Akkermansia muciniphila helps prevent memory problems in sleep-deprived mice by changing how brain immune cells remove connections between neurons

The animal experiments were approved by the Institutional Animal Care and Use Committee of Xi’an Jiaotong University (Xi’an, China). Male C57BL/6J mice were purchased from Charles River (Beijing, China) and maintained in a specific pathogen-free environment with a strict 24-h reverse light-dark cycle (lights turned on from 8:00 to 20:00). The mice were fed sterile water and commercial standard feed (GB14924.3, China) ad libitum.

Antibiotic treatment started when mice were five weeks old. An antibiotic cocktail²¹ was administered by oral gavage for 14 consecutive days (50 mg/kg vancomycin, 100 mg/kg neomycin, 100 mg/kg metronidazole, and 1 mg/kg amphotericin B). Ampicillin was provided in drinking water (1 g/L).

The SD mouse model was developed using a propeller-based automatic SD system (XR-XS108, Shanghai Xin Ruan) (Supplementary Figure S1a). The system contained a cage in which mice could feed freely while a clockwise-counterclockwise rotating bar was set rotating at the bottom of the cage, 20:00 until 16:00 the next day at 5 rpm, preventing the mice from sleeping. There are 3 turns clockwise and counterclockwise and 18-s intervals every 6 rotations for mice to achieve food and water when needed. The subjects were sleep-deprived for 20 h every day for seven consecutive days (Supplementary Figure S1b). Control (Con) animals were placed in the same cage without a rotating bar to disturb their sleep.

Before transplantation, the mice were treated with the antibiotic cocktail for 14 consecutive days, as previously described. Thereafter, mice were given 200 µL of the microbiota suspension from either the sleep-deprived or control mice. Three times a week (FMT was administered on days 1^st, 3^rd and 5^th) for 3 weeks, starting 48 h after the last gavage of the antibiotics. The microbiota suspension preparation was based on the method described in previous studies.^22,23 Mice in the empty transplant group received the same antibiotic treatment and were transplanted only with reduced PBS.

Behavioral tests were conducted 7 days after SD or 24 h after the last session of FMT. Performance was tracked and evaluated using a video tracking system (SMART 3.0; Panlab Harvard apparatus). The open-field test (OFT) and elevated plus maze (EPM) tests were used to assess the anxiolytic properties of mice (Supplementary Figure S2a-b), and the novel object recognition (NOR) and Y-maze tests were performed to examine recognition memory and spontaneous rodent behaviors (Supplementary Figure S2c-d). Detailed information is provided in the supplementary material.

Fecal samples were collected at four key times to identify the effectiveness of antibiotic treatment, including the first day without any treatment, 14 days after antibiotic treatment, 7 days after SD, and the day after all behavioral tests were completed. The samples were immediately frozen at −80°C until DNA extraction. After stool samples were weighed, total genomic DNA was extracted from feces using the Stool DNA Kit (Omega, D4015, China) following the manufacturer’s instructions. DNA quantity was determined fluorometrically using a dsDNA Broad Range-Fluorescence Quantification Assay with a T20 Fluorometer (Life Real, China).

Fecal samples for the 16S sequencing were collected from mice, either with normal sleep or 7 days after SD. 16S rDNA gene sequencing was performed on the Illumina sequencing platform in a Biotechnology company (Beijing, China). Briefly, PCR amplification was performed spanning the V3-V4 region of the 16S rDNA using primers 341F/806 R (forward 5′-ACTCCTACGGGAGGCAGCAG-3′ and reverse 5′-GGACTACHVGGGTWTCTAAT-3′). Subsequently, sequencing was performed on a MiSeq platform (Illumina) using a 2 × 300 bp-end protocol.

Cecal contents were collected on the day after all behavioral tests were completed. Total genomic DNA was extracted from cecal contents using the Stool DNA Kit (Omega, D4015, China) as described above. After diluting the DNA templates to 1 ng/µL using RNase-Free Distilled Water, each sample was prepared in triplicate using 2 µL (2 ng of the total DNA template). After diluting the DNA templates to 1 ng/µL with RNase-Free Distilled Water, 2 µL (2 ng of total DNA template) of each sample was loaded in triplicates. The forward primer 5′-CAGCACGTGAAGGTGGGGAC-3′ and reverse primer 5′ -CCTTGCGGTTGGCTTCAGAT-3′²⁴ were premixed with SYBR Green Nucleic Acid Gel Stains (SYBR Green) Premix (AG11701, Accurate Biology, China) according to the manufacturer’s protocol. Real-time PCR (qPCR) was performed using a Bio-Rad CFX96 thermal cycler (Bio-Rad, Hercules, California, USA). Reaction mixtures (20 µL total volume) were held at 95°C for 30 s, followed by 40 cycles of 95°C for 5 s and 60°C for 30 s. Melting curve analysis was conducted to confirm the specificity of amplification. Plasmid DNA (gene sequence shown in the supplementary materials), including the corresponding conservative sequence of A. muciniphila was prepared in a dilution series to create a standard curve. The absolute copy number of the 16S rDNA gene of each sample was then calculated.

A. muciniphila culture and oral supplementation were performed as previously described²² with a few modifications. A. muciniphila (ATCC BAA-835) was grown in fresh pre-reduced Brain Heart Infusion (BHI) broth (Sigma-Aldrich, BD 237,500, USA) under anaerobic conditions (75% N₂, 20% CO₂, and 5% H₂) at 37°C and incubated for 3–4 days. Afterward, cultures were centrifuged and condensed in anaerobic PBS containing 20% (vol/vol) glycerol to a concentration of ~ 1 × 10¹⁰ c.f.u/mL under strictly anaerobic conditions and stored at −80°C until use. The bacteria were incubated in a pre-reduced Tryptic Soy Agar medium (Sigma-Aldrich, BD 236,950, USA) for 4–5 days to determine the A. muciniphila counts (c.f.u./mL). A. muciniphila glycerol stocks were diluted with anaerobic PBS to a final concentration of 2 × 10⁸ viable c.f.u. per 0.2 mL. Mice were treated by oral gavage with 200 μL of either A. muciniphila suspension or anaerobic PBS three days a week (on days 1^st, 3^rd and 5^th) for four weeks, starting 48 h after the last gavage of antibiotics.

SCFA mix (67.5 mM acetate, 40 mM Butyrate) (Sigma-Aldrich, China) was added to drinking water as described previously.²⁵ Mice were treated with drinking water with/without SCFAs for 4 weeks before SD until the behavior tests were completed.

Primary microglial cultures were prepared from the cortical tissue of mouse pups on postnatal day 3 as previously described.²⁶ Microglia were plated at a density of 2 × 10⁴ cells/cm² and maintained in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum, penicillin (100 IU/mL), and streptomycin (100 μg/mL). SCFAs treatment was performed as previously described with a few modifications.²⁷ SCFAs mixtures containing sodium acetate (Sigma-Aldrich, S2889, China) (236 μmol/L) and sodium butyrate (117 μmol/L) (Sigma-Aldrich, 303410, China) or vehicle (V) solution (H₂O) were added to the cultured microglia for 15 min. The cells were then stimulated with LPS (100 ng/mL) for 24 h before all cultures were fixed for immunostaining.

For mouse cortical neuron cultures, cortices from embryonic days (e) 14–17 embryos were dissected in a cold dissection buffer (1×HBSS).²⁸ The meninges, blood vessels and choroid plexus were carefully separated; the tissue was finely minced into approximately 1-mm³ pieces with iris scissors for digestion. The suspension cells were obtained according to the manufacturer’s protocol.²⁸ The cells were plated on poly l-lysine/laminin-coated coverslips at a density of 2.4 × 10⁴ cells/cm² and cultured for 7 days in vitro. At DIV 7, microglia were plated onto primary neurons in a 1:3 microglia-to-neuron ratio. As described above, the SCFA mixture (sodium acetate 236 μmol/L and sodium butyrate 117 μmol/L) was added to the co-culture system or neuron cultures for 15 min, and LPS (100 ng/mL) was added for 24 h before all cultures were fixed using 4% paraformaldehyde (PFA) for immunostaining.

Western blotting was performed as previously described.²⁹ Protein samples were extracted from hippocampal tissue in RIPA buffer supplemented with a protease and phosphatase inhibitor cocktail. After assessing the protein concentration with a BCA Protein Assay kit (Thermo Fisher Scientific, 23227, USA), they were degenerated by heating at 95°C for 10 min with loading buffer before cooling to room temperature and loaded onto 10% SDS-PAGE separation gels. The protein bands were then transferred to polyvinylidene fluoride (PVDF) membranes. Appropriate primary antibodies and HRP-conjugated secondary antibodies were used to detect the proteins of interest (details are shown in Supplementary Table S1). Immunoreactive bands were visualized using a chemiluminescent substrate. β-actin was used as the internal standard. Antibody information can be found in Supplementary Table S1. Immunoblots were quantified using ImageJ analysis software (Version 2.0, NIH, USA).

Mice were deeply anesthetized using isoflurane and transcardially perfused with 20–30 mL of ice-cold normal saline, followed by 15–20 mL of 4% PFA. The brains were removed and post-fixed in 4% PFA at 4°C for 6–8 h. After dehydration in 30% sucrose in 0.1 M Phosphate buffer at 4°C for 3–4 days, 30-µm thick coronal brain sections were obtained using a cryostat (Thermo, CryoStar N×50OPD). Free-floating sections were blocked with 5% donkey serum/0.3% Triton X-100 in PBS for 2 h at room temperature (RT), followed by incubation with primary antibodies (antibody information is shown in Supplementary Table S1) diluted at 4°C overnight. The sections were rinsed three times in PBS the morning before incubation with the corresponding secondary antibodies for 2 h at RT.

Immunolabeled proteins were imaged using a Nikon A1 confocal microscope (Nikon, Japan). The images for dentate gyrus synaptic marker analysis were captured using a 63× oil immersion objective lens and 3× zooms with an interval of 5 μm along the Z-axis. Iba1 and CD68 were captured using a 40× objective lens. Iba1 and synaptophysin (SYP) were captured using a 63× objective. VGLUT1 and PSD95 of primary neurons were captured using 60× objectives and 4× zooms using a Leica TCS SP8 STED 3X confocal microscope (Leica, Germany). Images were acquired from three sections representing the rostral, middle, and caudal regions of the hippocampal dentate gyrus. For imaging, we used three brain slices per mouse from three mice per experimental group, and three images were taken for each slice at random locations in the dentate gyrus, as defined by DAPI staining. For in vitro experiment, images from five fields per coverslip distributed in each quadrant or center with three replicates were taken for analysis. The acquired Z-stacks were then individually projected onto a two- dimensional plane. For the synaptic marker count per unit, the area was analyzed using the ImageJ program with a threshold of normalized fluorescence intensity. For synapses inside microglia, synaptic puncta labeled with SYP and Iba1 were imaged, as described above. Subtracted background and smooth images were obtained using ImageJ software. Microglia engulfment was quantitatively measured using the Imaris software 9.0.1. All experimental analyses were performed by a researcher who was blinded to the treatment groups.

The microbiomes were analyzed at the Illumina sequencing platform by the Beijing Genomics Institute (Beijing, China). High-quality reads for bioinformatics analysis were selected, and all of the valid reads from all samples were clustered into operational taxonomic units (OTUs) based on 97% sequence similarity using the Usearch (v7.0.1090_i86linux32). Then, OTUs were annotated by comparing representative sequences to the Greengene (V201305) and Ribosomal Database Project (RDP) (Release16, 20160930) using the RDP classifier (1.9.1). The difference comparisons among different groups were displayed by the principal coordinate analysis (PCoA) based on the UniFrac distance matrix. Alpha-diversity (Shannon index and Simpson index) indices were calculated by Wilcoxon Rank-Sum Test. Beta-diversity analysis was performed using weighted UniFrac distances. LEfSe (Linear discriminant analysis effect size) analysis was performed under the following conditions: the P-value for the factorial Wilcoxon tests among classes was 0.05 and the threshold on the logarithmic LDA score for discriminative features was 2.0 (the code from https://github.com/SegataLab/lefse↗). The analysis was conducted using R (V3.5.1).

Results are presented as the mean ± SEM, and differences were considered significant when P < .05. Statistical analyses were performed using Prism 7.0 (GraphPad, USA). Student’s t-test was used to compare two groups with a normal distribution, while One-way ANOVA with Tukey’s multiple comparisons was used for comparisons of three or more groups. Correlations were analyzed using linear regression.

Furthermore, to better understand the connection between SCFAs levels and cognitive function, a correlation analysis was performed between the levels of several SCFAs and percentage of exploration time. The results demonstrated that the serum levels of acetate and butanoic acid were positively correlated with the percentage of exploration time (Figure 7j,k). Additionally, the results of SCFA-targeted metabolomics showed that the serum concentrations of acetate and butanoic acid were decreased in SD mice compared with those in control mice, which were significantly restored by oral administration of AKK (Figure 7l,m). We then tested the effects of SCFAs (acetate and butanoic acid) pre-supplements in SD mice and conducted behavioral analysis (Supplementary Figure S9a). The NOR test showed that the reduction in exploration time for the novel object in SD mice was prevented by SCFAs pretreatment (Supplementary Figure S9b, SD/SCFAs 69.16 ± 3.658%, vs. SD/V 55.81 ± 3.70%, P = .0178). The Y-maze results indicated that the spontaneous alternation index drop observed in SD mice was also blocked by SCFAs pretreatment. (Supplementary Figure S9d, SD/SCFAs 56.87 ± 2.97%, vs. SD/V 46.19 ± 2.03%, P = .0165), while the total distance traveled in the NOR and Y-maze tests was similar among groups (Supplementary Figure S9c and e). The above results indicate that SCFAs pretreatment alleviated SD-induced cognitive impairment. These results suggest that AKK treatment restored acetate and butanoic acid levels, which may be the basis for its protective effect against SD-induced cognitive dysfunction.

VGLUT1 and PSD-95 protein levels were significantly increased in SD mice pretreated with SCFAs compared with the SD/V group (Supplementary Figure S10g-i). Therefore, to investigate whether SCFAs prevent synapse loss via reducing synapse engulfment of microglia. We used a co-culture system in which microglia were added to primary neurons cultured for 7 days in vitro. SCFAs were pre-administered to the co-culture system 15 min before LPS treatment. After 24 h, cells were fixed for immunostaining (Figure 8b). Quantitative analyses of neuron-microglia co-cultures revealed decreased synaptophysin density in neuron/microglia treated with LPS compared to that in control neurons (Figure 8f,g). Moreover, a significant increase in synaptophysin-positive puncta inside microglia was found in co-cultures treated with LPS compared to the control cells. However, pretreatment with SCFAs blocked this change in co-culture (Figures 8f–h). In primary cultured neurons, LPS treatment did not alter VGLUT1 or PSD95 expression (Supplementary Figure S11), indicating that other necessary factors outside the neurons are responsible for synaptic loss. In contrast, VGLUT1 and PSD95 were significantly reduced in neuron/microglia co-cultured with LPS compared to control neurons. SCFAs administration significantly rescued VGLUT1 and PSD95 expression in the neuron/microglia co-culture after LPS stimulation (Figures 8i and k). Together, these results suggest that SCFAs protect against synapse loss by inhibiting synaptic engulfment of microglia.