Gut microbiota-derived metabolites mediate the neuroprotective effect of melatonin in cognitive impairment induced by sleep deprivation

All animal experiments in this study were approved by the Animal Welfare Committee of the Agricultural Research Organization, China Agricultural University (approval no. CAU201709112). Male ICR mice (8 weeks old; 35–40 g) were purchased from the Beijing Vital River Laboratory (Beijing, China). The mice were placed in cages and maintained under standard environmental conditions of temperature (21 ± 1 °C) and relative humidity (50 ± 10%), with a regular 14-h light/10-h dark cycle. The light was turned on at 7:00 h and turned off at 21:00 h. All mice were acclimated for 1 week before the experiments.

To prepare FMT material, fresh feces were collected from donor mice and immediately diluted with sterile PBS. PBS (1 mL) was used to dilute 50 mg of fecal pellets. Briefly, the stool was steeped in sterile PBS for approximately 15 min, shaken, and then centrifuged at 1000 rpm and 4 °C for 5 min. The supernatant was collected and centrifuged at 8000 rpm and 4 °C for 5 min. This supernatant was discarded, and the bacteria were retained and resuspended in PBS and filtered twice. The final bacterial suspension was mixed with an equal volume of 40% sterile glycerol to a final concentration of 20% and stored at –80 °C until transplantation [31]. For each mouse, 100 μL of bacterial suspension (10⁸ colony-forming unit [CFU]/mL) was transplanted into each recipient mouse by gavage each day for 14 consecutive days [32, 33]. Mice in the V-FMT group received an equal volume of sterile PBS containing 20% glycerol.

To investigate the effects of A. veronii on cognitive impairment induced by SD, the mice were divided into three groups: the control (CON), A. veronii colonization (Aero), and A. veronii colonization supplemented with Mel (20 mg/kg) (A + Mel) groups (Fig. 1B). The Aero and A + Mel groups were provided with drinking water containing the same antibiotics used for FMT recipient mice for 10 days. Aero and A + Mel mice were then orally gavaged with 10⁸ CFU/mL of A. veronii in 0.1 mL PBS on the 11th day at 8 am. The CON group was orally gavaged with 0.1 mL PBS. In the A + Mel group, 20 mg/kg Mel was administered to mice via intraperitoneal injection 60 min before A. veronii colonization. CON and Aero mice were intraperitoneally injected with an equal volume of sterile saline containing 2% ethanol.

To investigate the effects of LPS on cognitive impairment induced by SD, mice were divided into the control (CON), LPS (LPS), LPS supplemented with Mel (20 mg/kg) (LPS + Mel), and LPS supplemented with the TLR4 inhibitor TAK-242 (LPS + TAK-242) groups (Fig. 1C). The LPS-supplemented mice were administered intraperitoneal injections of LPS (27,840; Sigma, St. Louis, MO) 2 mg/kg every day, a single dose per day at 8:00 am. For treatment with Mel and TAK-242, 20 mg/kg Mel (LPS + Mel) and 150 mg/kg TAK-242 (LPS + TAK-242) were administered to mice by intraperitoneal injections 60 min before LPS supplementation as a single dose once per day for 3 days. Mice in the CON, LPS, and LPS + TAK-242 groups were intraperitoneally injected with an equal volume of sterile saline containing 2% ethanol.

To investigate the effects of butyrate on cognitive impairment induced by SD, mice were divided into six groups: the SD (SD), SD + antibiotics (SD + Abs), SD + Mel supplementation (SD + Mel), SD + antibiotics + Mel supplementation (SD + Abs + Mel), SD + antibiotics + butyrate supplementation (SD + Abs + butyrate), and non-sleep-deprived control (CON) groups (Fig. 1D). The SD-treated mice were administered intraperitoneal injections of Mel 0 mg/kg (vehicle, SD, and SD + Abs) and 20 mg/kg (SD + Mel and SD + Abs + Mel) once, 60 min before SD, and a single dose per day at 7:00 am for a total of 3 days. For treatment with butyrate, 40 mM sodium butyrate (Sigma-Aldrich, St. Louis, MO, USA) was administered orally to mice by gavage (SD + Abs + Butyrate), 60 min before SD, and a single dose per day at 7:00 am for a total of 3 days. Mice in the CON, SD, SD + Abs, and SD + Abs + butyrate groups were intraperitoneally injected with an equal volume of sterile saline containing 2% ethanol. For substantial depletion of the microbiota, mice in the SD + Abs, SD + Abs + Mel, and SD + Abs + butyrate groups were provided drinking water containing the same antibiotics used for the FMT recipient mice for 10 days.

All mice were euthanized under anesthesia using 10% chloral hydrate after the experiment ended at 8:00 am on the final day. Hippocampal tissue, colonic contents, and fecal samples were collected. This experiment was repeated twice.

Feces were collected in tubes and diluted with a 10-weight volume of PBS. Each tube was vigorously vortexed and centrifuged for 10 min at 800 rpm. The supernatant was serially diluted 10²–10⁸ fold, and aliquots were streaked with a cell spreader on brain–heart infusion agar. After overnight incubation at 37 °C, CFUs were determined [34].

The Morris water maze (MWM) test was used to assess spatial learning and memory. The maze consisted of a round tank (120 cm in diameter, 50 cm in height) filled with warm water (23 ± 1 °C). Black nontoxic carbon ink was added to make the water opaque. The pool was divided into four quadrants (I, II, III, and IV). A moveable, hidden, circular platform was placed at a fixed location in quadrant IV and submerged approximately 1 cm below the water surface.

After acclimatization for 1 week, all mice were placed in a water pool without a platform for 1 min and allowed to swim. All experiments were performed at 8:00 a.m. To minimize the effects of stress on the experimental outcomes, behaviorally and physically healthy mice without any stereotypical characteristics were selected for further study. On the second day, each mouse was placed in water in all four quadrants in a fixed order to perform four training trials per day. The maximum trial duration was 60 s. Mice that failed to locate the hidden platform were manually guided. Once they reached the platform, they were allowed to remain there for 15 s. All mice received this training for five consecutive days. The mice were subjected to behavioral tests (training and detection periods) after FMT or A. veronii experiments. The mice in the butyrate and LPS supplementation groups were first subjected to behavioral training and then assessed after the test treatments.

MWM parameters included latency (s), path length (m), path velocity (mm/s) to reach the hidden platform, time spent in the target zone, and the number of crossings over the previous platform location when the platform was removed. The experiment was performed at the same time of the day, under the same environmental conditions. The animal movement was tracked using a computerized tracking system (XR-XM101; Shanghai Softmaze Information Technology Co., Ltd., China).

BV2 immortalized murine microglia cells were maintained at 37 °C in a 5% CO₂ humidified incubator in Dulbecco’s modified Eagle’s medium (Gibco, Franklin Lakes, NJ, USA) supplemented with 10% fetal bovine serum and 100 U/mL penicillin and streptomycin (Gibco). The cells were cultured in 12-well plates (5 × 10⁵ cells/mL). Some cells treated with 200 μM LPS were also treated with 5 mM butyrate (Sigma-Aldrich; LPS + butyrate). After butyrate supplementation for 30 min, some LPS + butyrate cells were sequentially treated with TLR4 inhibitor (10 μM TAK-242; MedChemExpress, Monmouth Junction, NJ, USA; LPS + TAK-242-cells), MCT1 inhibitor (10 nM AZD3965; MedChemExpress; LPS + butyrate + AZD3965), HDAC3 agonist (50 μM ITSA-1; MedChemExpress; LPS + butyrate + ITSA-1-cells), or nuclear factor-kappa B (NF-κB) antagonist (100 μM pyrrolidine dithiocarbamate [PDTC]; MedChemExpress; LPS + PDTC cells). Each plate of treated cells was incubated for 24 h. To generate conditioned media, BV2 cells treated with drug were grown in a serum-free growth medium for 24 h. Culture supernatants were collected, then added to HT-22 cells, and cultured in a humidified 5% CO₂/95% air environment at 37 °C for 24 h. After BV2 microglia were treated under different conditions, culture supernatants were collected for ELISA analysis. BV2 cells were collected for western blotting for the detection of signaling pathway proteins (HDAC3, p-IκB, and p-P65). HT22 cells were collected for western blot analysis of cleaved caspase-3 levels. Each assay was repeated eight times.

The hippocampus tissues, BV2 cells, and HT-22 cells were rapidly isolated and lysed in RIPA lysis buffer (CW2333S; CWBIO, Beijing, China) containing 1% protease inhibitor cocktail (CW2200S; CWBIO, Beijing, China) and 1% phosphatase inhibitor cocktail (CW2383S; CWBIO, Beijing, China). The lysates were centrifuged at 14,000 × g for 15 min at 4 °C. The supernatants were collected, and the amount of protein was measured using a bicinchoninic acid kit (CW0014; CWBIO, Beijing, China), before the protein concentration was standardized. The protein samples were resolved using 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and electro blotted onto a polyvinylidene fluoride membrane (Millipore; Billerica). Nitrocellulose membranes were blocked for 60 min using TBST (a mixture of Tris-buffered saline [TBS] and 0.05% Tween-20) containing 5% fat-free dry milk. They were then incubated in rabbit primary antibodies (TLR4, 1:1000, Abcam; cleaved caspase-3, 1:1000, CST; HDAC3, 1:1000; p-IκB,1:1000; p-P65, 1:1000; β-actin,1:8000, Abcam) overnight at 4 °C. After washing in TBST, they were incubated in horseradish peroxidase conjugated goat anti-rabbit IgG (1:5000; CW0103; CWBIO, Beijing, China) for 2 h at 37 °C. The protein bands were detected using an enhanced chemiluminescence kit (CW0049; CWBIO, Beijing, China). The protein band intensities were quantified using ImageJ software (version 1.4, National Institutes of Health, Bethesda). The protein level was normalized to the density ratio of β-actin, while the relative protein level in the CON group in vivo or in the control cells in vitro was defined as 100%. Each sample was assayed three times.

Paraffin sections were incubated in rabbit anti-Iba1 (1:500; ab178846; Abcam) primary antibody overnight at 4 °C. The sections were then rinsed in 0.01 M PBS (pH 7.4) and incubated in biotinylated goat anti-rabbit IgG (1:300, sc-2020; Santa Cruz) for 2 h at room temperature. After washing, the tissues were incubated in streptavidin–horseradish peroxidase (1:300; Vector Laboratories, Burlingame) for 2 h at room temperature. Immunoreactivity was visualized by incubating the tissue sections in 0.01 M PBS containing 0.05% DAB (Sigma) and 0.003% H₂O₂ for 10 min in the dark. The sections were then stained with hematoxylin and mounted. Control slides without the primary antibody were examined in all cases. Immunoreactive cells presented with yellow–brown staining in the cytoplasm. The localization and distribution of immunoreactive positive materials in the hippocampus were observed using a microscope (BX51; Olympus). For each mouse, representative coronal brain sections with similar coordinates (Bregma, 2.0 mm) were selected, five slices each from five animals/group were used in the analysis. ImageJ software (version 1.4, National Institutes of Health, Bethesda) was used to analyze the levels of microglia. Results are expressed as a mean integral optical density (IOD) of Iba1-positive cells, normalized to that in controls.

Mice feces sample DNA was extracted using the stool DNA Kit (DP328-02, TIANGEN) according to the manufacturer’s instructions. PCR amplification was performed using the AceQ qPCR SYBR green master mix (Q111-02; Vazyme Biotech, USA). Primers specific to 16S ribosomal RNA were used as an endogenous control to normalize loading between samples. The relative amount of 16S ribosomal DNA in each sample was estimated using the ΔΔCT. Primer sequences were designed using Primer-BLAST. RT-PCR primers sequences are as follows:Aeromonas: Fwd-5’GCGACTTCAAGCTGCAAGAG 3′, Rev 5’TTCAGTCGCTCGATGGTCTG 3’.Bacteria 16S: Fwd-5’TCCTACGGGAGGCAGCAGT 3’, Rev 5’GGACTACCAGGGTATCTAATCCTGTT 3’.

Hippocampus samples or culture supernatants of BV2 cells were collected for the detection of inflammatory factors (TNF-α, IL-6, IL-4, and IL-10) concentrations using a competitive ELISA (Uscn Life Science, Inc., Wuhan, China). The ELISA kits used to detect the levels of mouse LPS was purchased from Jianglai Industrial Limited By Share Ltd, Shanghai, China. All the tests were performed according to the manufacturer’s instructions. Each sample was tested in triplicate. The intra-assay coefficient of variation (CV) was < 10% and the inter-assay CV was < 12%.

Fresh colon contents were obtained from the mouse colon using 4 mL of sterile PBS. After centrifugation (1500 rpm, 5 min), the supernatant was discarded, and all samples were stored at − 80 °C until gut microbial analysis. Genomic DNA from the colon tissue of 21 mice was isolated using a PowerSoil DNA Isolation kit (MO BIO Laboratories, Carlsbad, CA, USA) according to the manufacturer’s instructions. The concentration and purity of isolated DNA were quantified using a Synergy HTX Multi-Mode Reader (Gene Company Ltd., Hong Kong, China). The V1-V9 region of the bacterial 16S rRNA gene was amplified using universal primers (27F: AGRGTTTGATYNTGGCTCAG; 1492R: TASGGHTACCTTGTTASGACTT) and purified with MagicPure® size selection DNA beads (TANGEN Biotech Corporation Ltd, Beijing, China). The abundance and diversity of the gut microbiota in mice were measured using the PacBio sequencing platform (Biomarker Technologies, Beijing, China). Circular consensus sequencing (CCS) was obtained from the raw subreads following minPasses ≥ 5 and minPredictedAccuracy ≥ 0.9 (SMRT Link version8.0). Then, 1200–1650 bp CCS was filtered into high-quality clean tags using Lima version 1.7.0, and chimera sequences were detected and removed using the UCHIME algorithm. Finally, the qualified sequences were clustered at a 97% similarity level using USEARCH (version 10.0), and 0.005% of the total sequences were identified as quality-filtered sequences to generate the operational taxonomic units (OTUs). Taxonomy was assigned to the OTUs using the SILVA database (v.123) with the RDP classifier at a 70% confidence threshold [35]. Alpha diversity indices, including the ACE, Chao1, Shannon, and Simpson indices, were calculated using QIIME2 software (Version 2020.8). Statistical significance between groups was determined by one-way analysis of variance (ANOVA). Beta diversity analysis was performed to investigate the structural variation in microbial communities across samples using binary Jaccard distance metrics and visualized via principal coordinate analysis (PCoA). Differences in the binary Jaccard distances among the groups were determined using analysis of similarities (ANOSIM). Linear discriminant analysis (LDA) effect size (LefSe) [36] was used to identify representative species. LDA was performed from the phylum to the genus level. LDA scores ≥ 3.0 and p < 0.05 were considered signature taxa and selected for plotting and further analysis. The Spearman rank correlation test was used (R package “psych”) to analyze the correlations between the signature microbial taxa and phenotypic variables. FDR adjusted p < 0.2 was considered statistically significant and visualized using the R package “corrplot”. Sequencing data were deposited in the Sequence Read Archive of the National Center for Biotechnology Information (Bioproject: PRJNA826223) for publication.

Colon contents (100 µL) were added to 500 μL of extract containing an internal standard (1000:2; volume ratio of methanol to acetonitrile = 1:1; internal standard concentration 2 mg/L). The solutions were mixed by vortexing for 30 s. After centrifuging at 12,000 rpm for 15 min, the supernatant was collected for liquid chromatography–mass spectrometry (LC–MS) analysis. High-resolution mass spectral data were obtained using a UPLC Acquity I-Class PLUS system coupled with a Xevo G2-XS QTof (Waters, Wilmslow, UK) and Acquity UPLC HSS T3 (1.8 µm, 2.1 mm × 100 mm; Waters). As the chromatographic parameters, the mobile phase consisted of (A) aqueous 0.1% formic acid and (B) methanol with the addition of 0.1% formic acid; gradient program: 0 min (2% B), 0.25 min (2% B), 10 min (98% B), 13 min (98% B), 13.1 min (2% B), and 15 min (2% B); flow rate, 400 μL/min; injection volume, 1 μL. The MS parameters were set at an ion spray voltage of 2000 V (positive) and − 1500 V (negative), cone voltage of 25 V, ion source temperature of 150 °C, desolvation temperature of 500 °C, and desolvation gas flow of 800 L/h. The LC–MS/MS raw data were processed using MassLynx software (V4.2, Waters) and then imported into Progenesis QI software (version 2.3) for peak alignment and selection. Metabolites were identified using retention time, exact mass, and tandem MS data against METLIN and a self-built database (Biomarker Technologies Corporation, Beijing). Theoretical fragments were used for MS/MS identification.

Fresh fecal contents (n = 8) samples were collected and stored at − 80 °C. Fecal samples were mixed with water and centrifuged. The supernatant was filtered and mixed with ether and sulfuric acid. After high-speed centrifugation, the ether layer was collected, and SCFA concentrations were measured using a model 6890 N gas chromatograph (Agilent, San Diego, CA, USA). The content is expressed as micrograms per milligram.

The data were expressed as the mean ± standard error and analyzed using GraphPad Prism version 9 (GraphPad Software, La Jolla, CA, USA). Experiments were performed at least in five independent biological and at least two independent technical replicates. Differences between groups were analyzed using one-way ANOVA followed by Turkey’s multiple comparisons tests. All p-values < 0.05 were considered statistically significant.

To investigate whether neuroinflammation was induced by SD-derived gut microbiota, we evaluated changes in inflammatory factors and microglia immunohistochemical staining in the hippocampus. The integrated optical density (IOD) of Iba1-positive cells in the hippocampal cornu ammonis (CA)1, CA3, and dentate gyrus (DG) areas was 51.9% (p < 0.001), 27.6% (p = 0.01), and 32.3% (p = 0.002) higher in the SD-FMT group than in the CON-FMT group, respectively (Fig. 2J,K). We also observed significant increases in IL-6 (36.5%, p = 0.03) and TNF-α (47.2%, p = 0.01) levels and a significant decrease in IL-4 (49.6%, p = 0.04) and IL-10 (35.5%, p < 0.001) levels in the hippocampus of SD-FMT mice compared to CON-FMT mice (Fig. 2L–O). However, FMT of the “SD + Mel microbiota” significantly suppressed the activation of microglial cells, increased pro-inflammatory factors, and decreased anti-inflammatory factors. Excessive inflammation can cause cytotoxicity, interfere with cell growth, and induce apoptosis. Therefore, we determined the expression levels of cleaved caspase-3, Bax, and Bcl-2 using western blotting. Compared with the CON-FMT group, there was a significant downregulation in the expression of Bcl-2 (p = 0.021) and a significant upregulation in the expression of cleaved caspase-3 (p = 0.002) and Bax (p = 0.001) in the SD-FMT group. However, the FMT of the “SD + Mel microbiota” reversed these changes (Supplemental Fig. 1). The FMT experiments suggest that the gut microbiota is required for the protective effect of Mel on neuroinflammation, apoptosis, and memory impairment induced by SD.

To identify the specific bacterial phyla associated with CON-FMT, SD-FMT, and SD + Mel-FMT groups, LDA and LEfSe were performed to identify the core taxa most likely to explain the differences between groups. As shown in Fig. 4C, Bacteroidetes and Proteobacteria were more abundant in the SD-FMT group than in the CON-FMT and SD + Mel-FMT groups (Bacteroidetes, p = 0.046, LDA score = 4.39; Proteobacteria, p = 0.012, LDA score = 3.98). Additionally, as shown in Fig. 4A, Firmicutes was decreased in the SD-FMT group compared to the CON-FMT and SD + Mel-FMT groups (p > 0.05). Furthermore, LEfSe analysis identified 64 taxa biomarkers in the three groups with an LDA score > 3 and p < 0.05. The relative abundances of Lachnospiraceae_NK4A136_group (p = 0.041, LDA score = 4.55), Eubacteriumxylanophilum_group (p = 0.027, LDA score = 3.39), Ruminococcus_1 (p = 0.018, LDA score = 3.56), and Lachnospiraceae_A2 (p = 0.001, LDA score = 3.64) were significantly lower in the SD-FMT group than in the CON-FMT and SD + Mel-FMT groups (Fig. 4G–J). In addition, the relative abundance of Turicimonas (p = 0.035, LDA score = 3.12) was significantly higher in the SD-FMT group than in the CON-FMT and SD + Mel-FMT groups (Fig. 4K), whereas there was no significant difference between the CON-FMT and SD + Mel-FMT groups (p > 0.05).

To investigate whether butyrate could mediate improved SD-induced neuroinflammation and apoptosis attributed to Mel, we examined the changes in the expression of Iba1 and the release of inflammatory cytokines and intracellular signaling proteins in the hippocampus. Compared with the CON group, neuroinflammation was evident in the SD and SD + Abs groups, which showed a significant increase in the IOD of Iba1-positive cells in the hippocampal CA1, CA3, and DG (p < 0.05; Fig. 9I–L) and IL-6 and TNF-α levels and a significant decrease in IL-4 and IL-10 levels (p < 0.05; Fig. 9M–P). For intracellular signaling proteins, there was an obvious upregulation in the expression of HDAC3 (p < 0.05; Fig. 9Q), p-IκB (p < 0.05; Fig. 9R), p-P65 (p < 0.05, Fig. 9S), and cleaved caspase-3 (p < 0.05, Fig. 9T) in the SD and SD + Abs groups compared to that in the CON group. Conversely, Mel supplementation reversed the changes in neuroinflammation and apoptosis caused by SD. The above indicators revealed no significant differences between the SD + Mel, SD + Abs + Mel, and CON groups. Similar to Mel supplementation, butyrate supplementation also improved SD-induced neuroinflammation and apoptosis. No difference was observed in any of the parameters between the SD + Abs + butyrate and CON groups.

Additional file 1:Supplemental Figure 1. The gut microbiota mediated the neuroprotective effect of melatonin in neuronal apoptosis induced by sleep deprivation. (A-D) Relative protein levels of Bcl-2, Bax and cleaved caspase-3 in the hippocampus (n = 6). CON-FMT: receiving control microbiota FMT mice, SD-FMT: receiving sleep deprivation microbiota FMT mice, SD+Mel-FMT: receiving SD+Mel (20 mg/kg) microbiota FMT mice. The data represent the mean ± SEM, p < 0.05 was set as the threshold for significance by one-way ANOVA followed by post hoc comparisons using Tukey’s test for multiple groups’ comparisons, *p < 0.05, **p < 0.01, ***p < 0.001. Supplemental Figure 2. Composition of the colonic microbiota in FMT-treated mice. (A) OTU number, (B) Simpson index, (C) Shannon index. CON-FMT: receiving control microbiota FMT mice, SD-FMT: receiving sleep deprivation microbiota FMT mice, SD+Mel-FMT: receiving SD+Mel (20 mg/kg) microbiota FMT mice. The data represent the mean ± SEM, p < 0.05 was set as the threshold for significance by one-way ANOVA followed by post hoc comparisons using Tukey’s test for multiple groups’ comparisons, *p < 0.05, **p < 0.01, ***p < 0.001. Supplemental Figure 3. The levels of LPS in the hippocampus (n = 7). CON: control group, SD: sleep deprivation group, SD + Mel: SD + melatonin (20 mg/kg) supplement group. The data represent the mean ± SEM, p < 0.05 was set as the threshold for significance by one-way ANOVA followed by post hoc comparisons using Tukey’s test for multiple groups’ comparisons, *p < 0.05, **p < 0.01, ***p < 0.001.