). Comparative analysis of the 52 proteins shared between the human and murine arrays identified 12 proteins with significant interspecies differences ( Table 5 , Figure 4 B). Among these, 12 and 10 proteins were differentially expressed in humans and mice, respectively ( Figure 4 B). Notably, B lymphocyte chemoattractant (BLC), fractalkine (CX3CL1) and IGFBP6, among others, were significantly altered in both species ( Table 3 and Table 4 ), showing downregulated expression in both the MDD patients and CUMS-exposed mice ( Figure 5 ). These findings suggest a possible conserved role for EV-associated proteins in the context of major depressive disorder and depressive-like behavior. To explore the biological relevance of these alterations, protein–protein interaction networks were analyzed using the STRING database. Functional enrichment analyses were performed using Gene Ontology (GO) and KEGG pathway annotations via WebGestalt. This integrative approach enabled the identification of enriched biological pathways, molecular functions, and interaction networks associated with the cytokine cargo of serum-derived exosome-enriched EVs from the healthy control participants and individuals with MDD, as well as from the control mice and the mice exhibiting depressive-like behavior following exposure to the CUMS protocol. Interspecies comparisons further revealed both shared and distinct EV-associated cytokine signatures between human and murine conditions ( Figure 6 , Figure 7 and Figure 8 ; Table 3 , Table 4 and Table 5 ). PPI analysis of the set of 36 human exosome proteins (PPI enrichment p -value: <1.0 × 10 −16 ; 0. 700 highest confidence value) and clustering ( Figure 6 A, Table 6 ) identified two subsets or clusters of proteins associated with different signaling pathways, where cluster 1 (red) and cluster 2 (green) integrated 14 proteins in the ERBB2-EGFR signaling pathway, growth factor, and inhibition of signaling by overexpressed EGFR pathways and, on the other hand, 10 proteins in the chemokine-mediated signaling pathway, viral protein interaction with cytokines and cytokine receptor and chemokine receptor-bound chemokines, respectively. Interestingly, significant protein–protein interaction between both clusters was identified: GDNF, EGF, EGFR, ICAM-1, and NGF in cluster 1, and CCL11, CCL3, CNTF, CX3CL1, CXCL13, CSF3, FLT3LG, IL6ST, and THPO in cluster 2 (dotted line, Figure 5 A). This also allowed us to associate them with other processes and signaling pathways located in the cluster network (graph Local network, STRING enrichment, Figure 6 A). In addition, GO analysis of the associated biological processes identified the participation of this set of proteins in the categories of stimulus response, biological regulation, and cellular communication for the cytokine-mediated signaling pathway, positive regulation of protein kinase B signaling, ERBB2-EGFR signaling pathway, and chemokine-mediated signaling pathway, mainly ( Figure 6 B, biological process, GO enrichment graph). Most cellular component categories exhibit activity at the cell surface, in the extracellular space, at the membrane, at the membrane of clathrin-coated endocytic vesicles, and in membrane rafts (as indicated by the bar graph and cellular component enrichment graph in Figure 6 B). In the case of molecular function categories, the main ones are protein binding and molecular transducing activity associated with cytokine activity, growth factor activity, receptor–ligand binding, and cytokine, chemokine, and chemoattractant binding, as well as CCR chemokine receptor binding (bar graph and graphic molecular function enrichment, Figure 6 B). Finally, the biological signaling pathways associated with this set of proteins, according to their enrichment ratio (FDR ≤ 0. 05), were viral protein interaction with cytokine and cytokine receptor, cytokine–cytokine receptor interaction, EGFR tyrosine kinase inhibitor resistance, TNF signaling pathway, MAPK signaling pathway, Ras signaling pathway, PI3K-Akt signaling pathway, JAK-STAT signaling pathway, and chemokine signaling pathway, among others ( Figure 6 C). According to the PPI enrichment analysis of the 19 proteins with significant differences from the murine exosomes (PPI enrichment p -value: <1.0 × 10 −16 ; 0.700, the highest confidence value), a leading protein cluster related to chemokine receptors, which binds chemokines and interacts with viral proteins and cytokines, was identified ( Figure 7 A, Table 4 , Table 7 ). Local network (STRING) enrichment analysis confirmed interactions with chemokine ligands, receptors, the JAK-STAT signaling pathway, the interleukin-2 family signaling pathway, and the tumor necrosis factor domain. Cluster analysis showed no interactions between protein subsets and only one big set of proteins (red cluster, Figure 7 A). In the GO analysis of biological processes, cell communication, response to stimuli, and biological regulation were identified as the main categories, particularly associated with chemotaxis and migration processes of granulocytes, leukocytes, neutrophils, and lymphocytes, as well as chemokine- and cytokine-mediated signaling pathways (biological process, GO enrichment analysis, Figure 7 A). The main cellular components associated with this set of proteins included the extracellular space and the cell surface (outer side of the plasma membrane) ( Figure 7 B, cellular component, GO enrichment analysis). The associated molecular functions included protein binding, primarily related to cytokine and chemokine activity, as well as cytokine and chemokine receptor binding (CXCR, CCR), growth factor and chemoattractant activity ( Figure 7 B, molecular function, GO enrichment analysis). Interestingly, the main associated biological signaling pathways were viral protein interaction with cytokines and cytokine receptors, cytokine–cytokine receptor interaction, chemokine, IL-17, and JAK-STAT signaling pathways, according to their enrichment ratios and FDR ≤ 0.05 ( Figure 7 C, Table 7 ). Finally, for the set of 12 proteins comparable between human and mouse species, PPI enrichment analysis (PPI enrichment p -value: <1.11 × 10 −16 ; 0, 700 highest confidence value) identified two small clusters of four and two proteins (red and green, Figure 8 A, Table 8 ) primarily associated with lymphocyte chemotaxis. Although there was interaction with a small cluster (two proteins) related to insulin-like growth factor signaling (pathway/binding complex), IL15 also interacted with this cluster. Most pathways and GO terms in the local network cluster (STRING) enrichment analysis showed strong associations with pathways such as chemokine receptor binding, chemokines, insulin-like growth factor-binding protein family 1–6, and the tissue inhibitor of metalloproteinase family, among others ( Figure 8 A). In the GO analysis, the biological process categories were primarily associated with leukocyte and lymphocyte migration/proliferation, as well as positive regulation of tyrosine phosphorylation of the STAT protein signaling pathway (GO analysis, Figure 8 B). The associated cellular components were mainly located in the extracellular space and membrane. They may be associated with the insulin-like growth factor-binding protein complex (Cellular component, GO enrichment analysis, Figure 8 B). As previously described, the associated molecular functions will also be primarily related to cytokine activity, receptor–ligand activity, CCR chemokine receptor binding, chemoattractant activity, and growth factor activity (molecular function, GO enrichment analysis, Figure 8 B). Finally, the primary biological signaling pathways enriched for this set of proteins, according to FDR, include TNF signaling, viral protein interactions with cytokines and cytokine receptors, cytokine–cytokine receptor interactions, and chemokine signaling pathways. However, other interesting pathways include Toll-like receptor signaling, the JAK-STAT pathway, Rap1 signaling, and Ras signaling ( Figure 8 C).

In this study, we explored the protein signature of exosomes derived from the blood serum of participants diagnosed with MDD and from mice exposed to the CUMS protocol, a widely used model to mimic depression-like behaviors, with the aim of obtaining translational insights into MDD from an animal model of chronic stress. MDD is a multifactorial disorder with severe and heterogeneous manifestations across patients, and its high prevalence makes it a major global health concern [ 80 , 81 ]. Although the CUMS model is widely used to study depression-like behaviors in rodents [ 15 , 82 ], the degree of molecular and proteomic homology between clinical depression and depression-like behavior remains unclear. Therefore, as with any in vivo model, its validity is limited to specific behavioral features and their association with selected biochemical and cellular changes at the CNS level [ 23 , 83 ]. In this context, the study of the proteomic content of exosomes in both clinical and murine settings may provide novel insights into the molecular and cellular mechanisms underlying this disorder, their relationship with symptom manifestation and severity, and the development and monitoring of more effective and comprehensive antidepressant treatments aimed at achieving remission and improving patients’ quality of life. Notably, our findings reveal that cytokine and growth factor profiles associated with EVs are significantly altered in both conditions, highlighting EVs as potential mediators and indicators of neuroimmune dysregulation in depression. Comparison of signaling pathways derived from enrichment analyses revealed that at least eight pathways are shared between humans and mice in MDD and the CUMS model: chemokine signaling, cytokine–cytokine receptor interaction, JAK–STAT signaling, pathways in cancer, Rap1, Ras, and TNF signaling, and viral protein interaction with cytokines and cytokine receptors. Notably, six of these pathways showed an FDR value < 0.05: chemokine signaling, cytokine–cytokine receptor interaction, JAK–STAT, Ras, and TNF signaling, and viral protein interaction with cytokines and their receptors ( Table 5 , Table 6 and Table 7 ). These findings suggest that these pathways may be relevant to both MDD and depression-like behavior induced by chronic stress. Importantly, one of the most notable findings of this study is the predominance of downregulated cytokines and growth factors in the EV cargo in both MDD-diagnosed participants and CUMS-exposed mice. This pattern contrasts with the classical view of depression as a purely hyper-inflammatory state [ 84 , 85 ]. Instead, our results support a more nuanced model in which MDD is characterized by dysregulation of immune signaling rather than a uniform increase in inflammatory mediators. The observed reduction in molecules such as BDNF, EGFR, CXCL13, and CX3CL1 suggests a disruption of intercellular communication pathways that are critical for maintaining neuroimmune homeostasis [ 86 , 87 , 88 , 89 , 90 ]. In this context, EVs may reflect systemic alterations and potentially contribute to the propagation of dysfunctional signals between peripheral and central compartments, including mechanisms such as blood–brain barrier (BBB) crossing and neuroimmune communication [ 91 , 92 , 93 , 94 ]. The chemokine superfamily (including CXC, CC, C, and CX3C subtypes) has been implicated in processes that may modulate depressive symptomatology, such as immune cell regulation and migration [ 95 , 96 ], blood–brain barrier permeability [ 97 ], and synaptic pruning [ 98 ]. These chemokines are secreted in response to inflammatory cytokines and selectively recruit monocytes, lymphocytes, and neutrophils by inducing chemotaxis through G protein-coupled receptors (GPCRs) [ 99 ]. Chemokines and their receptors are widely expressed in the CNS under both physiological and pathological conditions [ 100 , 101 , 102 ]. Glial cells—including astrocytes, oligodendrocytes, and microglia—as well as neurons, express several chemokines such as CCL2, CCL3, CCL19, CCL21, CXCL10, and CX3CL1 [ 103 , 104 , 105 ], which may be upregulated under pathological conditions such as MDD [ 23 ]. One of the primary roles of chemokines in the CNS is the regulation of neuronal development and plasticity, influencing key processes such as proliferation, migration, and differentiation of neural progenitor cells. For instance, CX3CL1 is released by mature neurons and astrocytes, while its receptor CX3CR1 is predominantly expressed in microglia [ 106 , 107 ], which plays a central role in synaptic pruning, neurotransmitter modulation, and neurogenesis [ 108 , 109 ]. These processes may be critically involved in neuroplasticity alterations observed in depressive states. Regarding cytokine–cytokine receptor interactions and TNF signaling pathways, elevated levels of circulating cytokines and their soluble receptors have been consistently reported in MDD patients and animal models [ 110 ]. For example, increased plasma levels of IL-6, IL-1β, TNF-α, and C-reactive protein (CRP) have been described. Some cytokines, including IL-6, IL-2, IL-1β, and TNF-α, have been proposed as biomarkers of depressive states or stress, and certain antidepressant treatments have been shown to reduce their levels [ 111 , 112 , 113 ]. Their effects in depression and stress depend on the specific pathways they regulate; notably, inflammasome activation has been proposed as a key mechanism [ 114 , 115 ]. This pathway responds to elevated levels of damage-associated molecular patterns (DAMPs), triggered by stimuli such as hormonal imbalance and oxidative stress, leading to activation of IL-1β and IL-18, peripheral immune activation, and increased blood–brain barrier permeability. These processes may contribute to astrocytic dysfunction and neuro-inflammation, including microglial activation, which, in turn, promotes the production of chemokines such as CXCL7 and CXCL8. These molecules have been identified in both animal models and human studies and are involved in cascades that modulate neuronal plasticity and neurogenesis [ 116 , 117 ]. Additionally, the interaction between the JAK/STAT and Ras signaling pathways has been associated with depressive disorders and depression-like behaviors in both clinical and preclinical studies [ 118 , 119 , 120 ]. These pathways often interact with others, such as ERK and MAPK, which were also identified in our enrichment analyses. For example, the Ras/MAPK pathway plays an important role in neuronal plasticity, oxidative stress, cell growth, differentiation, and apoptosis [ 121 , 122 ]. Among these, the JAK/STAT pathway is particularly relevant in neurological disorders due to its direct involvement in inflammatory processes and its distribution across brain regions such as the cortex and hippocampus [ 123 , 124 , 125 ]. Furthermore, this pathway plays a crucial role in cell growth, survival, development, differentiation, and gene regulation [ 126 ]. As a downstream effector of cytokines, growth factors, and reactive oxygen species (ROS), its dysregulation represents a key factor in neurodegenerative and neuropsychiatric conditions [ 124 , 125 ]. Collectively, these findings suggest that protein sets identified in both humans and mice are associated with signaling pathways that may modulate depression and stress [ 127 , 128 ], providing some translational insights and highlighting some similarities between the CUMS model and clinical observations. Although differences in protein expression levels exist, overlapping associations were observed between 36 human and 18 murine proteins involved in these pathways ( Figure 5 and Figure 6 ). Notably, a subset of proteins ( Table 7 , Figure 8 ), including CXCL13 (BLC) and CX3CL1 (fractalkine), was consistently downregulated in both species, suggesting conserved mechanisms of neuroimmune communication in response to chronic stress and depressive states. CX3CL1–CX3CR1 signaling plays a critical role in neuron–microglia communication, including microglial activation and synaptic plasticity [ 129 , 130 ]. In contrast, although CXCL13 (BLC) is well known for its role in immune responses and has been associated with complex diseases such as cancer [ 131 , 132 ], its role in psychiatric disorders remains poorly understood. Together, these findings support the hypothesis that alterations in immune signaling and neuro-inflammatory communication are central features of depression [ 120 , 127 ]. Conversely, several proteins displayed species-dependent or even opposite regulation patterns, including ICAM-1, IL-15, MIP-1α, and SCF. These discrepancies highlight important limitations in the direct translation of animal model findings to human conditions. While the CUMS model reproduces key behavioral features and some molecular aspects of depression, our results suggest that systemic immune responses to chronic stress may differ between species, possibly due to differences in physiology, immune system organization, or stress perception. Another important aspect of this study is the exclusive inclusion of female subjects in both human and animal cohorts. Although this approach prevents the exploration of sex differences, it introduces important biological considerations related to hormonal status. Significant differences were observed in FSH levels and menstrual cycle phase, with a predominance of the follicular phase in the MDD group and the luteal phase in controls. Although the differences were not statistically significant in the latter parameter, they suggest that hormonal fluctuations may influence EV cargo composition [ 133 , 134 ]. Estrogen and progesterone are known to modulate immune responses and hypothalamic–pituitary–adrenal (HPA) axis activity [ 135 , 136 ]. The follicular phase, characterized by lower progesterone levels, has been associated with increased stress sensitivity and altered inflammatory responses [ 137 ], whereas the luteal phase may exert stronger immunomodulatory effects [ 138 ]. Therefore, the observed EV-associated cytokine profiles may be partially influenced by hormonal status rather than exclusively reflecting disease-related changes. This represents an important factor to consider in data interpretation. Future studies with larger sample sizes and stricter control of menstrual/estrous cycles and hormonal levels will be essential to disentangle endocrine and pathological contributions. From a methodological perspective, it is important to note that EV isolation was performed using a precipitation-based approach, which enriches exosome-like vesicles but may also co-isolate non-vesicular components, such as protein aggregates and lipoproteins. In addition, the use of antibody-based arrays provides a semi-quantitative assessment of protein expression, which may limit sensitivity and dynamic range compared to quantitative techniques such as ELISA or mass spectrometry. On the other hand, the differences in the results of EV array markers and slot blot for the identification of CD63 may suggest differences in antibody specificity between the array marker and slot blot, as well as in the use of lysed (array marker) or non-lysed (slot blot) exosome-enriched EVs. Therefore, further validation using orthogonal approaches will be necessary to confirm these findings. Despite these limitations, we consider that our study provides evidence supporting the role of EVs as carriers of biologically relevant information in MDD and chronic stress modeled in mice. The identification of conserved and species-specific EV signatures highlights the complexity of neuroimmune interactions in depression and underscores the importance of integrative approaches combining clinical and experimental models [ 17 , 139 ]. Furthermore, the association of EV cargo with key pathways involved in inflammation [ 22 , 134 , 140 ], cell communication [ 24 , 37 ], and growth factor signaling [ 24 , 25 ] suggests that EVs may function not only as a source of potential biomarkers but also as active participants in the pathophysiology of depression [ 23 , 26 , 38 , 39 ].

Finally, our findings support the relevance of the protein cargo of serum-derived exosomes from both humans and mice in the context of depression and depression-like behavior, representing an important—although not exclusive—source of information for understanding the molecular mechanisms underlying depressive disorders. Our results suggest that both MDD and chronic stress are associated with a dysregulation of EV-mediated intercellular communication, particularly in pathways related to the immune system and the CNS. This dysregulation is further characterized by a decrease in the expression of cytokines and growth factors, suggesting coordinated alterations in specific signaling networks rather than a simple inflammatory activation. The similarities observed in the bioinformatic analyses indicate partial homology between human and murine expression profiles associated with inflammatory and neuro-inflammatory responses, potentially affecting specific brain regions and cellular substrates within the CNS. This partial overlap between human and murine EV profiles could support the partial translational value of the CUMS model. Taken together, these findings contribute to a more comprehensive understanding of the molecular mechanisms underlying depression and chronic stress and support the potential of EVs as promising biomarkers in neuropsychiatric disorders.