Intestinal interoceptive dysfunction drives age-associated cognitive decline

Mice

Unless noted otherwise, young female (8 weeks old) and male (4 weeks old) C57BL/6 mice were obtained from The Jackson Laboratory and old mice (18 months old) were acquired from the National Institute on Aging. For co-housing, mice were housed either two young + three old, or three young + two old per cage. Males were co-housed before puberty to avoid fighting. Non-co-housed control young and old mice had their cages mixed so that two mice were swapped between each cage to control for the social effect of cage mixing. Where possible, mice were randomized to experimental groups, and experimenters were blinded to the experimental condition. Animals were housed in facilities at the University of Pennsylvania, the Arc Institute or Stanford University in 7 am to 7 pm light–dark cycles, 20–25 °C and 30–70% humidity. All experimental procedures were performed according to approval by the IACUC committees at the University of Pennsylvania, Stanford University and the Arc Institute.

Germ-free C57BL/6 mice were maintained in sterile isolators at the University of Pennsylvania Gnotobiotic Animal Facility. For FMT, one stool pellet per recipient mouse was homogenized in 1.5 ml of sterile phosphate-buffered saline (PBS) in an anaerobic hood and filtered through 70 μm filters. This homogenate (200 μl) was orally gavaged into germ-free mice housed in positive pressure isocages; in addition to oral gavage, 50 ml of used bedding from donor cages was added to the isocages.

The following mouse strains were used in manuscript and were purchased from The Jackson Laboratory: C57BL/6J (000664), CD45.1 (002014), DBA/2J (000671), Trpv1^Cre (017769), Snap25-GCaMP6s (025111), DTA (009669), Phox2b^Cre (016223), Phox2b^FlpO (022407), Il1r^fl/fl (028398), Ccr2^−/− (004999), Nlrp3^−/− (021302), Tnfr^−/− (003243), Cre-dependent hM4Di (026219), Cre/FlpO-dependent hM4Di (029040) and Ai65 (Cre/FlpO-dependent tdTomato) (021875). Gpr84^−/− mice were a kind gift from I. Kimura⁶⁰.

Antibiotic treatment, bacterial colonization and stool collection

Mice were given a combination of neomycin (1 g l^–1; Research Products International); ampicillin (1 g l^–1; Research Products International); vancomycin (0.5 g l^–1; Mylan); metronidazole (0.5 g l^–1; Research Products International); imipenem or cilastatin (0.5 g l^–1; Fresenius Kabi); and ciprofloxacin (0.2 g l^–1; Sigma-Aldrich) in their drinking water for two weeks. Parabacteroides goldsteinii (ATCC) and Alistipes shahii (ATCC) were grown under anaerobic conditions at 37 °C in Tryptic Soy Broth (BD Biosciences) supplemented with 5% defibrinated sheep’s blood (Hardy Diagnostics). Lachnospiraceae bacterium (DSMZ) was grown under anaerobic conditions at 37 °C in fastidious anaerobe broth (FAB) (Neogen). Lactobacillus acidophilus (ATCC) and Escherichia coli K12 (ATCC) were grown under aerobic conditions according to supplier instructions. Bacteria were gavaged into mice after two weeks of antibiotics treatment 24 h after the cessation of treatment. Mice were gavaged every other day the first week, and then once a week after. Control mice received PBS gavage. Stool samples for all experiments were collected fresh in 1.7 ml Eppendorf tubes and immediately snap frozen on dry ice before storage at −80 °C until DNA extraction.

Microbial metabolite isolation

P. goldsteinii and A. shahii were grown for two days in 50 ml tubes before centrifugation for 20 min at 3,220 × g at 4 °C. The supernatant was sterilized through 0.45 μm filters before 30 min of centrifugation through a 3 kDa size filter (Thermo Fisher Scientific) at 3,220 × g at 4 °C. The <3 kDa fraction was then treated with 25 μg ml^–1 proteinase K (Thermo Fisher Scientific) for 1 h at 37 °C and then boiled for 10 min at 99 °C. This solution (300 μl) was gavaged into mice for five days before testing 1 h after the final gavage.

Bromodeoxyuridine labelling

Mice received 50 mg kg^–1 bromodeoxyuridine (BrdU) (MedChem Express) dissolved in sterile PBS via intraperitoneal injection for five consecutive days, brains were collected four weeks after final injection.

Capsaicin, cholecystokinin, GLP1 agonist and CCKAR antagonist treatment

Capsaicin (Sigma-Aldrich) was dissolved at 25 mg ml^–1 in 10% Tween-80, 10% ethanol and 80% PBS. Mice were intraperitoneally injected with 5 μg kg^–1 capsaicin for five days before testing. Cholecystokinin octapeptide (2 μg kg^–1 CCK; Bachem), GLP1 (amino acid positions 7–36) (4 mg kg^–1, Cayman Chemical, 15069) and liraglutide (1 mg kg^–1; Toronto Research Chemicals) were dissolved in PBS and injected 1 h before testing. The CCKAR receptor antagonists devazepide (MedChem Express) and A-65186 (Biorbyt) were dissolved in 10%:10% DMSO:Tween-80 in PBS and injected intraperitoneally for five days at 1 mg kg^–1 before testing.

MCFA, embelin and PBI-4050 treatment

3-HOA (100 mg kg^–1; 1PlusChem), embelin (100 mg kg^–1; Ambeed), decanoic acid (20 mg kg^–1; Sigma-Aldrich) and PBI-4050 (20 mg kg^–1; TargetMol) were dissolved in 5% DMSO, 5% Tween-80 and 90% PBS. An oral gavage was conducted for five days, with 1 h after the final treatment. For water supplementation, decanoic and dodecanoic acid (Sigma-Aldrich) were dissolved at 100 mg ml^–1 in Tween-80 and then added to drinking water at 0.2 mg ml^–1 for ten days; daily water consumption was assumed to be 3 ml per mouse.

In vivo antibody treatment

All antibodies were obtained from BioXCell. For cytokine depletion, mice received intraperitoneal injections of anti-TNF (150 μg, BP0058) or anti-IL-1β (50 μg, BE0246) on days 0, 4 and 7 before testing on day 8. For T cell depletion, mice received 200 μg intraperitoneal injections of anti-CD4 (BE0003) on days 0 and 4 and 200 μg anti-CD8α intraperitoneally (BE0061) on day 0 before testing on day 7. For natural killer cell depletion, mice received 200 μg intraperitoneal injections of anti-NK1.1 (BE0036) on days 0 and 4 before testing on day 7. For neutrophil depletion, mice received daily intraperitoneal injections of anti-Ly6G (25 μg, BP0075-1) and anti-rat immunoglobulin G (50 μg, BE0122) every other day for one week before testing⁶¹.

PLX-3397 treatment

PLX-3397 (Selleck Chemicals, S7818) was formulated in AIN-76A (Research Diets) at 290 mg kg^–1 and fed to mice for ten days before testing.

Clodronate liposome treatment

Control and clodronate liposomes (Liposoma) were injected at 10 mg kg^–1 equivalent of clodronate dose on days 0 and 3. Treatment with 3-HOA started on day 1; mice were tested on day 5 and tissues were collected on day 6.

Recombinant TNF and IL-1β treatment

TNF or IL-1β (1 μg; PeproTech) dissolved in PBS was injected intraperitoneally once daily for five days, with testing performed 1 h after the final injection.

In vivo 2-photon nodose ganglia imaging

In vivo imaging

In vivo imaging was conducted using a two-photon microscope (Bruker) equipped with a galvanometer for image acquisition and a piezo objective combined with a galvo/resonant scanner, allowing image capture at 29 frames per second. The set-up included a Somnosuite anaesthesia system (Kent Scientific) for isoflurane delivery, a homoeothermic control warming pad to maintain body temperature and a programmable syringe pump (Harvard Apparatus PHD 2000) for nutrient infusions into the intestine. Imaging was performed using a 20× water-immersion upright objective.

Surgical procedure

The surgical procedure followed a previously described protocol²⁹. Snap25-GCaMP6s mice were fasted for 1 h before the onset of the dark cycle. Mice were placed under continuous anaesthesia (isoflurane/oxygen) and maintained on a heating pad throughout the procedure. At least one hour after the onset of the dark cycle, an approximately 2-cm incision was made above the sternum and below the jaw to expose the carotid artery and vagus nerve by separating the salivary glands. Retractors were used to pull the sternomastoid, omohyoid and posterior belly of the digastric muscles aside, allowing visualization of the nodose ganglion.

The vagus nerve was transected above the nodose ganglion, which was carefully separated from the hypoglossal nerve and small adjacent branches. The vagus nerve was then dissected away from the carotid artery and surrounding tissues. The right nodose ganglion was gently positioned onto a stable imaging platform consisting of a 5 mm diameter coverslip attached to a metal arm with a magnetic base. Surgical silicone adhesive (Kwik-Sil, WPI) was applied to secure the vagus nerve in place, and the ganglion—immersed in DMEM (Corning)—was covered with a second coverslip before imaging.

To deliver nutrients into the intestine, a small abdominal incision was made in the anaesthetized mouse to expose the stomach. Silicone tubing was inserted through a small incision in the stomach wall and advanced into the proximal duodenal lumen. Super glue was applied to secure the tubing to the stomach wall and prevent it from sliding out. Infusions were delivered using a precision pump connected to syringes filled with Ensure and attached to the silicone tubing. Baseline neuronal activity was recorded for 30 s, followed by a 30 s of continuous infusion of 200 µl of Ensure, and an additional 90 s of post-infusion recording.

In certain experiments, the surgery followed treatment with either vehicle (Tween-80) or decanoic acid 0.2 mg ml^–1 in drinking water for one week before recording. In experiments using PLX diet, decanoic acid and PLX were administered concurrently for one week.

Fluorescence analysis

GCaMP6s fluorescence changes were analysed by outlining regions of interest (ROIs) corresponding to individual neurons throughout the imaging session. Pixel intensities within ROIs (averaged across pixels) were measured frame-by-frame using ImageJ, and the data were exported to Excel for further analysis. The z-score for each neuron was calculated as follows: the mean baseline fluorescence (over a 30-s period) was subtracted from the fluorescence intensity at each time point, and the result was divided by the standard deviation of the baseline fluorescence. This normalization quantified fluorescence changes as the number of standard deviations above baseline.

Neurons were classified as responsive if they met the following criteria: (1) peak GCaMP6s fluorescence reached a z-score of ≥2.5, and (2) mean GCaMP6s fluorescence was more than 2.5 above the baseline mean for at least five consecutive seconds during or after infusion. Nodose ganglion neurons that displayed no baseline activity were excluded from the analysis.

CCK–SAP nodose ganglia injection

Mice were anaesthetized with isoflurane (2–5%). After hair removal and disinfection, an incision was made in the neck area and the sternohyoid and sternomastoid muscle connective tissue blunt dissected and moved aside to expose the carotid artery. The vagus nerve was gently exposed and followed to reach the nodose ganglion as it enters the foramen. The glass micropipette was advanced into the ganglion and CCK–SAP (Advanced Targeting Systems) or blank-SAP (Advanced Targeting Systems) was injected (125 ng in 0.5 μl per ganglion). The procedure was repeated on the other side, and the incision was closed by using a nylon suture. Mice were monitored until fully awake and then placed back in the home cage and tested two weeks after surgery.

Coeliac–superior mesenteric ganglionectomy

To determine splanchnic contributions to gut–brain sensing, splanchnic afferents underwent a gangliectomy at the prevertebral coeliac and superior mesenteric ganglia. Mice were anaesthetized and treated with analgesia and an abdominal midline incision was made through skin and muscle. The coeliac and superior mesenteric arteries were exposed by targeting the aorta at the arterial branching point medial to the left kidney. At the intersection of the three blood vessels, coeliac and superior mesenteric ganglia and the surrounding tissue were gently teased away until the area between was cleared of nerve and lymphatic tissue. In sham animals, the area was exposed, and a small tear was made in the lymphatic tissue, but the ganglia were left intact. Mice were given at least one week to recover and regain presurgical weight.

Dorsal root ganglia ablation and hot plate testing

Resiniferatoxin (RTX; AdipoGen) was dissolved in PBS with 0.25% DMSO/0.02% Tween-80/0.05% ascorbic acid and 50 ng RTX in 10 μl was injected intrathecally at L5–L6 into four-week-old male wild-type mice under isoflurane anaesthesia⁶². Mice were tested four weeks after injection, and ablation was confirmed by hot-plate testing, where mice were placed on a 56 °C plate and latency to paw licking was timed with a maximum of 30 s.

Bone marrow chimera

Sex-matched CD45.1 mice received two 550 cGy doses of irradiation spread 4 h apart and received one million live bone marrow cells 24 h after the first irradiation. For mixed chimeras, mice received a 50:50 mix of either wild-type or Gpr84^−/− bone marrow combined with Ccr2^−/− bone marrow. Mice were tested six weeks after transplantation.

Intestinal barrier permeability testing

Fluorescein isothiocyanate-dextran (3 kDa; Sigma-Aldrich) was dissolved in PBS to a concentration of 80 mg l^–1. Mice were fasted for 4 h before being subjected to gavage with 150 μl, and blood was centrifuged at 10,000 × g for 5 min at 4 °C. Serum was collected and fluorescence was quantified at an excitation wavelength of 485 nm and emission wavelength of 535 nm.

Whole mouse imaging

A four-week-old Trpv1^Cre Phox2b^FlpO tdTomato mouse was fasted for 24 h to reduce luminal gut content, euthanized and then perfused with 25 ml cold PBS followed by 25 ml cold 4% paraformaldehyde (PFA); the mouse was then post-fixed in 4% PFA at 4 °C for 24 h. The skin was then removed and the mice were further processed by LifeCanvas Technologies. Briefly, the body was decalcified in 10% EDTA at 4 °C for ten days, refreshing the solution every two days before SHIELD fixation (six days SHIELD OFF at 4 °C followed by 24 h SHIELD ON at 37 °C). Delipidation buffer was then applied at 45 °C for ten days before organic solvent delipidation with a THF/PBS-Quadrol (25% Quadrol in PBS) dehydration gradient. Dehydration was performed at 4 °C with 50%, 70%, 80%, 90% and 95% THF solutions, for 24 h per step.

The sample was then moved to dichloromethane for five days, refreshing daily. After that the sample was rehydrated following the reverse gradient and washed in PBS for three days. The sample was then index matched in 50% EasyIndex for three days, before being moved to 100% EasyIndex and index matched for a further three days, mounted and then imaged after removing the paws. Imaging was performed on a SmartSPIM microscope with a 1.625× objective using a 4 μm z-step.

CCK and GLP1 ELISAs

CCK (Abbexa) and GLP1 (EMD Millipore) ELISA kits were used according to manufacturer’s instructions. Mice were fasted for 24 h, and tissue and serum were collected 45 min after refeeding. Tissue samples were homogenized in RIPA buffer (Cell Signalling Technology) with protease inhibitors (Sigma-Aldrich).

Isolation of monocytes

Monocytes were extracted from whole blood using a mouse monocyte isolation kit according to manufacturer’s instructions (Biolegend).

Isolation of intestinal leukocytes

The intestinal lumen was flushed with cold PBS, opened longitudinally and cut into 1 cm pieces. Epithelial cells were removed by incubating in dissociation buffer (3% foetal bovine serum (FBS) (Corning) and 10 mM EDTA in Hank’s balanced salt solution) at 4 °C with shaking for 30 min. After removing epithelial cells by vortexing and straining through a 70-μm filter, the remaining intestinal pieces were washed thoroughly in cold PBS and then digested in Hank’s balanced salt solution containing 0.1 mg ml^–1 Liberase (Sigma-Aldrich) and 1 mg ml^–1 DNase (ROC) for 20 min at 37 °C, with shaking at a rate of 180 r.p.m. The digested cells were washed with PBS and passed through a 70 μm cell strainer.

Flow cytometry

Intestinal (as described above) or blood leukocytes were collected; blood leukocytes were collected via cardiac puncture and red blood cells lysed with ACK lysing buffer (Quality Biological). Cell viability staining was performed using the LIVE/DEAD Fixable Aqua Cell Stain Kit (Thermo Fisher Scientific; 1:500); blocking with TruStain FcX PLUS (BioLegend; 1:50) for 30 min at 4 °C. Staining was performed with BioLegend antibodies at 1:100 dilution against CD45, CD45.1, CD45.2, TCRβ, CD4, CD8a and NK1.1 for 15 min at 4 °C. All samples were acquired by flow cytometry (LSR II, BD Biosciences) and analysed using the FlowJo software (Tree Star).

DREADD adenovirus and clozapine-N-oxide injection

Viral genomes (1 × 10¹¹) of either AAV9-PHP.S.hSyn.DIO.hM3D(Gq)-mCherry.WPRE.hGH (Addgene) or AAV9-PHP.S.hSyn.DIO.hM4D(Gi)-mCherry.WPRE.hGH (Addgene) were diluted in PBS and injected retro-orbitally into heterozygous Trpv1^Cre mice^63,64. Testing was performed 3–4 weeks after injection. Baseline performance was established 1 h after intraperitoneal injection of vehicle (0.1% DMSO in sterile PBS). Mice were then tested 1 h after injection of CNO (Cayman Chemical Company) at 1 mg kg^–1. Phox2b^hM4Di mice were injected once daily for five days with 0.5 mg kg^–1 CNO and tested 3 h after injection.

Frailty index

Frailty in young and old mice at baseline and after co-housing was assessed using an established 25-item frailty index⁶⁵.

Novel object recognition

On the day of testing, mice were placed into a clean rat cage with bedding; after 1 h of acclimatization, two identical objects were placed in the cage for 10 min. One hour after removing the familiar objects, the familiar object was reintroduced with a novel object. Interaction, defined as sniffing or direct nose contact, was timed until 30 s of total interaction time was reached. If a mouse did not reach 30 s of total exploration within 10 min, it was excluded. For germ-free mice, the experiment was performed within their sterile home isolator. Mice were confirmed to have no innate preference within the object pairings⁶⁶, which are:

1. 1.

   Elmer’s 22 g glue sticks and 5 cm binder clips (Amazon ASIN).

2. 2.

   100 ml glass bottle (Sari-Glas) filled with 100 ml water and plastic funnel (Thermo Fisher Scientific).

3. 3.

   A custom-made 2.5 cm square aluminium block, 15 cm tall on 5 cm × 5 cm clear plastic base, and 2.5 cm diameter white PVC cylinder, 15 cm tall on 5 cm × 5 cm clear plastic base.

The recognition index was calculated as follows: (novel object time (s) – 15 s)/30 s.

Barnes maze

Mice were placed in the maze (San Diego Instruments) opposite the escape hole and allowed to explore for up to 2.5 min. After each trial, if the mouse did not find the escape hole, it was gently guided to it and allowed to stay there for 15 s. Mice were tested twice a day, 1 h apart, for four consecutive days. Tracking and analysis were conducted using the Ethovision software v.15 (Noldus).

Open field test

Mice were handled one day before testing and acclimatized to the room for 30 min before testing from 8 pm to 10 pm. Mice were placed into a 40 × 40 cm box and allowed to explore for 15 min. Tracking and analysis were conducted using the EthoVision software v.15 (Noldus).

Stool DNA extraction

Stool DNA was purified using DNeasy PowerSoil Pro Kit (Qiagen) according to manufacturer’s instructions. For the longitudinal C57BL/6J cohort, DNeasy PowerSoil Kit (Qiagen) was used.

Metagenomic library preparation and analysis

Libraries from longitudinal samples over the lifespan were prepared with DNA extract from stool using the Illumina Nextera DNA Flex Library Prep kit (Illumina) with TruSeq unique dual indices (Illumina) according to manufacturer’s instructions. Quality and quantity control of DNA and libraries were performed using an Agilent 4200 TapeStation and Qubit 4, respectively. Sequencing was performed on a NextSeq 550 instrument using a 75-cycle kit (Illumina). Demultiplexing was performed using the illumina Bcl2fastq software (v.2.20.0.422). Reads processing was performed using the snakemake Sunbeam pipeline (v.2.1.1 + dev81.gd0e29cd)⁶⁷. Briefly, reads were trimmed for minimal length of 36-base pairs using Trimmomatic (v.0.39). Adapter sequences were removed. Reads with a sequence complexity score lower than 0.55 were removed using Komplexity (v.0.3.6). Reads mapped against mouse host genome (mm10) were filtered out. Taxonomic assignment of the decontaminated and quality-controlled reads was performed using Kraken 2 (ref. ¹⁴) to the Kraken 2 database PlusPFP. Subsequent analysis was performed using the statistical computing environment R v.4.1.2 (ref. ⁶⁸) in RStudio (v.4.2) using the following packages: ape (v.5.5), vegan (v.2.6.4), DESeq2 (v.1.32.0), matrixStats (v.0.61.0), cowplot (v.1.1.1), broom (v.0.7.8), dplyr (v.1.0.7), tidyr (v.1.1.3) and tidyverse (v.2.0.0).

16S rDNA library preparation and microbiome analysis

DNA was amplified using the KAPA Hifi HotStart ReadyMix (Roche) and the F27/R338 primer-pair targeting the V1V2 region of the 16S gene (F27, 5′-AGAGTTTGATCCTGGCTCAG-3′; R338, 5′-TGCTGCCTCCCGTAGGAGT-3′).

After PCR amplification, the pooled library underwent bead-based size selection using AMPure XP beads (Beckman Coulter) and was sequenced using 500 bp paired-end sequencing (Illumina MiSeq). QIIME 2 (v.2021.2.0) was used to process 16S sequencing reads⁶⁹. Reads were demultiplexed⁷⁰, quality-filtered⁷¹ and denoised using deblur with a trim-length of 200′ (ref. ⁷²). For taxonomic classification, reads were extracted from the Silva 138 database^73,74 using the following parameters: ‘–p-f-primer AGAGTTTGATCCTGGCTCAG –p-r-primer TGCTGCCTCCCGTAGGAGT –p-trunc-len 200’. These reads were used to train a naive Bayes classifier that was then applied to our dataset. Downstream analysis was performed in R v.4.1.0 using tidyverse, phyloseq and vegan packages⁷⁵.

To nominate candidate members of the microbiome associated with cognitive decline, we calculated a combined score by multiplying the P values of each taxonomic element from co-housing experiments and faecal microbiota transfer experiments, as well as its longitudinal association with age.

In vitro myeloid differentiation and treatment with GPR84 agonist

Haematopoietic progenitors (LSKs) were isolated from bone marrow and expanded for ten days as previously described⁷⁶. Progenitors were then differentiated into myeloid cells for one week in IMDM (Thermo Fisher Scientific), 20% FCS (Thermo Fisher Scientific), 1% penicillin–streptomycin–glutamine (Gibco), mouse GM-CSF (10 ng ml^–1; PeproTech), mouse SCF (10 ng ml^–1; PeproTech), mouse IL-3 (5 ng ml^–1; PeproTech), mouse IL-6 (5 ng ml^–1; PeproTech), mouse FLT3L (5 ng ml^–1; PeproTech) and mouse TPO (2 ng ml^–1; PeproTech). Following differentiation, myeloid cells were replated into 24-well plates at 5 × 10⁵ cells per millilitre and treated with LPS (10 mg ml^–1; Sigma) or vehicle for 4 h. After LPS treatment, embelin (100 mM; AmBeed) was added and supernatants were collected after 30 min. IL-1β was measured from the supernatant using ELISA (Invitrogen).

Proteomics sample preparation

Samples were prepared mostly as previously described^77,78. Briefly, longitudinal faecal samples of mice at 1, 4, 7, 9, 11, 15, 18, 22 and 25 months old were suspended in lysis buffer (3% SDS, 50 mM (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid) (HEPES) pH 8.5, 75 mM NaCl + protease inhibitors). The samples were put on ice and lysed using Cole-Parmer Tissue Homogenizer for five cycles (60 s on and 60 s off cycle), followed by centrifugation at 20,000 × g for 15 min at 4 °C. The supernatant was taken and reduced with 5 mM dithiothreitol, alkylated with 15 mM iodoacetamide for 30 min at 25 °C and quenched with 5 mM dithiothreitol. Proteins were purified by methanol–chloroform precipitation. The dried protein pellet (50 μg each) was resuspended in 50 mM HEPES at pH 8.5 with 8 M urea for denaturation. The protein mixture was diluted with 50 mM HEPES pH 8.5 to 1 M urea and digested overnight with 33.3 ng μl^–1 Trypsin/LysC mix (Promega) at 37 °C, 1000 r.p.m. Samples were desalted using SepPak cartridges (Waters) and then vacuum-dried. All samples were then labelled with tandem mass tag (TMT) pro 18plex reagent set (Thermo Fisher Scientific) at a TMT:peptide ratio of 5:1. Samples were mixed and incubated for 1 h at 25 °C; they were then quenched at 0.5% hydroxylamine, mixed and incubated at room temperature for 15 min. A small amount of the samples was combined and analysed via LCMS to confirm the labelling efficiency and ratio balancing across the channels before recombining into one set. The combined samples were then partially dried down to remove acetonitrile, acidified with trifluoroacetic acid to a final concentration of 1%, and then desalted using C18 StageTips, with peptides eluted using 40% acetonitrile with 5% formic acid, and then 80% acetonitrile with 5% formic acid, and dried down under a vacuum at room temperature. The dried pellet was resuspended in 5% acetonitrile, 10 mM ammonium bicarbonate at pH 8.0, sonicated for 10 min and then fractionated by medium pH reverse phase high-performance liquid chromatography (HPLC; Zorbax 300Extend C18, 4.6 × 250 mm column, 5 µm particle size, Agilent) at 25 °C. Peptides were eluted with a gradient with initial starting condition of 100% buffer A (5% ACN, 10 mM ammonium bicarbonate) and 0% buffer B (95% ACN, 10 mM ammonium bicarbonate). Buffer B was increased to 35% over 60 min and then ramped up to 100% B in 6 s, where it held for 5 min. Buffer B was then decreased to 0% over 6 s and held for 10 min to re-equilibrate the column to original conditions. The samples were fractionated into 96 fractions, and then pooled into 24 fractions as previously described. The fractions were dried under vacuum and resuspended in 5% ACN with 5% formic acid, and approximately 1 μg per sample was injected for analysis on an Orbitrap Eclipse as described below.

Proteomics peptide measurement

Samples were analysed on an EASY-nLC 1200 (Thermo Fisher Scientific) HPLC coupled with an Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scientific) with Tune version 3.4. Peptides were separated on an Aurora Series emitter column (25 cm × 75 μm inner diameter, 1.6 μm C18; Ionopticks) and held at 60 °C during separation using an in-house-built column oven over 180 min. Mass spectrometry data were collected in data-dependent acquisition mode. A high-resolution MS¹ scan (500–1,200 m/z range, 60,000 resolution, automatic gain control (AGC) 5 × 10⁵, 100 ms maximum injection time, RF for S-lens 30) was collected in the Orbitrap, and the top ten precursors were selected for MS² and MS³ analysis. Ions were isolated using a 0.5 m/z window for MS² spectra. The MS² scan was performed in the quadrupole ion trap (CID, AGC 1 × 10⁴, 30% normalized collision energy, 35 ms maximum injection time) and the MS³ scan was analysed in the Orbitrap (HCD, 60,000 resolution, maximum AGC 5 × 10⁴, 250 ms maximum injection time, 50% normalized collision energy). The maximum cycle time was set to 5 s. For TMT reporter ion quantification, up to six fragment ions from each MS² spectra were selected for MS³ analysis using synchronous precursor selection.

Proteomics data analysis

The data was analysed using Masspike software licensed from Harvard (GFY core v.3.8) with a built-in statistical package⁷⁹. Thermo Fisher Scientific raw files were converted to mzXML using ReAdW.exe. The assignment of MS² spectra was performed using the SEQUEST algorithm v.28 by searching the data against the combined reference proteomes for Mus musculus acquired from SwissProt and microbiome sequences from Multiple Bacteria Genome Compressor database⁸⁰ along with common contaminants such as human keratins and trypsin. The MS/MS spectra were matched with fully tryptic peptides from this composite dataset using a precursor ion tolerance of 20 ppm and a product ion tolerance of 1 Da. TMT modification of peptide N-termini and lysine residues (+304.2071 Da) and carbamidomethylation of cysteine residues (+57.02146 Da) were set as static modifications. Oxidation of methionine residues (+15.99492 Da) was set as a variable modification. Peptide spectral matches were filtered to a 1% false discovery rate using linear discriminant analysis as previously described⁸¹. Non-unique peptides that matched to multiple proteins were assigned to proteins that contained the largest number of matched redundant peptides sequences using the principle of Occam’s razor⁸¹. Quantification of TMT reporter ion intensities was performed by extracting the most intense ion within a 0.003 m/z window at the predicted m/z value for each reporter ion. Peptide level intensities and the corresponding reporter ion scans were analysed using the previously described framework for multiplexed mass spectrometry proteomics analysis⁷⁹.

Hippocampus RNA extraction and sequencing

Mice were anaesthetized with isoflurane 1 h after exposure to a novel object and the left hippocampus was rapidly dissected and flash frozen. Total RNA was extracted using the trizol–chloroform–isopropanol method. Libraries were prepared using the Illumina Stranded mRNA Prep kit, according to manufacturer’s instructions, and sequenced on a NextSeq 550 instrument using a 75-cycle kit (Illumina). Quality and quantity control of RNA and libraries were performed using an Agilent 4200 TapeStation and Qubit 4, respectively. Libraries were sequenced on an Illumina NextSeq instrument to produce 74-base-pair single end reads with an average sequencing depth of ten million reads per sample. Raw reads were mapped to the mouse reference transcriptome (GRCm38) using Kallisto v.0.46.0 (ref. ⁸²). Subsequent analysis was performed using the statistical computing environment R v.4.1.2 in RStudio v.1.4.17 and Bioconductor v.3.13 (ref. ⁸³) in a similar approach as recently published⁸⁴. Briefly, transcript quantification data were summarized to genes using the tximport package⁸⁵ and normalized using the trimmed mean of M values method in edgeR⁸⁶. Genes with a trimmed mean of M values of less than 1 in n + 1 of the samples (where n is the size of the smallest group of replicates) were filtered.

Quantitative real-time PCR

RNA was isolated with the trizol–chloroform–isopropanol method and complementary DNA was made a high-capacity reverse transcription kit (Applied Biosystems). Quantitative real-time PCR was performed by using Luna Universal qPCR Master Mix (New England Biolabs) with a 10 μl reaction volume according to manufacturer’s instructions on an Applied Biosystems (Quantstudio v.6) thermal cycler. The following primer sequences were used: Gpr84 Fwd (5′-AGGTGACCCGTATGTGCTTC-3′) and Rev (5′-GTTCATGGCTGCATAGAGCA-3′), Tnfa Fwd (5′-GGTGCCTATGTCTCAGCCTCTT-3′) and Rev (5′-GCCATAGAACTGATGAGAGGGAG-3′), Il1b Fwd (5′-TGGACCTTCCAGGATGAGGACA-3′) and Rev (5′-GTTCATCTCGGAGCCTGTAGTG-3′), Il6 Fwd (5′-TACCACTTCACAAGTCGGAGGC-3′) and Rev (5′-CTGCAAGTGCATCATCGTTGTTC-3′). Gapdh or 18S were used as housekeeping genes: Gapdh Fwd (5′-GGGTGTGAACCACGAGAAATATG-3′) and Rev (5′-TGTGAGGGAGATGCTCAGTGTTG-3′), 18S Fwd (5′-CTTAGAGGGACAAGTGGCG-3′) and Rev (5′-ACGCTGAGCCAGTCAGTGTA -3′). Relative gene expression was calculated by 2^{(Ct housekeeping – Ct gene)}. P. goldsteinii relative abundance was quantified by extracting DNA from stool pellets as described above and the following primers⁸⁷: P. goldsteinii Fwd (5′-GCAGCACGATGTAGCAATACA-3′) and Rev (5′-TTAACAAATATTTCCATGTGGAAC-3′) and 16S Fwd (5′-GTGSTGCAYGGYTGTCGTCA-3′) and Rev (5′-ACGTCRTCCMCACCTTCCTC-3′) and calculated by 2^{(Ct 16S – Ct P. goldsteinii)}.

Parabacteroides goldsteinii RNA sequencing

P. goldsteinii was grown in FAB (Neogen) at 37 °C to OD₆₀₀ = 0.2 before being infected with multiplicity of infection (MOI) = 1 PDS1 phage and grown for 24 h. Bacteria were pelleted at 10,000 × g for 5 min and RNA extracted with the trizol–chloroform–isopropanol method. rRNA was depleted with MICROBExpress Bacterial mRNA Enrichment Kit (Thermo). Libraries were prepared using the Illumina Stranded mRNA Prep kit skipping the poly-A mRNA capture step and sequenced on a NextSeq 2000 instrument using a 100-cycle kit (Illumina). Quality and quantity control of RNA and libraries were performed using an Agilent 4200 TapeStation and Qubit 4, respectively. Libraries were sequenced to produce 118-base-pair single end reads with an average sequencing depth of eight million reads per sample. Raw reads were mapped to a reference genome (NCBI Taxonomy ID 328812) using Kallisto.

Single-cell RNA sequencing analysis

To determine Gpr84 expression in intestinal immune cells, a single cell RNA sequencing dataset of intestinal leucocytes was re-analysed³⁸ using Seurat v.4 (ref. ⁸⁸). The data were deposited in GEO (GEO: GSE229321). In brief, data were filtered to remove cells with high mitochondrial reads (> 20%), low gene detection (< 200) and high gene detection (> 4,000). Data were then normalized using the logNormalize method, highly variable features were determined using the ‘vst’ selection method, and data were scaled. Principal component analysis was then performed on scaled data as a form of linear dimensional reduction; 50 principal components were computed and stored. As the majority of true signal was found to be captured in the first 20 principal components, only those were used for cell clustering. Clustering was performed using the FindClusters function of Seurat v.4 at a resolution of 0.25 and visualized using UMAP. After normalization and clustering, we used adaptively thresholded low-rank approximation⁸⁹ to impute the RNA count matrix and fill in technical dropouts.

In vivo phage treatment

The 18–20-month-old mice received 1 × 10⁹ pfu of phage supplemented with 100 μl 9% (w/v) sodium bicarbonate via oral gavage on days 1, 3 and 5 before testing on day 8.

In vitro phage treatment

P. distasonis (strain APCS1XY) and P. goldsteinii were grown in FAB to OD₆₀₀ = 0.2 at 37 °C anaerobically; 200 μl of this culture was added to 9.8 ml of fresh media with MOI = 1 phage and OD₆₀₀ monitored. The same was done with E. coli in lysogeny broth aerobically.

Immunohistochemistry and immunofluorescence of mouse brain

For all FOS quantification experiments, mice were placed into a 1.5 l white bucket (Gramercy) with clean bedding and a novel object (an upside-down 50 ml conical tube filled with dark liquid glued to a 5 cm × 5 cm clear plastic base) for 10 min and then euthanized 1 h later.

Mice were anaesthetized with isoflurane and perfused with ice-cold PBS and 4% PFA before decapitation and brain dissection. Brains were then fixed in 4% PFA overnight at 4 °C. For free floating immunofluorescence sections, brains were cut coronally using a Leica 1000S Vibratome at 50 μm, except for the NTS sections and hippocampal sections in Figs. 2e,f and 3o, which were cut at 100 μm. For immunohistochemistry, brains were processed into paraffin and cut at 6 μm.

For immunohistochemistry, after antigen retrieval for 15 min at 99 °C with citrate buffer, slides were incubated in 5% hydrogen peroxide in water to quench endogenous peroxidase activity. Slides were washed for 10 min in running tap water, 5 min in 0.1 M Tris and then blocked in 0.1 M Tris with 2% FBS. Slides were incubated in primary antibody overnight against glial fibrillary acidic protein (GFAP) (Dako; 1:25k). For GFAP, every 30th section was stained. For BrdU/NeuN quantification, two hippocampal sections were stained per mouse. Primary antibody was rinsed off with 0.1 M Tris for 5 min, and then incubated with goat anti-rabbit (Vector) or horse anti-mouse (Vector) biotinylated IgG in 0.1 M Tris/2% FBS 1:1000 for 1 h. Biotinylated antibody was rinsed off with 0.1 M Tris for 5 min, then incubated with avidin-biotin solution (Vector) for 1 h. Slides were then rinsed for 5 min with 0.1 M Tris and then developed with ImmPACT DAB peroxidase substrate (Vector) and counterstained briefly with haematoxylin. Immunohistochemistry sections were scanned with a 3DHISTECH Laminar Scanner (Perkin Elmer) and quantification was done with QuPath⁹⁰.

Free-floating sections were blocked with PBS with 0.1% Triton (PBSTB; Thermo Fisher Scientific) and 1% bovine serum albumin (BioWorld) for 30 min at room temperature. After blocking, primary antibody staining was conducted overnight at 4 °C in PBSTB. The primary antibodies used were FOS (CST; 1:1500), NeuN (Millipore; 1:500), Iba1 (Wako; 1:500) and BrdU (Abcam; 1:500). Sections were washed three times in PBS and incubated in PBSTB with secondary antibodies (donkey anti-rabbit, donkey anti-mouse or donkey anti-rat fluorescent antibodies; Invitrogen Alexa Fluor; 1:500) for 2 h at room temperature. After three washes in PBS, the sections were mounted onto charged glass slides (Globe Scientific) and coverslipped with Vectashield (Vector) antifade aqueous mounting medium.

Imaging and analysis

Sections were imaged on a Zeiss LSM 710 confocal microscope with a 10 × 0.45 NA objective. The entire thickness of the section was imaged at 5-μm intervals and maximum intensity projections were used for analysis. FOS positive cells were quantified using ImageJ. Each data point is a single mouse, with one to two sections used per mouse. Scale bars indicate 100 μm, unless specified otherwise.

Golgi stain, imaging and analysis

Golgi staining was performed according to an established protocol⁹¹, imaged on a microscope and analysed using the AnalyzeSkeleton plugin in ImageJ (ref. ⁹²).

Untargeted metabolomics using LC–MS

Metabolites in 20 µl of bacteria culture supernatant were extracted with 80 µl ice-cold acetonitrile:methanol:water (40:40:20) solution. Following vortexing and centrifugation at 16,000 g for 10 min at 4 °C, 50 µl of supernatant was loaded into MS vials. Metabolites were analysed by quadrupole-orbitrap mass spectrometer (Q Excative Plus, Thermo Fisher Scientific) coupled to hydrophilic interaction chromatography via electrospray ionization. Liquid chromatography separation was on an Xbridge BEH amide column (2.1 mm × 150 mm, 2.5 µm particle size, 130 Å pore size; Waters) at 25 °C using a gradient of solvent A (5% acetonitrile in water with 20 mM ammonium acetate and 20 mM ammonium hydroxide) and solvent B (100% acetonitrile). The flow rate was 150 µl min^–1. The liquid chromatography gradient was: 0 min, 90% B; 2 min, 90% B; 3 min, 75% B; 7 min, 75% B; 8 min, 70% B; 9 min, 70% B; 10 min, 50% B; 12 min, 50% B; 13 min, 25% B; 14 min, 20% B; 15 min, 20% B; 16 min, 0% B; 20.5 min, 0% B; 21 min, 90% B; 25 min, 90% B. The autosampler temperature was set to 4 °C and the injection volume of the sample was 3 µl. Mass spectrometry data were acquired in negative and positive ion modes with a full-scan mode from 70 to 830 m/z at m/Δm = 140,000 resolution. Data were analysed using the Compound Discoverer (Thermo Fisher Scientific) software for automated peak picking. For known metabolites, the identify was confirmed by authentic standards based on retention time, precursor ion, and tandem-mass spectrum generated in-house and Human Metabolome Database. For data analysis, we focused on compounds from negative mode acquisition that were identified by name.

Targeted LC–MS of MCFAs

Approximately 10 mg of lyophilized, powdered caecal contents were homogenized in 200 µl of 80% methanol. A 50 µl aliquot of homogenate was spiked with 10 µl of ¹³C-labelled fatty acid internal standards followed by 50 µl of 200 mM O-benzylhydroxylamine in pH 5 ammonium acetate buffer and 10 µl of 1000 µM 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide for derivatization. Samples were vortexed and derivatized at room temperature for 10 min. A 100 µl aliquot of deionized water and 600 µl of ethyl acetate were added and the sample was vortexed and centrifuged at 18,000 × g at 4 °C. A 50 µl aliquot of the ethyl acetate layer was dried under nitrogen at 35 °C and reconstituted with 100 µl of 50% acetonitrile in water in a 96-well plate. Calibration standards were similarly prepared. Decanoic acid and dodecanoic acid were quantitated by multiple reaction monitoring using an Agilent 1290 Infinity UHPLC/6495B triple quadrupole mass spectrometer.

Statistical analysis and reproducibility

Data are presented as mean ± s.e.m. Replicates represent biologically independent samples. The significance of the differences between groups was evaluated using ANOVA or a Mann–Whitney U-test, unless indicated otherwise; a P value of <0.05 was considered significant. Sidak’s post-hoc test was used to correct for multiple comparisons. For sequencing experiments and metabolomics, P values were adjusted using the Benjamini–Hochberg procedure. Two-sided testing and paired testing were used where indicated. Statistical analysis was performed in GraphPad PRISM 10.

To rank members of the microbiome for subsequent functional testing, we have used the following priority score:

where \(\frac{\sum \,({x}_{i}-x)({y}_{i}-y)}{\sqrt{\sum {({x}_{i}-x)}^{2}\sum {({y}_{i}-y)}^{2}}}\) is the correlation coefficient between age x and taxonomic abundance y.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.