Analysis of gut microbiota and intestinal mucosal neurotransmitter changes and their correlation in adolescent depression mice Authors: Jinrui Xing, Ziwei Jiang, Xushu Jing, Ying Li, Fengtao Guo, Penghong Liu, Zhifen Liu, Ning Sun Published: 2026 ABSTRACT AbstractBackgroundAdolescent depression is a major mental health disorder with increasing prevalence and substantial long-term consequences. Although growing evidence suggests that the gut-brain axis is involved in depression, the relationships among gut microbiota, intestinal mucosal neurotransmitters, and adolescent depression remain insufficiently understood. This knowledge gap limits a better understanding of the pathophysiological mechanisms underlying adolescent depression and the identification of potential microbiota-related targets. Therefore, this study aimed to investigate alterations in gut microbiota and intestinal mucosal neurotransmitters, as well as their correlations, in an adolescent mouse model of depression.MethodsWe established an adolescent depression mouse model using chronic unpredictable mild stress (CUMS), and collected data with the Smart video tracking system. We collected intestinal contents and mucosal tissues from mice. We analyzed gut microbial composition using metagenomic sequencing and quantified mucosal neurotransmitters with liquid chromatography–tandem mass spectrometry (LC–MS/MS). We analyzed correlations among gut microbiota, intestinal mucosal neurotransmitters, and behavioral indicators.ResultsMice in the CUMS group exhibited a significantly reduced sucrose preference rate in the sucrose preference test (P < 0.001); a significantly prolonged immobility time in the forced swim test (P < 0.01); and a significantly decreased total movement distance in the open field test (P < 0.01). No significant intergroup difference was observed in the tail suspension test. Regarding the gut microbiome, the CUMS group showed significantly lower Simpson index (P = 0.018) and Pielou’s evenness index (P = 0.022). Beta diversity analysis indicated a statistically significant but modest between-group difference in community structure (ANOSIM R = 0.145, P = 0.03); this finding was supported by PERMANOVA (Bray–Curtis; pseudo-F = 1.675, R² = 0.0897, P = 0.033). LEfSe (Linear discriminant analysis Effect Size) analysis suggested 27 candidate taxa with discriminatory signals between groups (nominal P < 0.05; exploratory). Neurotransmitter analysis demonstrated that levels of 5-HIAA (5-hydroxyindoleacetic acid), 5-HT (serotonin), 5-HTP (5-hydroxytryptophan), and Kyn (kynurenine) in the colon were significantly decreased in the CUMS group, whereas levels of PA (phenylethylamine) and NE (norepinephrine) were significantly elevated (P < 0.05). Spearman correlation analysis found that Lactobacillus and Lactobacillus acidophilus correlated positively with sucrose preference and negatively with immobility in the forced swim test. Lactobacillus acidophilus also showed a positive correlation with 5-HT pathway metabolites: 5-HIAA, 5-HT, 5-HTP, and Kyn.ConclusionAdolescent mice exposed to CUMS showed depression-relevant behavioral alterations, shifts in gut microbiota composition, and changes in 5-HT pathway metabolites. Gut microbiota dysbiosis was significantly associated with alterations in 5-HT pathway metabolites. Because this study is correlational, causal relationships require validation in future interventional studies.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12991-026-00686-x.Keywords: Adolescent CUMS-induced depression model mice, Gut microbiota, Intestinal mucosal neurotransmitters Background Major depressive disorder (MDD) in adolescents is a common, complex disorder. Its clinical features significantly lower patients’ quality of life and place a major socioeconomic burden on families and society [1, 2]. In recent years, both the rate and severity of MDD among adolescents have increased markedly [3]. Although adolescent depression is more prevalent, effective treatments remain scarce. Most antidepressant drugs work less well in adolescents than in adults, often leading to poor remission rates. This suggests that adolescent and adult depression differ in important neuropathophysiological ways [4]. Understanding the pathophysiology of adolescent MDD is crucial for early identification and precise intervention. Recently, research highlights the brain–gut–microbiome axis in MDD. Clinical studies show clear differences in gut microbiota between MDD patients and healthy controls (HCs) [5, 6]. Nikolova et al. found sequencing gut microbiota can distinguish MDD patients from HCs [7]. Another study showed adolescent MDD patients have lower relative levels of Faecalibacterium, Roseburia, Collinsella, Faecalibacterium prausnitzii, and unclassified Lachnospiraceae. Changes in Roseburia abundance strongly predict adolescent depression [8]. In a study of 171 participants (101 MDD, 70 HCs), the MDD group had less microbial richness and diversity, fewer beneficial bacteria, and more potentially harmful bacteria such as Escherichia–Shigella and Flavonifractor [9]. Animal studies also link adolescent MDD to gut dysbiosis. For example, fecal transplants from healthy adolescent donors to depressed mice (induced by chronic restraint stress) relieved depression-like behavior [8]. These findings suggest microbiota modulation may help intervention. In addition to microbiota-related mechanisms, dysregulation of the hypothalamic–pituitary–adrenal (HPA) axis is a core neuroendocrine feature of depression and stress susceptibility. Acute stress activates corticotropin-releasing hormone (CRH) neurons in the hypothalamic paraventricular nucleus, triggering pituitary adrenocorticotropic hormone (ACTH) release and subsequent glucocorticoid secretion from the adrenal cortex. Under physiological conditions, glucocorticoids provide negative feedback to the hypothalamus, pituitary, and limbic–prefrontal regions to terminate the response; however, chronic stress can impair this feedback, resulting in sustained glucocorticoid exposure and downstream alterations in amygdala–prefrontal circuits that are relevant to mood regulation in depression [10–12]. Importantly, HPA-axis-related dysfunction and maladaptive remodeling of stress-regulatory circuits are observed not only in stress-based depression models but also in models intended to mimic endogenous vulnerability. For example, chronic stress paradigms and their interventions have been shown to reshape prefrontal plasticity and affective behaviors, whereas neonatal clomipramine exposure produces long-lasting affective phenotypes accompanied by pathological changes in the amygdala, highlighting convergent limbic mechanisms across “endogenous” and stress-based depression models [13, 14]. More broadly, pharmacological strategies that restore synaptic plasticity and counteract stress-related neurobiological cascades have been proposed as adjunctive approaches in mood disorders [15]. Neurotransmitters are chemical messengers that transmit information between neurons. They play a pivotal role in regulating higher brain functions such as mood, cognition, and behavior. Key neurotransmitters include 5-HT, norepinephrine, dopamine, γ-aminobutyric acid (GABA), and acetylcholine. Recent studies have shown that gut microbiota can participate in the synthesis and metabolism of various neurotransmitters. For instance, some Lactobacillus and Bifidobacterium can secrete GABA, which helps regulate physiological and psychological processes. GABA also plays a significant role in anxiety and MDD. Many gut microorganisms can synthesize other neurotransmitters as well. For example, Escherichia coli, Bacillus, and Saccharomyces can synthesize norepinephrine. Candida, Streptococcus, Escherichia coli, and Enterococcus can produce 5-HT. Bacillus can generate dopamine, and Lactobacillus can synthesize acetylcholine [16]. Importantly, most classic neurotransmitters do not freely cross the blood–brain barrier; instead, microbial effects on host neurotransmission are thought to be mediated mainly through modulation of host biosynthesis and metabolism in the gut, and through neural, immune, and endocrine signaling pathways within the microbiota–gut–brain axis [17]. Among the mechanisms by which gut microbiota contribute to depression, the serotonin imbalance hypothesis within the neurotransmitter theory has drawn much attention. Serotonin is a key neurotransmitter involved in immune regulation, inflammation, and gastrointestinal functions. It also plays a central role in emotional disorders such as anxiety and depression. About 95% of the body’s serotonin is synthesized by enterochromaffin (EC) cells. EC cells are specialized endocrine cells within the intestinal epithelium. They detect various luminal stimuli and regulate physiological functions through serotonin secretion. The activity and synthetic capacity of EC cells are significantly influenced by the gut microbiota [18]. Another animal study showed that oral administration of Roseburia intestinalis (R. intestinalis) to depressed mice induced by chronic restraint stress (CRS) significantly alleviated depression-like behaviors. The possible mechanism involves promotion of tryptophan hydroxylase-2 (TPH2) and tryptophan hydroxylase-1 (TPH1) expression by R. intestinalis, increasing serotonin levels in the prefrontal cortex and colonic tissues [8]. Gut microbiota can influence host neurobiology by shaping the production and metabolism of neuroactive metabolites and by modulating stress-response circuits, thereby potentially affecting brain function and behavior [19]. Current research on the link between gut microbiota and intestinal mucosal neurotransmitters in adolescent MDD is limited. To address this, we created an adolescent depression mouse model with the CUMS paradigm. We examined changes in gut microbiota and intestinal mucosal neurotransmitter levels to analyze their correlation. Because circulating and brain endpoints were not assessed, the current work primarily provides gut-level associative evidence. These results will help clarify the pathophysiology of adolescent MDD and may inform intervention strategies targeting the microbiota–neurotransmitter axis. Materials and methods Experimental animals A total of 20 male C57BL/6 mice, specific pathogen-free (SPF) grade, 5–6 weeks old (PND35–42), with a body weight of 18 ± 3 g, were supplied by Beijing Vital River Laboratory Animal Technology Co., Ltd. (License No.: SCXK (Jing) 2021-0006). All animals were housed under SPF conditions, subjected to a 12-hour light/dark cycle, with the ambient temperature maintained at 20–26 °C and humidity at 40–70%. Bedding was changed every three days. In rodents, adolescence is commonly operationalized using postnatal age; therefore, we report mouse age in both weeks and postnatal days (PND) throughout the manuscript. Animal grouping and treatment Prior to model establishment, 20 mice were randomly assigned to groups based on baseline body weight (10 per group), resulting in two groups: a normal control group and a CUMS depression model group. During the protocol, one mouse died; therefore, 19 mice were included in the final analyses (control n = 9; CUMS n = 10). The depression model was established using the CUMS protocol. Each day starting at 8:00, mice were subjected to two different randomly selected stressors from the following list: [1] 24-hour food deprivation; [2] 24-hour water deprivation; [3] 3-hour physical restraint; [4] 3-minute tail clamping; [5] 24-hour exposure to a novel object; [6] 24-hour light/dark cycle reversal; [7] 24-hour exposure to damp bedding; [8] 24-hour housing in an unfamiliar cage; [9] 24-hour empty cage stimulation; [10] 24-hour cage tilt at 45°; [11] 3-hour stroboscopic stimulation; [12] 5-minute cold water swim at 4 °C. The same stressor was not applied consecutively, and the total frequency of each stressor did not exceed three times throughout the 4-week period to ensure the unpredictability of the stressors. During this time, mice in the normal control group were maintained under standard housing conditions. Behavioral tests were conducted one day after the completion of the CUMS modeling period. The Smart video tracking system was used to record the movement trajectories of the mice and collect experimental data. Before testing, mice were allowed to acclimate to the testing environment for at least 2 h. The experiments were performed in a quiet environment. The behavioral apparatus was wiped with alcohol after testing each mouse to eliminate residual odors that might affect the results. All behavioral assays were performed in parallel in age-matched mice from both groups at the end of the 4-week CUMS period. Because adolescent mice exhibit distinct baseline activity patterns and stress sensitivities, we did not perform pre-CUMS baseline behavioral testing to avoid repeated test-related stress and learning effects. Instead, mice were randomized by baseline body weight and all outcomes were interpreted relative to the age-matched control group tested under identical conditions. Successful establishment of the adolescent depression-like phenotype was defined a priori. The primary criterion was a significant reduction in sucrose preference in the sucrose preference test (SPT) compared with the control group, indicating an anhedonia-like state. In addition, mice were required to exhibit at least one supportive depressive-like behavioral alteration, including increased immobility time in the forced swim test (FST) or decreased locomotor activity (total distance traveled) in the open field test (OFT). Changes in general condition (e.g., reduced food/water intake, dull/rough coat, reduced spontaneous activity) and reduced body-weight gain were recorded as supportive indicators. Behavioral tests Monitoring of general condition and body weight After the initiation of the experiment, the mice were observed daily for water intake, food consumption, mental state, coat color, and defecation. Body weight was measured once per week. Sucrose preference test (SPT) The SPT was widely used to assess the level of anhedonia, a core symptom of depression. Mice underwent a 3-day sucrose preference adaptation training. On day 1, two bottles of 1% sucrose solution were provided for 24 h. On day 2, one bottle of 1% sucrose solution and one bottle of pure water were provided for 24 h, with the positions of the bottles switched midway. On day 3, mice were subjected to water deprivation but not food deprivation for 12 h. Subsequently, the sucrose preference test was conducted for 15 h, during which two bottles containing equal volumes of 1% sucrose solution and pure water, respectively, were provided. The consumption of water and sucrose solution was measured by weighing the bottles before and after presentation to the mice. The positions of the two bottles were counterbalanced across groups and switched midway to avoid potential side preference bias [20]. The sucrose preference rate was calculated using the following formula: Sucrose Preference Rate = (Sucrose Solution Consumption)/(Water Consumption + Sucrose Solution Consumption) × 100%. Forced swimming test (FST) The FST involves placing an animal in a confined space filled with water from which it cannot escape. After a period of struggling, the animal typically ceases active escape-oriented behavior and adopts an immobile posture, a phenomenon termed “behavioral despair”, which is commonly used to assess depressive-like behavior in animals. During the test, each mouse was individually placed into a transparent cylindrical water tank maintained at room temperature (24 °C ± 1 °C) for a total of 6 min. The immobility time during the last 4 min was recorded. The water was changed after testing each mouse. Tested animals were dried with clean paper towels and placed under a heat source (37 °C) to restore body temperature. A longer immobility time indicates a greater degree of depressive-like behavior [21]. Tail suspension test (TST) The TST involves suspending an animal by its tail in an uncontrollable position, inducing a “behavioral despair state” characterized by a typical “immobile posture”. It is widely used for the preliminary screening of antidepressant drugs. Prior to the test, medical adhesive tape was applied to the distal one-third of the mouse’s tail. The mouse was then suspended inside the tail suspension apparatus, and its behavior was recorded for 6 min. The immobility time during the last 4 min was recorded. A longer immobility time indicates a greater degree of depressive-like behavior [22]. Open field test (OFT) The OFT is based on the natural exploratory behavior of animals in a novel environment and their innate fear of open spaces. It is used to evaluate locomotor activity, exploratory behavior, and anxiety levels in experimental animals. The open field arena is divided into a central zone and a peripheral zone. At the beginning of the test, each mouse was placed in the center of the arena facing away from the experimenter. The total distance traveled, total movement time, immobility time, and movement speed over a 10-minute period were recorded. After each trial, the arena was thoroughly cleaned with ethanol to eliminate olfactory cues left by the previous animal [22]. Experimental methods Collection of mouse intestinal contents Following deep anesthesia with isoflurane, the abdominal cavity was opened and the entire intestine was excised and placed on a pre-cooled ice tray. The intestine was longitudinally opened, and luminal contents (fecal pellets) were collected first using sterile instruments and transferred into pre-cooled cryotubes. Samples were immediately snap-frozen in liquid nitrogen and stored at − 80 °C until metagenomic sequencing. Collection of mouse intestinal mucosal tissues After removal of luminal contents, the intestine was gently rinsed with pre-cooled (4 °C) physiological saline to remove residual contents. On an ice tray, mesenteric fat and blood vessels were carefully removed. The intestinal mucosal tissues were cut into ~ 0.5 cm segments, blotted dry with filter paper, snap-frozen in liquid nitrogen within 5 s, and stored at − 80 °C for subsequent LC–MS/MS analysis. Separate sterile instruments were used for luminal and mucosal sampling to minimize cross-contamination, and all procedures were performed on ice. Metagenomic sequencing of gut microbiota Genomic DNA extraction and quality control Samples were retrieved from the freezer, and approximately 0.2–0.5 g of each sample was transferred into a centrifuge tube containing lysis buffer. The samples were homogenized using a Tissuelyser-48 tissue homogenizer (Shanghai Jingxin Industrial Co., Ltd.) at 60 Hz. Genomic DNA was extracted from the pre-processed samples using the MagBeads FastDNA Kit for Soil (116564384; MP Biomedicals, CA, USA). DNA concentration was measured using a Qubit 4 Fluorometer (Invitrogen, USA), and DNA quality was assessed by 1% agarose gel electrophoresis. DNA concentrations were adjusted as needed before downstream processing. To monitor potential contamination and technical variation, negative controls and a positive control were processed in parallel with all study samples. For each DNA extraction batch, an extraction blank containing lysis buffer only was carried through the complete extraction procedure without fecal input. For each library preparation batch, a library blank containing molecular-grade water was processed through the full library preparation workflow without template DNA. In addition, a commercially available mock microbial community was processed using the same workflow to assess overall pipeline performance. Negative-control libraries were sequenced alongside biological samples, and taxa consistently enriched in blank controls were considered potential contaminants and were excluded from downstream analyses or interpreted with caution. Sample handling steps before and after PCR were conducted in physically separated areas, and aerosol-resistant filter tips were used throughout to minimize cross-contamination. Library preparation and sequencing Metagenomic shotgun sequencing libraries with an insert size of approximately 400 bp were constructed using the Illumina TruSeq Nano DNA LT Library Preparation Kit. The libraries were sequenced on an Illumina NovaSeq platform in paired-end 150-bp mode (PE150) at Personal Biotechnology Co., Ltd. (Shanghai, China). For shotgun metagenomic sequencing, sequencing depth was evaluated at the sample level by sequencing output rather than by coverage of an individual microorganism. Raw sequencing reads were subjected to quality control prior to downstream analyses. Bioinformatic processing and annotation Raw sequencing reads were quality-filtered using fastp (v0.23.2) to remove adapter sequences, low-quality bases, and short reads, using the following parameters: “-l 50 -g -W 5 − 3 −5 -q 20 -u 30”. Quality-filtered reads were taxonomically classified using Kaiju (v1.9.0) [23] in Greedy mode, with a maximum of 5 allowed mismatches, an E-value threshold of 1 × 10^−5, and low-complexity filtering enabled. Taxonomic abundance tables at multiple taxonomic levels were generated for downstream analyses. Alpha-diversity indices, including Chao1, Observed species, Shannon, Simpson, Pielou’s evenness, and Good’s coverage, were calculated based on the metagenomic taxonomic profiles. For alpha-diversity analysis, taxonomic profiles were rarefied to the lowest sequencing depth across samples before index calculation to reduce bias introduced by differences in sequencing output. Beta-diversity was evaluated using Bray-Curtis dissimilarity and visualized by non-metric multidimensional scaling (NMDS). Group differences in microbial community composition were assessed using ANOSIM and PERMANOVA. Differentially abundant taxa between groups were identified using LEfSe. Detection of intestinal mucosal neurotransmitters Standard solution preparation Weigh an appropriate amount of neurotransmitter standard and prepare a single standard mother solution with methanol or water. Measure an appropriate amount of each mother liquor to prepare a mixed standard, and dilute it with 2% formic acid methanol-water 1:4 to an appropriate concentration to prepare a working standard solution. The mother liquor and working standard solution were kept refrigerated. Weigh an appropriate amount of the isotope standard (Trp-d5), and prepare the internal standard mother solution with water to a concentration of 1000 ng/mL. Metabolite extraction Appropriate sample was taken into 2 mL centrifuge tube, a steel ball was added, and 100 µL 2% formic acid methanol solution was accurately added. Put into the tissue grinder, grind at 55 Hz for 60 s, then accurately add 400 µL water solution, put into the tissue grinder, grind at 55 Hz for 60 s. Centrifuge at 12,000 rpm for 10 min at 4℃, the supernatant was passed through 0.22 μm filter membrane. Take 100 µL of supernatant, add 20 ng/mL of Trp-d5 internal standard 100 µL, swirl well, and add it into the test bottle as a low concentration substance detection sample. Take appropriate amount of supernatant and add 2% formic acid methanol solution - water (1:4, v/v) to dilute for 200 times, and swirl well. Take diluent 100 µL, add 20 ng/mL of Trp-d5 internal standard 100 µL, swirl well, and add it into the test bottle as a sample for high concentration substance detection [24]. The calculation formula are as follows: The content of sample = C*0.5*2/Amount. The content of sample diluted 200 times = C*0.5*2*200/Amount. The unit of content: µg/g. The unit of C: ng/mL. The unit of amount: mg. Instrumental analysis Targeted quantification was performed on an EXion LC system coupled to an AB 6500 Plus triple quadrupole mass spectrometer (AB Sciex, USA). Liquid chromatography conditions: ACQUITY UPLC®HSS T3 column (2.1 × 150 mm, 1.8 μm, Waters, USA), the injection volume was 5 µL, the column temperature was 40 °C, the mobile phase A-Water (with 0.1% formic acid), B-Methanol (containing 0.1% formic acid). The gradient elution conditions were 0 ~ 2 min, 5%B; 2 ~ 6 min, 5 ~ 50% B; 6 ~ 7 min, 50 ~ 95% B; 7 ~ 10 min, 95% B; 10 ~ 10.1 min, 95 ~ 5% B; 10.1 ~ 12 min, 5% B; flow rate: 0.25 mL/min. Mass spectrum conditions: Electrospray ionization (ESI) source, positive and negative ion ionization mode. The ion source temperature was 500 °C, the ion source voltage was 4500 V, the collision gas was 6 psi, the curtain gas was 30 psi, and the atomizing gas and auxiliary gas were both 50 psi.Scans were performed using multiple reaction monitoring (MRM) [25, 26]. Calibration and QC design Calibration was performed using a 12-point external standard series (cal1 to cal12) prepared from the mixed working standard solution by serial dilution in 2% formic acid methanol–water (1:4, v/v). Linearity was assessed by linear regression, and all analytes showed correlation coefficients greater than 0.99; analyte-specific linear ranges and limits of quantification were determined accordingly. A pooled quality control (QC) sample was prepared by mixing pooled study supernatants with the mid-level calibrator (cal5) at a 1:1 volume ratio. QC injections were performed at regular intervals throughout each analytical batch to monitor system stability and analytical repeatability. QC stability was evaluated using relative standard deviation (RSD) calculated from standard peak areas, and all targeted compounds showed QC RSD below 15%. Method performance verification Multiple reaction monitoring transitions and compound-dependent parameters were optimized for each analyte. Batch acceptance was assessed using the pooled QC results; when QC criteria were not met, affected injections were reviewed and repeated when sample material was available. Statistical analysis In this study, statistical analysis was conducted using GraphPad Prism software (version 10.0). Independent sample T-tests or Mann–Whitney U-tests were used to assess statistical significance. The Spearman correlation analysis was employed to assess the correlations between different gut microbiota and neurotransmitters. A P value of less than 0.05 was considered statistically significant. For weekly body-weight measurements, between-group comparisons at each time point were conducted using Welch’s unpaired t-test, with Holm–Šidák correction for multiple comparisons across time points. Unless otherwise stated, microbiome diversity metrics, LEfSe differential abundance testing, and Spearman correlation analyses report unadjusted (nominal) P-values without multiple-comparison correction. In addition, taxonomic profiles derived from shotgun metagenomics are compositional, and LEfSe does not explicitly model compositionality, library-size effects, or false discovery rate. Therefore, LEfSe-identified taxa are presented as hypothesis-generating candidates. Results General condition and behavioral changes in the adolescent depression animal model Prior to model establishment, all adolescent mice exhibited normal drinking and feeding behaviors, were mentally active, had clean and glossy fur, and displayed regular defecation patterns. After 4 weeks of CUMS intervention, mice in the CUMS group showed marked changes in overall appearance and behavior, including reduced food and water intake, mental lethargy, decreased spontaneous activity, dull and rough fur, and slowed behavioral responses. At baseline, no statistically significant difference in body weight was observed between the CUMS and control groups (P > 0.05). After 4 weeks of CUMS exposure, body weight gain in the CUMS group was significantly lower than that in the control group (Fig. 1). Body weight was recorded weekly from baseline (day 0) to day 28. To improve interpretability of between-group differences, individual data points were overlaid on the group mean ± SEM in Fig. 1a. Between-group comparisons were performed at each time point using Welch’s unpaired t-test, and Holm–Šidák correction was applied to adjust for multiple comparisons across time points. Adjusted P-values are provided in Supplementary Table S1. In the final week of modeling, body weight gain in the CUMS group was minimal, whereas the control group continued to gain weight. Suggesting that CUMS exposure attenuated normal body-weight gain in adolescent mice. Because additional developmental endpoints were not assessed, we interpret this change as reduced somatic growth rather than definitive impairment of overall development. Body weight changes during the CUMS procedure. (a) Weekly body weight (g) from baseline (day 0) to day 28 for Control (n = 9) and CUMS (n = 10) shown as mean ± SEM with individual mice overlaid. Between-group comparisons at each time point were performed using Welch’s unpaired t-test with Holm–Šidák correction for multiple comparisons across time points. (b) Body weight time course plotted as mean ± SEM. Significance: ns, P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001 The impact of CUMS on the behavioral performance of adolescent mice was evaluated using a series of behavioral tests. Compared to the normal control group, the sucrose preference rate in the sucrose preference test was significantly lower in the CUMS group (P < 0.001) (Fig. 2a). In the forced swim test, the immobility time was significantly longer in the CUMS group (P < 0.01) (Fig. 2b). In the tail suspension test, the immobility time did not differ significantly between groups (P > 0.05) (Fig. 2c). In the open field test, the total movement distance was significantly reduced in the CUMS group (P < 0.01) (Fig. 2d), and representative trajectories are shown in Fig. 3. The above behavioral outcomes met the pre-defined model validation criteria. Based on the pre-defined behavioral combination criteria, the CUMS model was successfully established. Behavioral test outcomes in Control and CUMS mice. (a) Sucrose preference rate (%) in the sucrose preference test (SPT). (b) Immobility time (s) during the last 4 min of the forced swim test (FST). (c) Immobility time (s) during the last 4 min of the tail suspension test (TST). (d) Total distance traveled (cm) during the 10-min open field test (OFT). Data are shown as mean ± SEM with individual mice overlaid (Control n = 9, CUMS n = 10). Group comparisons were performed using unpaired t-tests or Mann–Whitney U-tests. **Significance: ns, P > 0.05; **P < 0.01; *P < 0.001 Representative locomotor trajectories in the open field test (OFT). Mice were allowed to freely explore the arena for 10 min and trajectories were recorded using the SMART video tracking system. Left, CUMS; right, Control. The outer boundary indicates the arena perimeter and the inner square indicates the center zone used for quantification. Trajectories are shown for visualization; quantitative group comparisons are provided in Fig. 2d Alpha and beta diversity of gut microbiota To assess the alpha diversity of the gut microbiota, several diversity indices, including the Chao1, Observed species, Shannon, Simpson, Pielou’s evenness, and Good’s coverage indices, were employed in this study. No significant differences were observed between the CUMS and Control groups in the Chao1, Observed species, Shannon, or Good’s coverage indices. However, the Simpson index (P = 0.018) and Pielou’s evenness index (P = 0.022) were significantly higher in the Control group than in the CUMS group (Fig. 4). Regarding beta diversity, Non-metric Multidimensional Scaling (NMDS) analysis suggested partial separation of the two groups in terms of microbial composition (Fig. 5). Alpha diversity of the gut microbiota in Control and CUMS groups. Each panel corresponds to a different alpha diversity index, indicated in the gray area at the top of the panel. The x-axis shows the experimental groups, and the y-axis shows the value of the alpha diversity index. Boxplots display the median (center line), interquartile range (IQR, box), and maximum/minimum values within 1.5×IQR (whiskers). Points outside the whiskers represent outliers. Group comparisons were performed using the Kruskal-Wallis test. For indices with ≤ 4 groups, Dunn’s post-hoc test was applied where appropriate. P-values from the Kruskal-Wallis test are displayed under the respective alpha diversity index labels. **Significance notation: *P < 0.05; **P < 0.01; *P < 0.001 NMDS ordination of gut microbial community structure based on Bray–Curtis dissimilarity derived from shotgun metagenomic relative abundance profiles. Control and CUMS samples are shown. Shaded ellipses represent 95% confidence intervals; the stress value is shown above the plot. Group differences were evaluated by ANOSIM (R = 0.145, P = 0.03) and PERMANOVA (Bray–Curtis; pseudo-F = 1.675, R² = 0.0897, P = 0.033; 999 permutations). Homogeneity of dispersion was assessed using PERMDISP The ANOSIM test indicated a statistically significant but small difference in community structure (R = 0.145, P = 0.03). Complementary PERMANOVA (Bray–Curtis, 999 permutations) indicated a statistically significant but small group effect (pseudo-F = 1.675, R² = 0.0897, P = 0.033). To verify that this finding was not driven by heterogeneous within-group dispersion, we conducted PERMDISP (betadisper, 999 permutations), which showed no significant difference in dispersion between groups (F = 0.5848, P = 0.482). Analysis of differences in gut microbiota species LEfSe was employed as an exploratory approach to highlight taxa showing discriminatory signals between the two groups and to generate candidate features for interpretation (Figs. 6 and 7). Because shotgun metagenomic taxonomic profiles are compositional and LEfSe does not explicitly account for compositionality, library size, or false discovery rate control, these findings should be interpreted with caution and viewed as hypothesis-generating. This method integrates the non-parametric Kruskal-Wallis test, Wilcoxon rank-sum test, and Linear Discriminant Analysis (LDA) to assess differences in species abundance and their effect sizes across different taxonomic levels (Figs. 6 and 7). The results revealed that at the family level, the abundances of Lactobacillaceae and Peptostreptococcaceae were significantly higher in the Control group compared to the CUMS group, whereas the abundance of Anaerovoracaceae was significantly elevated in the CUMS group. At the genus level, Lactobacillus, Romboutsia, and Dubosiella were significantly enriched in the Control group. Conversely, the abundances of Schaedlerella, Mobiluncus, Emergencia, and Phocaeicola were significantly higher in the CUMS group. At the species level, the abundances of Dubosiella newyorkensis, Ligilactobacillus murinus, Limosilactobacillus urinaemulieris, and Romboutsia ilealis were significantly greater in the Control group than in the CUMS group. In contrast, Schaedlerella sp. 910,574,865, Emergencia sp. 009936045, and Adlercreutzia muris were significantly enriched in the CUMS group. LEfSe analysis of differentially abundant taxa between Control and CUMS groups. Bar plot shows LDA scores for taxa with significant differences (Kruskal–Wallis P < 0.05; LDA threshold = 3.0) based on shotgun metagenomic taxonomic profiles. Bars indicate taxa enriched in Control or CUMS. Prefixes o_, f_, g_, and s_ denote order, family, genus, and species, respectively. LEfSe results are exploratory and based on nominal P-values without false discovery rate correction LEfSe taxonomic cladogram showing differentially abundant taxa between Control and CUMS groups. Nodes represent taxa across ranks; highlighted branches indicate taxa enriched in Control or CUMS based on shotgun metagenomic profiles. This cladogram reflects exploratory LEfSe findings and should be interpreted cautiously Analysis of intestinal mucosal neurotransmitters Using targeted LC–MS/MS, we quantified a panel of neurotransmitters and related metabolites in intestinal mucosal tissue. Concentrations were normalized to wet tissue weight and analytical stability was monitored using pooled QC injections across the batch. To provide an overview of group-level differences, we first applied multivariate pattern recognition (PLS-DA), followed by univariate comparisons of individual analytes. Partial least squares-discriminant analysis (PLS-DA) was used to visualize overall between-group differences in the intestinal mucosal neurotransmitter profile. The PLS-DA score plot showed separation between the CUMS group and the Control group (Fig. 8). Compared to the Control group, the levels of 5-HIAA, 5-HT, 5-HTP, and kynurenine (Kyn) in intestinal mucosal tissue were significantly decreased in the CUMS group (Fig. 9a–d), whereas the levels of phenylethylamine (PA) and norepinephrine (NE) were significantly increased (Fig. 9e–f). No significant differences were observed in the levels of other analytes in the targeted panel. Partial least squares-discriminant analysis (PLS-DA) score plot of intestinal mucosal neurotransmitter measured by LC–MS/MS. PLS-DA was performed using the quantified analyte concentrations to visualize between-group separation. Control and CUMS samples are shown; shaded ellipses represent 95% confidence intervals Intestinal mucosal neurotransmitter concentrations in Control and CUMS mice. (a) 5-HIAA, (b) 5-HT, (c) 5-HTP, (d) kynurenine (Kyn), (e) norepinephrine (NE), and (f) phenylethylamine (PA). Bars represent mean ± SEM with individual mice overlaid. Group comparisons were performed using unpaired t-tests or Mann–Whitney U-tests as appropriate. **Significance: *P < 0.05; P < 0.01 Correlation analysis between gut microbiota and intestinal mucosal neurotransmitters Spearman correlation analysis was performed to assess the relationships among different gut microbial communities, intestinal mucosal neurotransmitter levels, and behavioral indices. The results revealed that, regarding behavioral aspects, the order Lactobacillales, the family Lactobacillaceae, the genus Lactobacillus, Limosilactobacillus urinaemulieris, the genus Romboutsia, and Romboutsia ilealis were all significantly positively correlated with the SPT percentage. Conversely, the family Anaerovoracaceae, the genus Emergencia, and its unspecified species Emergencia sp. 009936045 were significantly negatively correlated with the SPT. Furthermore, Lactobacillales, Lactobacillaceae, Lactobacillus, and Limosilactobacillus urinaemulieris were negatively correlated with the immobility time in the FST. The genus Romboutsia, Romboutsia ilealis, and Limosilactobacillus urinaemulieris were positively correlated with the total movement distance in the OFT. Regarding neurotransmitters, Lactobacillales was positively correlated with kyn levels. Lactobacillaceae was positively correlated with both 5-HIAA and kyn levels. The genus Lactobacillus was positively correlated with 5-HTP levels. Limosilactobacillus urinaemulieris was significantly positively correlated with the levels of 5-HT, 5-HIAA, 5-HTP, and kyn. The genus Romboutsia and Romboutsia ilealis were also positively correlated with 5-HT and kyn levels. In contrast, the family Anaerovoracaceae, the genus Emergencia, the genus Kineothrix, the unspecified species Schaedlerella sp. 910,574,865, and Emergencia sp. 009936045 were negatively correlated with kyn levels. Lactobacillaceae, the genus Lactobacillus, and Limosilactobacillus urinaemulieris were significantly negatively correlated with the levels of PA and NE (Fig. 10). CSpearman correlation heatmap among LEfSe-identified differential taxa, behavioral indices, and intestinal mucosal neurotransmitter levels. Color indicates Spearman’s correlation coefficient. Significance: *P < 0.05; **P < 0.01 (nominal, unadjusted) Discussion Although CUMS induced robust anhedonia in the sucrose preference test and increased immobility in the forced swim test, we did not observe a significant difference in the tail suspension test (TST). Divergent outcomes between the FST and TST have been reported in chronic stress paradigms, and this may be particularly evident during adolescence, when developmental stage, coping strategies, and task-specific sensitivities can influence behavioral readouts. Importantly, the lack of a TST change also suggests that the increased FST immobility is unlikely to be solely explained by a global impairment in motor function. In the present work, model success was defined a priori based on a reduction in sucrose preference together with at least one supportive behavioral alteration. Therefore, we interpret the phenotype as depression-relevant but acknowledge that not all despair-based assays captured group differences in this cohort. Future studies incorporating additional behavioral endpoints and physiological measures of stress responsivity will further strengthen phenotypic robustness in adolescent mice. In recent years, the role of the gut microbiota in the pathophysiology of depression during adulthood has gained increasing attention. Because the microbiota changes throughout the lifespan, it is important to study its impact on depressive-like behaviors at different developmental stages. This study showed significant differences in gut microbiota diversity between adolescent mice exposed to CUMS and control mice. Specifically, the control group showed higher alpha diversity of gut microbiota than the CUMS group. This agrees with some previous studies [9, 27]. However, other studies have reported conflicting results. For example, one study using adolescent mice found chronic stress led to increased alpha diversity [28]. Other studies observed no difference in alpha diversity between groups [29, 30]. In terms of beta diversity, this study found a statistically significant but small difference between the CUMS and control groups. This aligns with most existing research. Previous reports found significant differences in gut microbiota beta diversity in patients with major MDD or in animal depression models [27, 29, 31]. In summary, this study confirms differences in gut microbiota composition and diversity between adolescent CUMS mice and controls. However, the trends of these changes are not the same across studies. These inconsistencies may stem from the unstable nature of adolescent gut microbiota and its susceptibility to various factors. They may also arise from differences in sample size, animal model setup, detection techniques, or statistical methods. LEfSe analysis was used as an exploratory, hypothesis-generating tool to highlight taxa with discriminatory signals between the two groups. In total, LEfSe suggested 27 candidate taxa across taxonomic levels (nominal P < 0.05), which were further explored by correlation analyses. Previous research is broadly consistent with these observations in adolescent depression. For example, one clinical study recorded lower levels of Faecalibacterium, Roseburia, Collinsella, Faecalibacterium prausnitzii, and some Lachnospiraceae in feces from adolescents with depression. The abundance of Roseburia had high predictive value [8]. For treatment, several studies suggest that targeting gut microbiota can improve depressive-like behaviors. For instance, fecal microbiota transplantation (FMT) from healthy adolescent donors to depression model mice reduced depressive symptoms [8, 32]. Another study showed that CUMS mice treated with oral Lactobacillus plantarum D-9 had changes in microbiota, modified tryptophan metabolism, HPA axis function, and inflammation. These changes went along with less anxiety and depressive behaviors [33]. The gut microbiota is now seen as a key marker in adolescent depression. Adjusting the microbiota with probiotics, prebiotics, diet, or FMT may affect the central nervous system and help manage symptoms [34]. These strategies provide new ideas for clinical management of depression and offer hope for better long-term outcomes. Neurotransmitters of intestinal origin are involved not only in the regulation of local intestinal motility and sensory functions but also in the remote modulation of central nervous system function and emotional states via pathways such as humoral circulation, immune regulation, and vagal nerve transmission. This study found that in adolescent mice subjected to the CUMS model, the levels of 5-HIAA, 5-HT, 5-HTP, and kynurenine were significantly lower than those in the control group, whereas the levels of PA and NE were markedly elevated. The role of the neurotransmitter system in psychiatric disorders has been extensively investigated. Previous studies have demonstrated that in CUMS-induced depression models, the content of 5-HT in the hippocampus, prefrontal cortex, and serum is significantly reduced [35–37]. Other studies have indicated that decreased plasma 5-HT levels are associated with the severity of depression, while its metabolite 5-HIAA is significantly correlated with suicidal behavior [38, 39]. It is widely accepted that metabolic disturbances of monoamine neurotransmitters, including 5-HT, dopamine (DA), and NE, contribute to the pathophysiology of depression. These substances influence the onset and progression of depression by modulating brain circuits involved in motivation, emotion, cognition, and stress responses [38, 40]. A study using magnetic resonance spectroscopy (MRS) reported elevated levels of γ-aminobutyric acid (GABA) and reduced levels of glutamate (GLU) in the cerebrospinal fluid of patients with first-episode MDD [40]. Another clinical study revealed significantly decreased levels of tryptophan–kynurenine pathway metabolites, including 5-HT, kynurenine, and kynurenic acid (Kyna), in the serum and urine of adolescent patients with depression [8]. Notably, the present study observed increased NE levels in the intestinal tissue of CUMS-exposed mice, which appears to differ from the commonly reported reduction of central NE in depression. This apparent mismatch may reflect compartment-specific and developmental-stage-specific dynamics. Chronic stress engages the HPA axis and can facilitate locus coeruleus–norepinephrine (LC–NE) hyperactivity, while amygdalar hyperactivity provides excitatory drive to hypothalamic stress centers and shapes monoamine metabolism in prefrontal circuits. In chronic immobilization stress models, inactivation of the basolateral amygdala alleviates negative affective behaviors and normalizes stress-induced alterations in monoamine metabolism, including NE, in the prefrontal cortex, underscoring a key role for amygdalar regulation within the HPA axis–monoamine network [11]. Similarly, amygdala-related pathological changes have been reported in non-stress models such as neonatal clomipramine exposure, supporting convergent stress-circuit dysfunction across models [14]. In the gut, dense sympathetic terminals and immune cells can release NE; therefore, sustained HPA-axis engagement and sympathetic outflow during adolescent CUMS may elevate intestinal NE despite potential central monoamine alterations. Moreover, stress-related low-grade intestinal inflammation may promote cytokine-driven sympathetic activation and local NE release [41, 42]. Notably, we observed concurrent reductions in 5-HTP、5-HT、5-HIAA and kynurenine in the intestinal mucosa. Although inflammatory activation is often associated with a shift from serotonin toward kynurenine metabolism, our pattern suggests that adolescent CUMS may reduce the upstream tryptophan pool or its intestinal availability, thereby lowering flux through both branches. An alternative explanation is enhanced downstream utilization of kynurenine, which was not quantified here. Importantly, mucosal metabolite levels reflect the integrated effects of host synthesis and metabolism together with microbial production or microbiota-mediated regulation; future studies combining targeted TRP/KYN panels and host enzyme expression will help clarify these mechanisms. We further analyzed the correlation between the gut microbiota and intestinal mucosal neurotransmitter levels. The results revealed that Limosilactobacillus urinaemulieris was significantly positively correlated with 5-HT, 5-HIAA, 5-HTP, and kynurenine. 5-HT is widely distributed in the central nervous system (CNS) and peripheral tissues, with approximately 95% synthesized in the gut and only about 5% originating from the CNS. In the CNS, 5-HT is a key neurotransmitter regulating mood, sleep, and appetite. In the gut, 5-HT is primarily synthesized and released by enterochromaffin (EC) cells, which form functional connections via direct synaptic contacts with nerve terminals of the enteric nervous system (ENS), collectively participating in gut–brain signaling. The synthesis of 5-HT depends on the hydroxylation and decarboxylation of tryptophan (TRP): TRP is first converted to 5-HTP by tryptophan hydroxylase (TPH), and then transformed into 5-HT via aromatic L-amino acid decarboxylase. TPH exists in two isoforms: TPH1 is mainly expressed in peripheral tissues (e.g., the gut) and regulates peripheral 5-HT levels, whereas TPH2 is predominantly found in neurons. Subsequently, 5-HT can be metabolized by monoamine oxidase (MAO) into 5-HIAA [43]. Gut microbes can influence TPH1 expression directly or indirectly, thereby modulating 5-HT synthesis in EC cells and further affecting central 5-HT levels. For instance, Clostridium ramosum significantly upregulates TPH1 expression and increases intestinal 5-HT content [44]; Lactobacillus rhamnosus and Bifidobacterium longum can elevate TPH2 and 5-HT levels, exerting antidepressant effects [45]. Clinical studies also support this mechanism: daily intake of Lactobacillus plantarum DR7 for 12 weeks significantly alleviated anxiety and improved mental state in adults [46]; a study on adolescent depression found reduced fecal microbial diversity, decreased short-chain fatty acid-producing bacteria, and concomitant declines in serum 5-HT levels, suggesting that alterations in gut microbiota may influence peripheral 5-HT levels via the tryptophan metabolic pathway [29]. Moreover, transplantation of Coprococcus eutactus from healthy adolescents into adolescent mice subjected to chronic restraint stress reversed depression-like behaviors and significantly increased colonic and serum levels of 5-HT and GABA, further confirming that specific probiotics exert antidepressant effects by modulating neurotransmitters [47]. In another important TRP metabolic pathway—the kynurenine pathway—TRP is catalyzed by tryptophan 2, 3-dioxygenase (TDO) or indoleamine 2, 3-dioxygenase (IDO) to form N-formylkynurenine, which is then converted to KYN. KYN can be further metabolized into neuroprotective kynurenic acid (KA) or neurotoxic quinolinic acid (QA) [48]. An animal study demonstrated that oral administration of Lactobacillus rhamnosus significantly improved depression-like behaviors in a mouse model of depression, potentially through upregulating key enzyme expression in the 5-HT synthesis pathway while suppressing the production of metabolites in the kynurenine pathway [49]. At the clinical level, studies also indicate a close relationship between gut microbiota and the kynurenine metabolic pathway in depression. One clinical trial observed a significant increase in the relative abundance of Actinobacteria in the gut of patients with depression, alongside an overall reduction in plasma levels of TRP and its metabolites in the KYN pathway [50]. In summary, targeting the gut microbiota to regulate the 5-HT and KYN metabolic pathways offers a novel strategy for adjunctive treatment of depression. Restoring microbiota–gut–brain axis function through probiotics, microbiota transplantation, or nutritional interventions may improve serotonergic neurotransmission and alleviate depressive symptoms [51]. Notably, in the present study we quantified metabolites only in intestinal mucosa; plasma transport, brain-region neurotransmitter content, and related pathway molecules were not measured, so any gut-to-brain mechanistic links discussed here are based on prior literature and require direct verification. Limitations This study has several limitations. First, our analyses are cross-sectional and correlational; therefore, causal relationships between CUMS-associated microbial shifts and changes in intestinal mucosal neurotransmitters cannot be inferred. Second, we did not quantify circulating neurotransmitters, neurotransmitter content in brain regions, or key pathway molecules, thus, the “gut–blood–brain” signaling chain cannot be directly evaluated. Third, microbiome differential abundance was assessed using LEfSe, which is widely used for biomarker discovery but does not fully address the compositional nature of shotgun metagenomic data or library-size effects; moreover, LEfSe and taxon–neurotransmitter correlation analyses were reported using nominal P-values without false discovery rate correction. Therefore, false-positive discoveries cannot be excluded and the reported taxa should be interpreted as exploratory candidates. Fourth, the sample size was modest, and independent replication in larger adolescent cohorts is needed. Fifth, only male mice were included, which may limit generalizability across sexes. Finally, neurotransmitters were measured at a single post-stress time point in intestinal mucosal; longitudinal sampling and interventional experiments are required to determine temporal dynamics and causality. Conclusion Adolescent mice exposed to CUMS showed depression-relevant behavioral alterations, shifts in gut microbiota composition, and changes in 5-HT pathway metabolites. Gut microbiota dysbiosis was significantly associated with alterations in 5-HT pathway metabolites. Because this study is correlational, causal relationships require validation in future interventional studies. Electronic Supplementary Material Below is the link to the electronic supplementary material. Supplementary file 1. Acknowledgements Not applicable. Abbreviations Chronic unpredictable mild stress Linear discriminant analysis effect size 5-hydroxyindoleaceticacid Serotonin 5-hydroxytryptophan Kynurenine Phenylethylamine Norepinephrine Major depressive disorder Healthy controls γ-aminobutyric acid Chronic restraint stress Tryptophan hydroxylase-2 Tryptophan hydroxylase-1 Specific pathogen-free Sucrose preference test Forced swimming test Tail suspension test Open field test Open reading frames Electrospray ionization Non-metric multidimensional scaling Linear discriminant analysis Partial least squares-discriminant analysis Kynurenic acid Hypothalamic–pituitary–adrenal Central nervous system Monoamine oxidase Tryptophan 2, 3-dioxygenase Indoleamine 2, 3-dioxygenase Quinolinic acid. PERMANOVA, permutational multivariate analysis of variance Author contributions N. Sun and Z.F. Liu designed the experiment; P.H. Liu, Z.W. Jiang, X.S. Jing, and Y. Li participated in the animal experiments and collected experimental data; J.R. Xing and F.T. Guo analyzed the data and wrote the paper. All authors reviewed the manuscript. Funding This study was supported by the National Natural Science Foundation of the China Youth Program (82201691), the National Natural Science Foundation of China (82471558), Shanxi Province Higher Education “Billion Project” Science and Technology Guidance Project, Key Laboratory of Cellular Physiology (Shanxi Medical University), Ministry of Education, China. This work was supported by the Central Guiding Local Technology Development Funding Projects (YDZJSX2024B009). Data availability The raw shotgun metagenomic sequencing data generated in this study have been deposited in the NCBI Sequence Read Archive (SRA) under BioProject accession PRJNA1440629. The corresponding BioSample accession numbers are SAMN56619081–SAMN56619099, and the run accession numbers are SRR37716095–SRR37716113. Declarations Ethics approval and consent to participate The animal experiments in this study were approved by the Ethical Review Committee for Laboratory Animal Welfare at the First Hospital of Shanxi Medical University (Approval No.: DWLL-2024-035) and complied with the Chinese National Laboratory Animal-Guidelines for Ethical Review of Animal Welfare (GB/T35892-2018). Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Penghong Liu, Email: liupenghong6@163.com. Zhifen Liu, Email: zhifenliu@sxmu.edu.cn. Ning Sun, Email: sunning@sxmu.edu.cn. References Associated Data Supplementary Materials Supplementary file 1. Data Availability Statement The raw shotgun metagenomic sequencing data generated in this study have been deposited in the NCBI Sequence Read Archive (SRA) under BioProject accession PRJNA1440629. The corresponding BioSample accession numbers are SAMN56619081–SAMN56619099, and the run accession numbers are SRR37716095–SRR37716113.