Gut microbiota dysbiosis drives depression-like behavior in adolescent rats via lysine-regulated mTOR autophagy pathway

Abstract

The prevalence of major depressive disorder (MDD) is increasing globally, particularly among adolescents. Although gut-brain axis dysfunction has been implicated in adolescent depression, the mechanisms by which gut microbiota dysbiosis drives depressive behaviors and potential antidepressant targets remain unclear. In this study, fecal microbiota transplantation (FMT) was performed from either healthy controls (HCs) or adolescents with MDD into antibiotic-treated adolescent rats. FMT from MDD adolescents induced depressive-like behaviors in recipient rats. Metagenomic sequencing revealed that FMT from MDD adolescents led to alterations in gut microbiota in recipient rats. While qPCR, Western blotting, immunofluorescence, and transmission electron microscopy (TEM) confirmed that these rats exhibited prefrontal cortex (PFC) autophagy hyperactivation, evidenced by a reduction in SQSTM1/p62 levels, an elevation in the LC3-II/LC3-I ratio, upregulated Beclin1, and increased numbers of autolysosomes. Similar autophagy-related transcriptional changes were observed in peripheral blood from MDD adolescents. Furthermore, ELISA showed reduced plasma lysine levels in MDD adolescents and decreased lysine concentrations in the PFC of FMT-MDD rats. The antidepressant effect of lysine and its interaction with autophagy were explored in a chronic unpredictable mild stress (CUMS) rat model with or without rapamycin (the autophagy activator, RAPA). Lysine supplementation alleviated depressive-like behaviors and suppressed PFC autophagy hyperactivation, while these effects were abolished by RAPA co-treatment. These findings reveal lysine deficiency as a metabolic bridge between gut microbiota imbalance and neuronal autophagy dysregulation, suggesting a gut microbiota-lysine-autophagy axis as an innovative mechanism and therapeutic focus for adolescent depression.

Introduction

Major depressive disorder (MDD) in children and adolescents poses a substantial global public health challenge. Epidemiological studies indicate a prevalence of approximately 4–5%, imposing a considerable disease burden within this age group [1, 2]. In addition to mood disturbances, affected youth often exhibit more pronounced functional impairments in academic performance and social integration compared with adults [3]. Critically, the substantial suicide risk associated with MDD makes it the leading cause of death in the 15–19-year-old age group [4]. Current research on depression encompasses diverse domains, including immune–inflammatory dysregulation, genetic and epigenetic alterations, changes in neurotrophin signaling and neurogenesis, monoaminergic neurotransmission, stress biology, metabolomic pathways such as kynurenine, and mitochondrial oxidative stress [5]; Nevertheless, developmental and age-specific mechanisms underlying MDD remain insufficiently characterized. Therapeutic challenges further exacerbate the problem: antidepressants show diminished efficacy in younger populations, and certain agents (e.g., venlafaxine) may even increase suicidality [6]. Collectively, the high prevalence, severe outcomes, limited mechanistic understanding, and constrained therapeutic options underscore the urgent need to elucidate the developmental pathophysiology of adolescent MDD to enable more targeted and effective interventions.

The microbiota-gut-brain (MGB) axis constitutes an essential communication network that operates bidirectionally, integrating physiological processes from both peripheral and central systems [7]. Accumulating evidence highlights the MGB axis as both a fundamental mechanistic pathway and a promising therapeutic target in neuropsychiatric disorders, particularly MDD [8]. Concurrently, macroautophagy (hereafter autophagy)—a conserved lysosomal degradation pathway critical for cellular homeostasis, organelle quality control, and stress adaptation—has been recognized as a significant factor in the pathophysiology of MDD [9]. Dysregulated autophagy within specific brain regions has been increasingly implicated in depressive symptomatology and antidepressant mechanisms [10].

Of particular relevance, recent studies demonstrate that gut microbiota directly modulate autophagic activity through multiple mechanisms, with microbial metabolites (e.g., SCFAs) activating AMPK/mTOR pathways [11]; pathogen-associated molecules regulate LC3-II recruitment [12]; and tryptophan catabolites influence neuronal autophagy [13]. This interplay situates autophagy as a mechanistic bridge within the MGB axis [14]. Notably, autophagy exhibits developmental stage-specificity. Adolescent brains, which are undergoing active synaptic pruning and cortical maturation, exhibit distinct patterns of autophagic flux patterns compared with adults [15]. Recent microbiome-based studies further demonstrate that adolescents with MDD display marked gut microbial and metabolic dysbiosis [16]. This developmental period coincides with heightened vulnerability to stress-induced depression and dynamic gut microbiome assembly [17], suggesting that interactions between the MGB axis and autophagy interactions may uniquely contribute to adolescent neuropathology [18]. Building on established links between depression and autophagy dysfunction, we propose that modulation of autophagy, particularly within neuronal and glial cells, represents a promising therapeutic strategy for adolescent MDD.

This study posits that dysregulation of neuronal autophagy plays a role in the onset of depressive symptoms, and suggests that lysine supplementation can restore autophagic balance and thereby ameliorate depressive-like behaviors. To examine this hypothesis, we first established an FMT model in antibiotic-treated recipient rats using fecal samples from adolescents with MDD or healthy controls, followed by behavioral assessment, amino acid metabolism, and autophagy evaluation in the PFC. We then employed a CUMS model with dietary lysine restriction to further examine the contribution of lysine deficiency to autophagy dysfunction. Finally, in CUMS-induced depressed rats, we investigated whether lysine supplementation or pharmacological autophagy activation with RAPA could modulate autophagy activity and depressive-like behaviors.

Materials and methods

Clinical study design and fecal sample collection

Sixteen adolescents aged 13–18 years meeting the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) criteria for major depressive disorder (MDD) were recruited from the Department of Psychiatry, the First Affiliated Hospital of Chongqing Medical University or Bishan Middle School, Chongqing, China in this study. Sixteen healthy controls (HCs), matched by age and sex without any psychiatric or somatic disorders were recruited from the Bishan Middle School, Chongqing, China or through public advertisements. To evaluate depressive and anxiety symptoms in MDD patients, the Child Depression Rating Scale-Revised (CDRS-R) and the 14-item Hamilton Anxiety Scale (HAMA-14) were employed, respectively. Exclusion criteria for all participants included: (1) use antibiotics or probiotics within the previous one months; (2) have other physical illness, illicit drug use, pregnancy or lactating. (3) have inflammatory bowel disease, colorectal tumor, or any other mental disorders. Fecal samples were collected from all participants and promptly stored at −80 °C until metagenomic sequencing. Peripheral blood samples were collected from the participants via venipuncture into EDTA-coated anticoagulant tubes (Becton, Dickinson and Company, USA) during the morning (8:00 AM-12:00 AM). Following centrifugation at 3000 rpm for 10 min at 4 °C, plasma supernatant was collected and maintained at −80 °C until ELISA. For peripheral blood mononuclear cells (PBMCs) isolation, samples were processed via the Ficoll method (Cytiva, 17144003) with centrifugation at 400 ×g for 20 min at 4 °C. PBMCs were cryopreserved in a solution of fetal bovine serum (FBS) and dimethyl sulfoxide (DMSO) (9:1 ratio) and stored in liquid nitrogen until the real-time quantitative PCR (RT-qPCR) analysis. This study was approved by the Ethics Committee of the First Affiliated Hospital of Chongqing Medical University (approval number: 2024-129-01). Written informed consent was obtained from all participants and their legal guardians prior to enrollment. All methods were performed in accordance with the relevant guidelines and regulations.

Animal experimental design and drug treatment

Male Sprague-Dawley (SD) rats aged three weeks were obtained from the Experimental Animal Centre of Chongqing Medical University. Prior to the initiation of the experiments, all rats were acclimatized for one week under standard laboratory housing conditions, which included a 12-h light/dark cycle (lights on 08:00–20:00), ambient temperature maintained at 20–26 °C, and relative humidity at 40–70%. Sterile chow and water were available ad libitum. All animal experiments received prior approval from the Institutional Animal Care and Use Committee of Chongqing Medical University (IACUC-CQMU-2024-0040) and were conducted in accordance with the relevant institutional guidelines and regulations.

Fecal microbiota transplantation (FMT) model

Rats were divided into two groups randomly: those receiving fecal microbiota from HCs (FMT-HCs, n = 12), and those receiving fecal microbiota from MDD patients (FMT-MDD, n = 12). Before FMT procedures, all rats were pretreated with a broad-spectrum antibiotic mixture (ABX; 25 g/L ampicillin, 25 g/L metronidazole, 25 g/L neomycin and 12.5 g/L vancomycin, 0.25 g/L amphotericin B, dissolved in phosphate-buffered saline [PBS]) administered via oral gavage twice daily (0.2 mL per dose at 9:00 and 19:00) for 14 consecutive days. For FMT preparation, frozen fecal samples from 16 HCs and 16 MDD patients were pooled within each group. Samples were resuspended in sterile saline (40 mL per 0.8 g feces), homogenized for 3 min using an ultrasonic disruptor, and followed by centrifugation at 1000 rpm for 20 min. The supernatant was collected, supplemented with glycerol to a final concentration of 15%, and stored at −80 °C. To ensure consistency, each rat received 1 mL of the prepared fecal supernatant by oral gavage at 09:00 AM once daily for 14 days. All procedures were performed aseptically within a biosafety cabinet using sterile instruments and personal protective equipment to prevent contamination. Body weight was recorded every week. After the 14-day transplantation, behavioral tests were conducted, including sucrose preference test (SPT), forced swim test (FST), open field test (OFT), elevated plus-maze test (EPM). Following behavioral assessment, rats were sacrificed by transcardial perfusion. The PFC was micro-dissected according to anatomical landmarks based on the rat brain atlas (Paxinos and Watson). Briefly, brain tissues were rapidly extracted and chilled on ice (4 °C), after removal of the olfactory bulbs, a coronal cut was made approximately at the level of the genu of the corpus callosum (about +3.0 mm anterior to bregma). The anterior part of the frontal lobes rostral to this cut was collected as the PFC sample. The sample was then snap-frozen in liquid nitrogen, and stored at −80 °C.

Lysine and RAPA treatment in a chronic unpredictable mild stress (CUMS) model

To explore the role of lysine-mediated autophagy, we employed both L-lysine (400 mg/kg; Sigma, P6407; dissolved in sterile saline) and the autophagy activator RAPA (2 mg/kg; MCE, AY-22989; prepared in a vehicle of 4% anhydrous ethanol, 5% Tween-80, and 91% saline). Rats were randomly divided into five groups (n = 8 per group): control group (CON + Saline), saline experimental group (CUMS + Saline), L-lysine and RAPA experimental group (CUMS + Lys + RAPA), lysine experimental group (CUMS + Lys), RAPA experimental group (CUMS + RAPA). All groups except the CON group were subjected to a 3-week CUMS protocol. The CUMS protocol consisted of 13 types of mild stressors, with one short-duration and one long-duration stimulus applied per day in a randomized schedule, ensuring that the same stimulus was not repeated on consecutive days. Short-duration stressors included tail pinch (1 min), foot shock, forced swimming (5 min) and cold exposure (1 h), while long-duration stressors included restraint (6 h), cage tilting at 45° (24 h), wet bedding (24 h), fasting (24 h), water deprivation (24 h) and crowding (24 h). Rats in the CON group were maintained under standard conditions for the same period. Body weight was monitored every week. After CUMS, behavioral tests (SPT, OFT, EPM and FST) were performed. Rats were then sacrificed by transcardial perfusion, and brain tissue was processed as described above. The PFC was precisely dissected, snap-frozen in liquid nitrogen, and stored at −80 °C for subsequent analyses.

Behavioral assessment

Sucrose preference test (SPT)

To assess anhedonia, the SPT was conducted following a 6-day adaptation protocol consisting of two days of dual pure water exposure, followed by two days of dual 1% sucrose solution exposure, and then two days with a choice between both fluids. On day 7, following 12 h of food and water deprivation, rats were presented with one bottle of 1% sucrose solution and one bottle of pure water placed on opposite sides of the cage. The bottles were reversed after 1 h to control for positional bias. Two hours later, both bottles were removed and the volumes of consumed fluids were measured. Sucrose preference was calculated as follows: Sucrose preference (%) = [sucrose intake / (sucrose intake + water intake)] × 100.

Open field test (OFT)

To evaluate locomotor activity and anxiety-like behavior, rats were acclimated to the behavioral testing room for 1-h prior to testing. Each rat was then positioned in the middle of a black chamber (100 × 100 × 40 cm) to explore for 5 min. Animal movement was tracked using a video camera and analyzed with ANY-maze software to quantify total locomotion distance and duration in the center. The black chamber was cleaned with 75% ethanol between trials.

Elevated plus maze (EPM)

To evaluate anxiety-like behavior, rats were transferred to the behavioral testing room for a 1-h acclimation period prior to testing. The plus maze was composed of four arms arranged in a cross configuration. The two open arms measured 50 cm in length and 10 cm in width, while the two closed arms measured 50 cm in length and 10 cm in width, with surrounding walls 40 cm in height. The entire plus maze was positioned at a fixed height of approximately 60–70 cm above the floor. Each rat was placed in the central area and permitted to explore freely for 5 min. Behavioral indices of anxiety were determined by quantifying both the duration and frequency of entries into the open and closed arms. The apparatus was cleaned with 75% ethanol between trials.

Forced swimming test (FST)

To evaluate the depression-like behavior, rats were acclimated in the behavioral testing room for 1 h prior to testing. The FST was conducted using a transparent acrylic cylinder (50 cm height, 20 cm diameter) filled with 30 cm of water maintained at 23 ± 1 °C. Rats were individually placed in the water and video-recorded for 6 min. The total immobility time was measured during the final 5 min of the test. Immobility was defined as the absence of active movement except those necessary to keep the animal’s head above water. Swimming was defined as horizontal movement in the water.

Fecal metagenomic sequencing and data analysis

Total microbial genomic DNA obtained from fecal samples with the FastPure Stool DNA Isolation Kit (Magnetic bead) (MJYH, Shanghai, China). The quality of DNA was assessed by measuring concentration and purity on a NanoDrop 2000 and Synergy HTX, with integrity verified via 1% agarose gels. Qualified DNA was then fragmented to an average size of ~350 bp using a Covaris M220 system (Gene Company Limited, China). A paired-end sequencing library was prepared with the NEXTFLEX Rapid DNA-Seq Kit (Bioo Scientific, Austin, TX, USA) and sequenced on an Illumina NovaSeq™ X Plus (Illumina Inc., San Diego, CA, USA) platform at Majorbio Bio-Pharm Technology Co., Ltd. (Shanghai, China). The resulting sequencing data have been deposited in the NCBI Short Read Archive under accession number SRP619016. Bioinformatic analyses were carried out on the platform of Majorbio Cloud Platform (www.majorbio.com). Adapter sequences were removed and low-quality reads (average quality <20 or read length <50 bp) were filtered out using fastp (v0.20.0) [19] (https://github.com/OpenGene/fastp, version 0.20.0), and host-derived sequences were removed by mapping to the Rattus norvegicus genome assembly Rnor_6.0,GCA_000001895.4 [20]. High-quality reads were then assembled into contigs with MEGAHIT (v1.1.2) [21] (https://github.com/voutcn/megahit, version 1.1.2), and only contigs ≥300 bp were retained. Open reading frames (ORFs) were predicted with Prodigal (v2.6.3) [22] (https://github.com/hyattpd/Prodigal, version2.6.3), and ORFs ≥100 bp were collected. A non-redundant gene set was generated using CD-HIT (v4.7) [23] (http://weizhongli-lab.org/cd-hit/, version 4.7) at 90% sequence identity and 90% coverage. Gene abundances were quantified with SOAPaligner (v2.21) [24] (https://github.com/ShujiaHuang/SOAPaligner, version soap2.21release) using a 95% identity threshold. Taxonomic assignments of the non-redundant genes were obtained by aligning against the NCBI NR database using DIAMOND (v2.0.13) [25] (http://ab.inf.uni-tuebingen.de/software/diamond/, version 2.0.13) with an e-value cutoff of 1e-5. Functional annotation of the non-redundant genes was carried out using multiple databases, including GO, KEGG, eggNOG, CAZy, CARD, and PHI. Based on the functional and taxonomic annotation as well as the abundance profile of non-redundant genes, differential analyses across groups were performed using the Kruskal-Wallis test.

Real-time quantitative PCR (RT-qPCR)

Total RNA was extracted from rat brain tissues using the RNAsimple Total RNA Kit (TIANGEN, DP419) according to the manufacturer’s protocol. Approximately 1 μg of total RNA was reverse-transcribed into complementary DNA (cDNA) using the PrimeScript RT kit (Takara, RR047A). Relative mRNA expression was quantified by RT-qPCR using TB Green Premix (Takara, RR820A) under the following conditions: initial denaturation at 95 °C for 30 s, followed by 40 cycles of 95 °C for 5 s and 60 °C for 30 s. Autophagy-related gene expression in rat brain tissue was analyzed using a targeted RT-qPCR array (Qiagen, 249950). Gene expression levels were normalized to β-actin (Actb) as the internal reference, and relative expression was calculated using the 2-ΔΔCt method. Reactions were performed on a LightCycler® 480 System (Roche Diagnostics, Basel, Switzerland), with at least three biological replicates analyzed. The primer sequences are provided in Table S1.

Western blotting

Western blot was performed according to standard protocols. Briefly, protein samples were denatured by boiling at 95 °C for 10 min and separated by SDS-PAGE. Proteins were then transferred onto PVDF membrane and blocked with 5% skim milk in TBST (Tris-buffered saline with 0.1% Tween 20) at room temperature. Membranes were incubated overnight at 4 °C with primary antibodies, followed by incubation with secondary antibodies for 1 h at room temperature the next day. Protein bands were visualized using the ECL Ultra Signal Chemiluminescence Kit (Abclonal, RM00020), and images were acquired using a Tanon 4600SF chemiluminescent imaging system (Tanon, Shanghai, China). Band intensities were quantified with ImageJ software. The primary antibodies used in this study included LC3A/B (CST, 12741, 1:1000), SQSTM1/p62 (Abmart, T55546, 1:1000), Beclin1 (CST, 3738, 1:1000), mTOR (Proteintech, 66888-1-Ig, 1:5000), phospho-mTOR (Ser2448) (CST, 5536, 1:1000), GAPDH (Abcam, ab9485, 1:8000). The secondary antibodies were goat anti-rabbit (Beyotime, A0208, 1:8000), and goat anti-mouse (Beyotime, A0216, 1:8000).

Immunofluorescence

Rats were deeply anesthetized and transcardially perfused with 4% paraformaldehyde (PFA). Brains were post-fixed in 4% PFA at 4 °C for 48 h, followed by dehydration in a transgradient sucrose solution (15 and 30%). Tissues were embedded in OCT compound and coronally sectioned at 30 μm thickness using a Minux® FS800 Cryostat (RWD Life Science Co., Ltd., China), and then transferred to ice-cold PBS. Sections were washed in PBST (0.3% Triton X-100 in PBS) for 15 min (5 min × 3 times) and blocked with blocking buffer (Beyotime, P0260) for 10 min at room temperature. Sections were incubated overnight at 4 °C with primary antibodies against SQSTM1/p62 (MBL, PM045MS, 1:500) and NeuN (Oasis Biofarm, OB-PRT013, 1:1000). After PBS washes (5 min × 3 times), sections were incubated with secondary antibodies Alexa Fluor 488 (Abcam, ab150077, 1:500) and Alexa Fluor 647 (Oasis Biofarm, G-RT647, 1:500) for 2 h at room temperature. Following additional PBS washes (5 min × 3 times), sections were incubated with DAPI staining solution (Beyotime, C1006) for 5 min and coverslipped. Fluorescence images were acquired using an Andor Dragonfly 200 Confocal Microscopy System (Andor Technology Ltd., Belfast, UK), processed with Imaris software (10.2.0) and analyzed with ImageJ software.

Enzyme-linked immunosorbent assay (ELISA)

Lysine concentrations in the PFC and plasma were measured using an ELISA kit (Bioobox, AKAM024M) according to the manufacturer’s protocol. Frozen PFC tissues were homogenized in extraction buffer and incubated in a water bath at 80 °C for 20 min. After centrifugation, the supernatant was collected, and 100 μL of sample was mixed with 200 μL of assay working solution, followed by incubation in a water bath at 80 °C for 30 min. For plasma samples, 100 μL of plasma was directly mixed with 200 μL of assay working solution and similarly incubated at 80 °C for 30 min. After the addition of stop solution, absorbance was measured at 530 nm using a Synergy LX Multi-Mode Microplate Reader (BioTek Instruments, Winooski, VT, USA).

Transmission electron microscopic (TEM)

Rats were deeply anaesthetized and sacrificed, and fresh PFC tissue (1 × 1 × 1 mm) was rapidly dissected and immersed in 2.5% glutaraldehyde solution in phosphate buffer at room temperature for 2 h prior to 4 °C storage. Sample was removed from the glutaraldehyde solution and washed with 0.1 M PBS (15 min × 3 times), and was post-fixed by 1% osmic acid at room temperature for 2 h. After additional PBS washes (15 min × 3 times), samples were dehydrated through a graded ethanol and acetone series (50, 75, 80, 95, 100% ethanol I and II, 100% acetone I and II; 15 min each). Complete dehydration was ensured while avoiding shrinkage. Sample was permeated in acetone: Epon 812 mixtures (1:1 for 2 h, then 1:2 overnight at RT), immersed in the 812 embedding agent (2 × 2 h, RT), and then embedded in the 8:12 embedding agent. After embedding, place sample in ovens at 37 and 60 °C, respectively, for 12 h to harden the embedding agent. Sections (60–80 nm) are prepared by a Leica EM UC7 ultramicrotome (Leica Microsystems GmbH, Wetzlar, Germany). The copper mesh containing the sections is immersed in a 2% acetic acid dioxyl uranium staining solution for 30 min in the dark. After rinsing with deionised water, sections were immersed in a lead citrate solution for 15 min to complete the lead-uranium double staining process. After rinsing and drying the copper mesh, we observe changes in organelle morphology by a JEOL JEM 1400 transmission electron microscope (JEOL Ltd., Tokyo, Japan), collect relevant parameters and save images for further analysis and processing.

Statistical analysis

Data were analyzed using GraphPad Prism 8. Results are presented as mean ± SD, and P < 0.05 was considered statistically significant. Normality was assessed using the Shapiro-Wilk test. Comparisons between two groups were made using unpaired Student’s t-tests or Mann-Whitney U tests where appropriate. For multiple-group comparisons, one-way or two-way analysis of variance (ANOVA) followed by Tukey’s or Bonferroni’s post hoc tests was applied. When normality assumptions were not met, the Kruskal-Wallis test followed by Dunn’s multiple comparison test was used. Specifically, baseline demographic and clinical characteristics (age, height, weight, BMI) were analyzed using one-way analysis of variance and Wilcoxon rank-sum test. Behavioral outcomes in rats (SPT, FST, TST, OFT) were analyzed by unpaired t-tests for two-group comparisons or by one-way ANOVA with Tukey’s post hoc tests when more than two groups were included (e.g., lysine supplementation). For molecular assays, including Western blotting, immunofluorescence, and TEM, protein expression levels of autophagy-related markers (LC3, SQSTM1/p62, Beclin1, mTOR, p-mTOR) were quantified and compared using unpaired t-tests (two groups) or one-way ANOVA with Tukey’s post hoc test (multiple groups). Quantification of autophagosomes from TEM images was analyzed using the same criteria. Additionally, in the CUMS model treated with lysine and RAPA, we utilized six biological replicates per group of rats. The samples were run on three separate blots, with two samples from each group loaded onto each blot. For each individual blot, the band intensity of the target protein was normalized to that of the loading control. The resulting ratio for each sample was then normalized to the average value of the control group (CON + Saline) on the same blot. Finally, the normalized values from all three blots were pooled for statistical analysis.

Result

Clinical characteristics of adolescent FMT donors

A total of 32 adolescents were enrolled in this study, including 16 patients with MDD and 16 age- and sex-matched HCs. There were no significant differences in demographic or anthropometric variables between the two groups. In detail, the mean age was comparable between groups (HCs: 16.56 ± 1.27 years; MDD: 15.93 ± 1.19 years; P = 0.221, t(30) = 1.385). Similarly, no group differences were found in height (1.63 ± 0.08 m vs. 1.64 ± 0.07 m, P = 0.746, t (30) = 0.3269), weight (53.73 ± 7.15 kg vs. 52.60 ± 7.19 kg, P = 0.637, t (30) = 0.4294), or BMI (20.00 ± 1.69 vs. 19.54 ± 2.08, P = 0.515, t (30) = 0.6589). The sex distribution was also identical, with females accounting for 75% in both groups (12/16). However, compared with the HCs group, the MDD group exhibited markedly higher scores on the CDRS-R (60.5 ± 8.54 vs. 17.94 ± 1.08, P < 0.001, t (30) = 19.14) and the HAMA-14 (26.31 ± 7.68 vs. 1.06 ± 0.89, P < 0.001, t (30) = 12.61), reflecting significant depressive and anxiety symptoms (Table 1).

Table 1.

Clinical characteristics of all subjects.

	HCs N = 16	MDD N = 16	P
Age, years, mean ± SD	16.56 ± 1.27	15.93 ± 1.19	0.221
Height, m, mean ± SD	1.63 ± 0.08	1.64 ± 0.07	0.746
Weight, kg, mean ± SD	53.73 ± 7.15	52.60 ± 7.19	0.637
BMI, mean ± SD	20.00 ± 1.69	19.54 ± 2.08	0.515
CDRS-R, mean ± SD	17.94 ± 1.08	60.5 ± 8.54	<0.001
HAMA-14, mean ± SD	1.06 ± 0.89	26.31 ± 7.68	<0.001
Famale, n(%)	12 (75.00%)	12 (75.00%)	-

BMI Body Mass Index, CDRS-R Children’s Depression Rating Scale, HAMA-14 Hamilton Anxiety Scale. Data are expressed as mean ± standard deviation (mean ± SD), and the results were analyzed by one-way analysis of variance and Wilcoxon rank-sum test. P < .05 was considered statistically significant.

FMT from depressed adolescents induces depressive-like behavior and gut microbiota alterations in recipient rats

To investigate whether gut microbiota dysbiosis contributes to depression-related phenotypes in adolescents, we performed FMT by transferring fecal samples from adolescents with MDD or HCs into antibiotic-treated adolescent rats (Fig. 1A). After a 7-day adaptation period to the housing conditions, all rats received antibiotic treatment twice daily for 2 weeks. Subsequently, the depression group rats were colonized with pooled fecal samples from patients with MDD (FMT-MDD), while the control group rats were colonized with pooled fecal samples from HCs (FMT-HCs). After 2 weeks of FMT, compared with the FMT-HCs group, the FMT-MDD group exhibited higher depressive-like behaviors, including a significant decrease in sucrose preference in the SPT (P = 0.0018, t (22) = 3.560) (Fig. 1B) and a significant increase in immobility time in the FST (P = 0.0082, t (22) = 2.907) (Fig. 1C). There were no significant differences between the FMT-HCs and FMT-MDD groups in the anxiety-like behaviors measured by the EPM (Fig. S1A) and OFT (Fig. S1B). Similarly, there was no significant difference in weight between the two groups (Fig. S1D).

Fig. 1. Study design, behavioral study, and fecal metagenomes of fecal microbiota transplantation in juvenile rats.

A Schematic representation of the animal experimental design. 24 rats were randomly allocated into two groups (n = 12/group), FMT-HCs group receiving fecal microbiota from healthy adolescents, and the FMT-MDD group receiving fecal microbiota from depressed adolescents. B Sucrose preference in SPT of FMT-HCs (mean ± SD: 89.69 ± 4.86) and FMT-MDD (mean ± SD: 81.41 ± 6.41) (n = 12/group). C Immobility time in FST of FMT-HCs (mean ± SD: 123.1 ± 20.86) and FMT-MDD (mean ± SD: 154.03 ± 30.39) (n = 12/group). D PCoA of fecal microbiota beta-diversity between FMT-HCs (n = 8) and FMT-MDD (n = 8). The community distance is based on Bray-Curtis dissimilarity metrics. Each symbol represents an individual sample. E LDA Score from fecal metagenomes analysis of fecal samples from FMT-HCs (red) and FMT-MDD (blue). F LDA score of functional KEGG Orthology (KO) analysis in fecal samples from FMT-HCs (red) and FMT-MDD (blue). Graphs were plotted using GraphPad Prism. Data are expressed as mean ± standard deviation (mean ± SD), and the results were analyzed using one-way ANOVA and Wilcoxon rank-sum test. P < 0.05 was considered statistically significant. (*) P < 0.05, (**): P < 0.01, (***): P < 0.001, (****).

Then, we profiled the fecal microbial composition post-transplantation. Alpha diversity indices (Simpson, Chao 1 and Shannon) revealed no significant differences in richness or evenness between the FMT-MDD and FMT-HCs groups (Fig. S1E). However, the principal coordinate analysis (PCoA) based on Bray–Curtis distances demonstrated clear separation between the FMT-MDD (n = 8, red dots) and FMT-HCs (n = 8, blue dots) groups (P < 0.001, F = 4.148, R² = 0.2286), indicating distinct gut microbial structures between the two groups (Fig. 1D). Taxonomic comparisons further showed that the FMT-MDD group exhibited enrichment of 26 bacterial species (LDA > 2, P < 0.05), including Akkermansia muciniphila and Ruminococcus bicirculans, and depletion of 29 species (LDA > 2, P < 0.05), primarily from the Bacteroides genus (e.g., B. ovatus, B. cellulosilyticus) (Fig. 1E). LEfSe analysis of KEGG functional pathways revealed significant metabolic alterations in the FMT-MDD group compared to the FMT-HCs group (Fig. 1F). In the MDD-FMT group, enzymes involved in amino acid metabolism and energy metabolism were notably reduced. Specifically, enzymes such as cysteine synthase (K01738), histidine ammonia-lyase (K01745), and threonine synthase (K01733) showed lower abundance, suggesting a diminished capacity for amino acid catabolism. Additionally, key enzymes in energy metabolism, including dihydrolipoamide dehydrogenase (K00382) and NADH-quinone oxidoreductase (K00330), were also found to be less abundant in the FMT-MDD group. This reduction likely reflects impaired energy production and altered redox balance, contributing to the metabolic dysfunction observed in the MDD group. Collectively, these findings indicate that FMT from adolescents with MDD induces depressive-like behaviors in recipient rats and is accompanied by substantial gut microbial alterations at both the compositional and functional levels.

Gut microbial dysbiosis induces autophagy hyperactivation in the PFC of rats with depressive-like behaviors

Autophagy is a vital process for cells to clean themselves and maintain internal balance. Emerging evidence suggests a close interplay among autophagy, the gut–brain axis, and depression [26]. To investigate whether FMT from depressed adolescents induces autophagy dysregulation, we first performed an autophagy-focused gene expression array in three brain regions of recipient rats, including the PFC, hippocampus, and nucleus accumbens. Significant alterations in autophagy-related gene expression were observed in the PFC of FMT-MDD rats compared with FMT-HCs rats (Fig. 2A), whereas no significant alterations were detected in the hippocampus (Fig. 2B) or nucleus accumbens (Fig. 2C), suggesting region-specific activation of autophagy in the PFC. Subsequently, western blot analysis of tissues from the PFC, hippocampus, and nucleus accumbens revealed that, in the PFC of the FMT-MDD group, the protein level of LC3-II/I (P = 0.0044, t (8) = 3.917) and Beclin1 (P = 0.0373, t (8) = 2.494) was significantly increased. Conversely, the expression of the autophagy substrate SQSTM1/p62 (P = 0.0276, t (8) = 2.687) was decreased (Fig. 2D). At the same time, the expression of SQSTM1/p62 in the hippocampus and nucleus accumbens of the FMT-MDD group was decreased (P = 0.0227, t (8) = 2.814; P = 0.0006, t (8) = 5.468), but there were no significant differences in the protein levels of LC3 and Beclin1 (Fig. S3). The results at the gene and protein levels indicate that the changes in autophagy in FMT rats are mainly focused on the PFC. Additionally, immunofluorescence staining revealed significantly decreased SQSTM1/p62 expression in NeuN-positive prefrontal neurons of the FMT-MDD group relative to FMT-HCs group (P < 0.0001, t (10) = 8.408) (Fig. 2E). Consistently with these findings, TEM analysis revealed an increased number of autolysosomes in the FMT-MDD group compared with the FMT-HCs group (P = 0.0147, t (4) = 4.111) (Fig. 2F). To examine whether similar alterations occur in humans, we assessed autophagy-related gene expression in the PBMCs of adolescents with MDD and HCs using RT-qPCR. Compared to HCs, MDD patients exhibited significantly lower expression of SQSTM1/p62 (P = 0.0048, t (30) = 3.042) and Bcl2l1 (P = 0.0304, t (30) = 2.273), and higher expression of Beclin1 (P = 0.0378, t (30) = 2.173), ATG5 (P = 0.0058, t (30) = 2.971), and ATG3 (P = 0.0082, t (30) = 2.830). While, LC3 (P = 0.7759, t (30) = 0.2872) expression showed no significant difference between groups (Fig. 2G). In summary, gut microbiota dysbiosis from depressed adolescents induced autophagy hyperactivation specifically in the PFC of recipient rats, with parallel alterations observed in PBMCs from adolescents with MDD.

Fig. 2. Autophagy disorder in FMT-MDD adolescent rats and adolescent MDD patients.

A-C PCR Array of autophagy-related genes in the PFC, hippocampus, and nucleus accumbens of FMT-HCs and FMT-MDD (n = 4/group). D Representative bars and Western blots showing Beclin1, SQSTM1/p62 and LC3 protein expression levels in the PFC of FMT-HCs (mean ± SD: 1 ± 0.15, 1 ± 0.14, 1 ± 0.08) and FMT-MDD (mean ± SD: 1.24 ± 0.16, 0.7 ± 0.21, 1.19 ± 0.08) (n = 5/group). E Schematic representation of brain regions. Immunofluorescence staining (40× magnification) and representative bars of SQSTM1/p62 expression in the PFC of FMT-HCs (mean ± SD: 1 ± 0.13) and FMT-MDD (mean ± SD: 0.33 ± 0.15) (n = 6/group). F Transmission electron microscopic (TEM) representative picture of the PFC in FMT-HCs (mean ± SD: 2.22 ± 0.51) and FMT-MDD (mean ± SD: 3.67 ± 0.33) (n = 3/group), with arrows indicating autophagosomes. G Expression levels of autophagy-related genes (p62, Beclin1, Atg5, Atg3, Bcl2l1, LC3) in peripheral blood mononuclear cells (PBMCs) of HCs and MDD (n = 16/group). Images from two independent experiments were analyzed using ImageJ software. Graphs were plotted using GraphPad Prism. Data are expressed as mean ± standard deviation (mean ± SD), and the results were analyzed using one-way ANOVA and Wilcoxon rank-sum test. P < .05 was considered statistically significant. (*) P < 0.05, (**): P < 0.01, (***): P < 0.001, (****).

Lysine deficiency links microbiota dysbiosis to PFC autophagy dysregulation

Dietary lysine restriction has been shown to aggravate depressive‑like behaviors in CUMS rats, highlighting lysine as a key amino acid in adolescent depression. In the above study, we applied a lysine-restricted diet to adolescent CUMS rats. The rats were randomly divided into 4 groups: controls with 100% dietary lysine (CON + L100, n = 8), controls with 70% dietary lysine (CON + L70, n = 8), CUMS with 100% dietary lysine (CUMS + L100, n = 9) and CUMS with 70% dietary lysine (CUMS + L70, n = 9) (Fig. S4) [16]. In this study, ELISA analysis in the PFC of FMT rats and in plasma from depressed patients demonstrated decreased lysine levels of FMT-MDD group (P = 0.0488, t (6) = 2.466) (Fig. 3A) and depressed adolescents (P = 0.0045, t (30) = 3.068) (Fig. 3B). Similarly, ELISA analysis in the PFC of CUMS rats with dietary lysine restriction shows that the CUMS + L70 group exhibited lower lysine levels in the PFC than the CUMS + L100 group (P = 0.0168), and the CON + L100 group (P = 0.0013), but no significant difference with the CON + L70 group (P = 0.1053) (F (3, 2) = 20.23) (Fig. 3C). Subsequently, western blot analysis of PFC revealed increased protein levels of LC3-II/I (P = 0.0459, t (4) = 2.860), along with decreased expression of the autophagy substrate SQSTM1/p62 (P = 0.0194, t (4) = 3.779) in the CUMS + L70 group compared with the CUMS + L100 group, while no significant difference in Beclin1 (P = 0.1891, t (4) = 1.581) (Fig. 3D). Together, these results implicate lysine deficiency as a key mediator connecting microbiota‑driven amino acid dysregulation to PFC autophagy abnormalities in adolescent depression.

Fig. 3. Lysine restriction exacerbated autophagic hyperactivation in the PFC of CUMS rats.

A The concentrations of Lysine in the PFC of FMT-HCs (mean ± SD: 4871.21 ± 1702.59) and FMT-MDD (mean ± SD: 2633.83 ± 628.6) (n = 4/group). B The concentrations of Lysine in the plasma of adolescents with MDD (mean ± SD: 42.96 ± 6.72) and healthy controls (mean ± SD: 52.08 ± 9.82) (n = 16/group). C The concentrations of Lysine in the PFC of CON + L70 (mean ± SD: 4397.58 ± 439.22), CON + L100 (mean ± SD: 6345.12 ± 684.18), CUMS + L70 (mean ± SD: 3434.16 ± 284.27) and CUMS + L100 (mean ± SD: 4803.37 ± 647.41) (n = 4/group). D Representative bars and Western blots showing Beclin1, SQSTM1/p62 and LC3 p-rotein expression levels in the PFC of CUMS + L70 (mean ± SD: 1 ± 0.87, 1 ± 0.74, 1 ± 0.16) and CUMS + L100 (mean ± SD: 0.2 ± 0.09, 4.23 ± 1.29, 0.71 ± 0.08) (n = 3/group). Images from two independent experiments were analyzed using ImageJ software. Graphs were plotted using GraphPad Prism. Data are expressed as mean ± standard deviation (mean ± SD), and the results were analyzed using one-way ANOVA and Wilcoxon rank-sum test. P < .05 was considered statistically significant. (*) P < 0.05, (**): P < 0.01, (***): P < 0.001, (****).

Lysine attenuates depressive-like behaviors in CUMS rats via the mTOR/Beclin1-autophagy pathway

To further investigate whether lysine exerts antidepressant effects through autophagy regulation, we tested the impact of lysine supplementation and RAPA in adolescent CUMS rats. Rats were randomly divided into five groups (n = 8 per group): control group (CON + Saline), saline experimental group (CUMS + Saline), L-lysine and RAPA experimental group (CUMS + Lys + RAPA), lysine experimental group (CUMS + Lys), RAPA experimental group (CUMS + RAPA) (Fig. 4A). After 4 weeks, body weights were reduced in all CUMS-exposed groups compared with CON + Saline (all P < 0.0001). Both the CUMS + RAPA and CUMS + Lys + RAPA groups had significantly lower body weights than the CUMS + Saline and CUMS + Lys groups (all P < 0.0001) (F (4, 35) = 65.38) (Fig. 4B). However, there was no significant difference in body weight between the CUMS + Saline and CUMS + Lys groups (P = 0.8821), as well as between the CUMS + Saline and CUMS + Lys groups (P = 0.2517). Importantly, the CUMS + Saline (P = 0.0068), CUMS + Lys + RAPA (P = 0.0028) and CUMS + RAPA groups (P = 0.0037) exhibited lower sucrose preference in the SPT than CON + Saline. The CUMS + Lys group exhibited higher sucrose preference in the SPT than the CUMS + Saline (P = 0.0389), CUMS + Lys + RAPA (P = 0.0175) and CUMS + RAPA groups (P = 0.0229) (F (4, 35) = 7.519) (Fig. 4C). Additionally, the immobility time in FST was significantly increased in the CUMS + Saline (P = 0.0100), CUMS + Lys + RAPA (P = 0.0118) and CUMS + RAPA groups (P = 0.0003) compared to the CON + Saline group. Meanwhile, the immobility time in FST was significantly decreased in the CUMS + Lys group compared to the CUMS + Saline (P = 0.0018), CUMS + Lys + RAPA (P = 0.0021) and CUMS + RAPA groups (P < 0.0001) (F (4, 35) = 11.30). (Fig. 4D). No significant group differences were observed in anxiety-like measures in the OFT (F (4, 35) = 1.809) (Fig. S2A) or EPM (F (4, 35) = 0.6144) (Fig. S2B). Collectively, lysine alleviated depressive-like behaviors, while RAPA aggravated them and blocked lysine’s beneficial effects.

Fig. 4. Study design, and behavioral study of chronic unpredictable mild stress model in juvenile rats.

A Schematic representation of the animal experimental design including five groups: CON + Saline, CUMS + Saline, CUMS + Lys + RAPA, CUMS + Lys and CUMS + RAPA (n = 8/group). B Relative body weight analysis in CON + Saline, CUMS + Saline, CUMS + Lys + RAPA, CUMS + Lys and CUMS + RAPA (n = 8/group). C Sucrose preference in SPT of CON + Saline (mean ± SD: 0.92 ± 0.03), CUMS + Saline (mean ± SD: 0.79 ± 0.09), CUMS + Lys + RAPA (mean ± SD: 0.78 ± 0.12), CUMS + Lys (mean ± SD: 0.9 ± 0.03) and CUMS + RAPA (mean ± SD: 0.78 ± 0.06) (n = 8/group). D Immobility time in FST of CON + Saline (mean ± SD: 96.63 ± 43.55), CUMS + Saline (mean ± SD: 161.79 ± 31.78), CUMS + Lys + RAPA (mean ± SD: 160.66 ± 44.57), CUMS + Lys (mean ± SD: 84.99 ± 33.4) and CUMS + RAPA (mean ± SD: 184.33 ± 29.08) (n = 8/group). (*): CON + Saline compared to CUMS + Saline, (α): CON + Saline compared to CUMS + Lys + RAPA, (β): CON + Saline compared to CUMS + Lys, (@): CON + Saline compared to CUMS + RAPA. Graphs were plotted using GraphPad Prism. Data are expressed as mean ± standard deviation (mean ± SD), and the results were analyzed by one-way ANOVA and Wilcoxon rank-sum test. P < .05 was considered statistically significant. (*) P < .05, (**): P < .01, (***): P < .001, (****): P < .0001; (@): P < .05, (@@): P < .01, (@@@): P < .001, (@@@@): P < .0001; (α): P < .05, (αα): P < .01, (ααα): P < .001, (αααα): P < .0001; (β): P < .05, (ββ): P < .01, (βββ): P < .001, (ββββ): P < .0001.

To further examine whether autophagy mediates the antidepressant effect of lysine, we analyzed autophagy-related protein expression in the PFC using Western blotting. Compared to the CON + Saline group, rats in the CUMS + Saline (P = 0.0002, P = 0.0843, P = 0.0016), CUMS + Lys + RAPA (P = 0.2017, P = 0.2419, P = 0.0261) and CUMS + RAPA groups (P < 0.0001, P = 0.0004, P < 0.0001) exhibited reduced protein expression levels of SQSTM1/p62 and p-mTOR (Ser2448), and increased expression of Beclin1. While compared to the CUMS + Saline (P = 0.0043, P = 0.0404, P = 0.0010), CUMS + Lys + RAPA (P = 0.8519, P = 0.1302, P = 0.0174) and CUMS + RAPA groups (P = 0.0011, P = 0.0002, P < 0.0001), the CUMS + Lys group exhibited increased protein expression levels of SQSTM1/p62 (F (4, 25) = 12.57) and p-mTOR (Ser2448) (F (4, 25) = 9.302), and reduced expression of Beclin1 (F (4, 25) = 18.57). However, there were no significant differences were observed between the CON + Saline and the CUMS + Lys groups in these proteins (Fig. 5A). Immunofluorescence staining of SQSTM1/p62 and neuronal marker NeuN in PFC tissues further revealed that SQSTM1/p62 expression was markedly decreased in the CUMS + Saline (P = 0.0002), CUMS + Lys + RAPA (P < 0.0001) and CUMS + RAPA groups (P < 0.0001) compared to the CON + Saline group, while increased protein expression levels of SQSTM1/p62 was observed in the CUMS + Lys group, compared to the CUMS + Saline (P = 0.0023), CUMS + Lys + RAPA (P = 0.0003) and CUMS + RAPA groups (P = 0.0004) (F (4, 25) = 16.80) (Fig. 5B). Consistently, transmission electron microscopy revealed a higher number of autophagic lysosomes in the PFC of the CUMS + Saline (P = 0.0173), CUMS + Lys + RAPA (P = 0.0032) and CUMS + RAPA groups (P = 0.0018) compared to the CON + Saline group. Whereas the CUMS + Lys group exhibited a lower number of autophagic lysosomes, compared to the CUMS + Saline (P = 0.0125), CUMS + Lys + RAPA (P = 0.0024) and CUMS + RAPA groups (P = 0.0013) (F (4, 10) = 15.62) (Fig. 5C). Together, these results indicate that lysine ameliorates depressive-like behaviors in CUMS rats by normalizing PFC autophagy through the mTOR/Beclin1 signaling pathway, while RAPA abrogates these effects.

Fig. 5. Lysine reversed the expression of autophagy proteins in the PFC of depressed rats, and RAPA abolished the effect of lysine on autophagy.

A Representative bars and Western blots showing p-mTOR(ser2448), mTOR, Beclin1, SQSTM1/p62 and LC3 protein expression levels in the PFC of CON + Saline (mean ± SD: 1 ± 0.07, 1 ± 0.07, 1 ± 0.14, 1 ± 0.08,1 ± 0.08), CUMS + Saline (mean ± SD: 0.73 ± 0.11, 0.98 ± 0.1, 1.26 ± 0.1, 0.57 ± 0.17, 1.06 ± 0.13), CUMS + Lys + RAPA (mean ± SD: 0.67 ± 0.12, 0.86 ± 0.15, 1.19 ± 0.1, 0.82 ± 0.16, 1.22 ± 0.14), CUMS + Lys (mean ± SD: 0.94 ± 0.1, 0.94 ± 0.19, 0.99 ± 0.1, 0.9 ± 0.12, 0.91 ± 0.18) and CUMS + RAPA (mean ± SD: 0.45 ± 0.11, 0.91 ± 0.14, 1.41 ± 0.05, 0.53 ± 0.16, 1.19 ± 0.26) (n = 6/group). B Immunofluorescence staining (40× magnification) and representative bars of SQSTM1/p62 expression in the PFC of CON + Saline (mean ± SD: 1 ± 0.17), CUMS + Saline (mean ± SD: 0.57 ± 0.16), CUMS + Lys + RAPA (mean ± SD: 0.51 ± 0.13), CUMS + Lys (mean ± SD: 0.92 ± 0.13) and CUMS + RAPA (mean ± SD: 0.52 ± 0.11) (n = 6/group). C Transmission electron microscopic (TEM) representative picture of the PFC in CON + Saline (mean ± SD: 2.67 ± 0.01), CUMS + Saline (mean ± SD: 6.83 ± 0.71), CUMS + Lys + RAPA (mean ± SD: 7.83 ± 0.71), CUMS + Lys (mean ± SD: 2.83 ± 0.71) and CUMS + RAPA (mean ± SD: 8.5 ± 3.54) (n = 3/group), with arrows indicating autophagosomes. Images from two independent experiments were analyzed using ImageJ software. Graphs were plotted using GraphPad Prism. Data are expressed as mean ± standard deviation (mean ± SD), and the results were analyzed using one-way ANOVA and Wilcoxon rank-sum test. P < .05 was considered statistically significant. (*) P < 0.05, (**): P < 0.01, (***): P < 0.001, (****).

Discussion

Here, we demonstrated that FMT from adolescents with MDD into antibiotic-treated recipient rats induced depressive-like behaviors, reduced lysine levels, and disrupted autophagy regulation in the PFC. We then employed a CUMS model with dietary lysine restriction, confirming that lysine deficiency contributed to autophagy hyperactivation and the development of depressive-like behaviors. Finally, in CUMS-induced depressed rats, lysine supplementation was shown to restore normal autophagic function, alleviate behavioral abnormalities and correct PFC autophagy dysfunction via modulation of the mTOR/Beclin1-autophagy pathway. In summary, our findings provide evidence that lysine supplementation exerts antidepressant-like effects by rebalancing the autophagy pathway and restoring mTOR signaling, positioning lysine as a potential therapeutic target for adolescent depression.

It should be noted that the two animal models employed in this study serve distinct but complementary purposes. FMT model was specifically designed to investigate whether gut microbiota derived from adolescents with MDD are sufficient to induce depressive-like behaviors, lysine deficiency, and the PFC autophagy dysregulation in recipient rats. In contrast, the CUMS model was used as a mechanistic validation system to test whether lysine deficiency-driven autophagy dysregulation causally contributes to depressive-like behaviors, independent of the initial upstream trigger. Although we did not perform microbiome sequencing in CUMS-exposed rats in the present study, multiple previous reports have demonstrated that CUMS induces significant alterations in gut microbiota composition and metabolic function [27, 28], including changes in taxa associated with depression and disruptions in amino acid metabolism. Importantly, recent evidence from pediatric cohorts further supports a link between depression-associated microbiota alterations and amino acid deficiency. In a multi-omics study of children and adolescents with MDD, Teng et al. reported a significant enrichment of Akkermansia species in patients compared with healthy controls, which was accompanied by reduced amino acid availability and depressive phenotypes. Notably, transplantation of this microbiota into rodents recapitulated both the increase in Akkermansia abundance and amino acid deficiency, highlighting cross-species consistency with the elevated Akkermansia muciniphila observed in our FMT-MDD rats [16]. Importantly, however, stress-induced microbiota changes are highly context-dependent and do not necessarily recapitulate the microbial profiles observed in depression-associated FMT models [29, 30]. Therefore, rather than being redundant, the FMT and CUMS paradigms in this study address different levels of the microbiota-lysine-autophagy axis: the former establishes microbiota-driven causality, whereas the latter confirms the functional importance of lysine-regulated autophagy in depressive-like phenotypes.

Autophagy, a conserved lysosomal degradation pathway, plays a dual role in central nervous system (CNS) homeostasis, contributing to neuronal quality control under physiological conditions while exacerbating neuropsychiatric pathology when dysregulated. Previous studies have reported altered autophagy signaling in multiple brain regions of depressed patients and animal models, including increased LC3-II/LC3-I ratio, reduced SQSTM1/p62 reduction, and elevated Beclin1 expression [31, 32]. Importantly, developmental stage appears tmodulate baselinene autophagy. During adolescence active synaptic pruning and cortical remodelig, are accompanied by heightened basal autophagic flux [33], potentially increasing vulnerability to stress-induced dysregulation. Our results align with this concept: FMT from adolescents with MDD triggered pronounced PFC autophagy hyperactivation in recipient adolescent rats, coinciding with depressive-like behaviors.

The gut microbiota’s role in regulating autophagy is increasingly recognized. AMPK-mTOR-ULK1 signaling is a central regulator of autophagy [34]. Microbiota-derived metabolites such as tryptophan catabolites have been shown to modulate this pathway and influence autophagic flux [35]. Our study extends this paradigm by identifying lysine as a critical microbiota-derived metabolite in adolescent depression. Lysine is an essential amino acid required for protein synthesis, neurotransmitter production, and mTOR pathway activation [36, 37]. Notably, although metagenomic functional annotation did not reveal significant differences in canonical microbial lysine biosynthesis pathways between FMT-MDD and FMT-HCs groups, this does not exclude a microbiota-driven reduction in host lysine availability. Increasing evidence indicates that gut microbiota regulate host amino acid homeostasis through mechanisms beyond direct biosynthesis, including modulation of intestinal barrier function, competition for dietary amino acids, regulation of amino acid transporters, and alterations in systemic metabolic utilization [38–40]. In our study, Compared to the functional potential seen in healthy microbiota, the profile in FMT-MDD rats was less pronounced in several metabolic categories, including genes encoding carbohydrate-active enzymes (e.g., K01190, K05349), nutrient-binding proteins (e.g., K21572), and transporters (e.g., K02013). This reflects a functionally attenuated state with compromised nutrient acquisition and environmental engagement. Meanwhile, we observed reduced lysine concentrations in both the plasma of adolescents with MDD and the prefrontal cortex of FMT-MDD rats, suggesting that lysine deficiency reflects altered host–microbiota metabolic interactions rather than a single microbial pathway defect [41]. In both MDD patients and FMT-MDD rats, we observed lysine deficiency alongside excessive autophagy, prompting the hypothesis that diminished lysine availability may impair mTOR activation and disinhibit autophagy. Supporting this, in our CUMS model lysine supplementation reversed PFC autophagy hyperactivation and alleviated depressive-like behaviors—effects abolished by autophagy activator RAPA.

The PFC appears particularly susceptible to autophagy-related alterations in depression. Chronic stress has been shown to cause excessive PFC autophagy leading to dendritic spine loss and impaired synaptic plasticity [42, 43]. Our TEM and immunofluorescence data confirmed elevated autophagosome/autolysosome formation and reduced SQSTM1/p62 reduction in the PFC of FMT-MDD rats, reinforcing the region-specific vulnerability of the frontal cortex in adolescent depression. Given that rs-fMRI studies have linked PFC connectivity deficits to amino acid metabolic changes in pediatric MDD [16], our findings suggest that lysine-dependent autophagy modulation may partly underlie these neurofunctional alterations.

Targeting autophagy in depression remains challenging, as both insufficient and excessive activity can be detrimental to neuronal integrity and synaptic function [31, 44]. Strategies to reduce pathological autophagy include inhibiting upstream initiators, such as ULK1 and Beclin1, or blocking autophagosome–lysosome fusion using agents like chloroquine, which reduces autophagy-mediated neurotoxicity in preclinical models [44]. Nutrient-sensing pathways offer an alternative approach: activation of the mechanistic target of RAPA (mTOR)—a central negative regulator of autophagy—can restorbalanced fluxux and promote neuronal survival [45]. Nutritional interventions are particularly attractive as non-invasive modulators of mTOR and autophagy. Essential amino acids, especially branched-chain amino acids and lysine, activate mTOR complex 1 (mTORC1), and suppress autophagy in neurons and peripheral tissues [46]. Lysine not only supports protein synthesis and neurotransmitter production but alsactsct as a metabolic signal influencing mTOR activity [37]. Prior studies show lysine supplementation can attenuate stress-induced autophagy overactivation by enhancing mTOR and p70S6K phosphorylation, reducing autophagosome formation, and preserving synaptic structures [36]. Other indirect mTOR-enhancing strategies include insulin/IGF-1 pathway activation, AMPK modulation, or neurotrophic factor administration [47]. Given mTOR’s role in regulating autophagy, neuroplasticity, inflammation, and metabolism, lysine supplementation may provide dual therapeutic benefits in depression: reducing pathological autophagy while enhancing synaptic plasticity and neuronal resilience. Lysine and the mTOR autophagy pathway are positioned as a promising target for future therapeutic strategies in adolescent MDD.

Several limitations should be acknowledged. First, although our findings implicate gut microbiota-driven lysine deficiency in PFC autophagy hyperactivation, most evidence comes from rodent studies and cross-sectional human data; longitudinal and interventional research is needed for clinical validation. Second, while we identified lysine as a mediator of microbiota–autophagy interactions, other amino acids and metabolic pathways likely contribute to adolescent MDD and were not systematically assessed. Third, FMT transfers a complex microbiota, preventing attribution of effects to specific taxa; gnotobiotic or defined-consortium approaches are needed to establish causality.. Fourth, the mTOR-autophagy axis is context-dependent and lysine’s antidepressant effects may involve additional mechanisms such as neurotransmitter or immune modulation. Finally, the human sample size though comparable to prior adolescent depression microbiome studies, may limit the generalizability; larger multi-center cohorts and integrative multi-omics will strengthen future conclusions.

In conclusion, this study identifies excessive PFC autophagy as a mechanistic link between adolescent gut microbiota dysbiosis and depression, mediated in part by lysine deficiency and downstream mTOR inhibition. These findings highlight the importance of metabolites in microbiota–brain axis research and suggest that targeted nutritional interventions, such as lysine supplementation, may complement existing antidepressant strategies for adolescent MDD. Future studies should investigate the translational potential of modulating amino acid–autophagy pathways in clinical populations and explore whether combined microbiota and dietary interventions can yield synergistic therapeutic benefits.

Author contributions

This study was designed for XZ and CW. JZ, FC, TT, LZ, and LZ recruited and supervised the participants and performed all clinical procedures. JZ, FC, and LZ performed the preparation, processing, and metagenomic sequencing of fecal samples. JZ and TT analyzed the metagenomic sequencing data. JZ, FC, LZ, BQ, XX, YL, HC, CL, and PL performed the experiments on rats. JZ performed the RT-qPCR experiments. JZ, FC, and LZ performed Western blotting experiments. JZ performed Immunofluorescence experiments. JZ and QL performed ELIA experiments. JZ and FC prepared rat tissue samples for TEM experiments. JZ, TT, CW, and XZ prepared the manuscript. All the authors approved the manuscript.

Funding

This work was supported by STI2030-Major Projects (2022ZD0212900), the National Key R&D Program of China (2024YFC2707800), the National Natural Science Foundation of China (82271565 and 82301714), New Chongqing Youth Innovation Talent Project (CSTB2025YITP-QCRCX0084), Chongqing Medical Scientific Research Project: Joint Project of Chongqing Health Commission and Science and Technology Bureau (2026QNKX006).

Data availability

The metagenomic sequencing data generated in this study have been deposited in the NCBI Sequence Read Archive (SRA) under accession number SRP619016. The datasets generated and analyzed during the current study, including behavioral, molecular, and imaging data, are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interest.

Ethics approval

This study involving human participants was approved by the Ethics Committee of the First Affiliated Hospital of Chongqing Medical University (approval number: 2024-129-01). Written informed consent was obtained from all participants and their legal guardians prior to enrollment. All methods were performed in accordance with the relevant guidelines and regulations. All animal experimental procedures were approved by the Institutional Animal Care and Use Committee of Chongqing Medical University (IACUC-CQMU-2024-0040) and were conducted in accordance with the relevant institutional guidelines and regulations.

Supplementary information

The online version contains supplementary material available at 10.1038/s41398-026-04095-2.