Early-life gut mycobiome core species modulate metabolic health in mice

Gnotobiotic mouse model

Seven to thirteen-week-old, germ-free C57Bl/6J mice were obtained from the International Microbiome Center (IMC) gnotobiotic mouse facility at the University of Calgary. Female mice were orally gavaged twice, 2 days apart, with 100 μl of a consortium of organisms, that was prepared in house. These consortia included (1) the Oligo-MM12³³ alone “B” (A. muris KB18, A. muciniphila YL44, B. caecimuris I48, B. animalis YL2, B. pseudococcoides YL58, C. innocuum I46, E. clostridioforme YL32, E. faecalis KB1, F. plautii YL31, L. reuteri I49, M. intestinale YL27, and T. muris YL45) or (2) in combination with C. albicans (This strain was isolated from infant stool samples enrolled in a clinical study at the Alberta Children’s Hospital in Calgary, Canada⁸⁰) “B + C”, (3) R. mucilaginosa (DSMZ; DSM 70825) “B + R”, or (4) M. restricta (ATCC; MYA4611) “B + M”. Experiments were completed with groups B, B + C and B + R run simultaneously, or B and B + M run simultaneously. Each colonization group was kept in a separate gnotobiotic isolator. Inoculum were prepared under anaerobic conditions by combining 100 μl of 3-day-old cultures of each bacterial species, 1-day-old culture of C. albicans, 2-day-old culture of R. mucilaginosa and 7-day-old culture of M. restricta. Bacteria were grown in fastidious anaerobe broth at 37 °C (Lab M Limited, UK); C. albicans and R. mucilaginosa were grown in yeast-mold broth at room temperature (YM; Becton Dickson, USA); and M. restricta was grown in modified Dixon media (mDixon)⁸¹ at 30 °C. mDixon was prepared in house by combining 36 g/L malt extract (Becton Dickson), 20 g/L desiccated Ox-bile (Sigma-Aldrich, USA), 10 mL/L tween-40 (Sigma-Aldrich), 6 g/L peptone (Sigma-Aldrich), 2 mL/L glycerol (VWR), 2 mL/L Oleic Acid (Sigma-Aldrich), and 15 g/L agar (Thermo Fisher, USA), and was adjusted to pH 6. After the initial gavage, females were paired with males for mating at a 2:1 ratio, respectively. To support early-life exposure to fungal species, fungal cultures were spread on the dams’ abdomen and nipple regions twice on days 2–6 after birth⁸². Mice were kept at a maximum of five animals per cage, under a 12-h light/dark cycle, 40% relative humidity, 22–25 °C and ad libitum access to sterile food and water. All experiments were conducted under protocols approved by the University of Calgary Animal Care Committee, following the guidelines of the Canadian Council on Animal Care (AC20-0176).

DIO model

At 21-days of age, mice were weaned onto a powdered SD (LabDiet JL Rat and Mouse/Auto 6F 5K52, Richmond, IN, USA) or HFHS diet (HFHS; DYET#1013915, Bethlehem, PA, USA), and remained on the diets until the experimental endpoint at 12 weeks of age (Males—SD: n_B = 23, n_B+C = 8, n_B+R = 10, n_B+M = 12; Females—SD: n_B= 18, n_B+C= 8, n_B+R = 8, n_B+M = 12; Males—HFHS: n_B= 22, n_B+C = 10, n_B+R = 8, n_B+M = 12; Females—HFHS: n_B = 20, n_B+C = 12, n_B+R = 7, n_B+M = 10). To ensure sterility, the HFHS diet was irradiated (5–20 kGy) and both diets were autoclaved at 132 °C for 30 min. Mice had ad libitum access to food and water, and food intake was measured at 3, 5, 7 and 9 weeks of age as the average difference in food jar weight over 3 consecutive days. Body weight was measured weekly and whole-body fat and lean mass were measured at 12 weeks of age, following export from gnotobiotic isolators by time-domain nuclear magnetic resonance (TD-NMR) with the Minispec LF90II (Bruker, USA). Wet weight of peri-gonadal adipose tissue, liver, kidney and heart was determined at the experimental endpoint.

Quantification of viable fungi

Fecal fungal load was first assessed by culturing samples collected at 6–7 weeks of age on selective medium. Mouse fecal pellets were weighed and homogenized in 500 μl of sterile phosphate-buffered saline (PBS; Corning, USA) at 15 Hz, for 2 min. Homogenates were serially diluted in sterile PBS. Dilutions were spread on YM (B + C, B + R) or mDixon (B + M) agar plates supplemented with gentamycin (10 μg/ml; Sigma-Aldrich) and chloramphenicol (200 μg/ml; Sigma-Aldrich). Plates were incubated at room temperature for 3 (YM) or 10 (mDixon) days before colonies were counted to determine viable fungal concentration. Blood along with kidney and liver homogenates (homogenized in 1 mL of sterile PBS with a steel bead at 30 Hz, for 3 min) from B + C mice were also plated as described above.

gDNA isolation from fecal samples

Genomic template DNA was isolated from mouse fecal samples at the specified time points with the DNeasy PowerSoil Pro kit (Qiagen, Germany) according to the manufacturer’s instructions with an added cell lysis step at TissueLyser II bead beater (Qiagen) and eluted in nuclease-free water. Samples were stored at −20 °C until further processing for quantitative polymerase chain reaction (qPCR) or 16S rRNA amplicon sequencing.

Fungal qPCR

Fungal DNA concentration in fecal samples at 12 weeks of age was determined by quantitative amplification of the fungal ITS spacer region⁸³. All qPCR reactions were carried out with the StepOne Plus Real-Time PCR System (Applied Biosystems, USA). Reactions were performed at 20 μl, consisting of 10 μl of iQ SYBR Green Supermix (BioRad Laboratories, USA), 100 nM ITS1F (5’-CTTGGTCATTTAGAGGAAGTAA-3’), 500 nM ITS4 (5’-TCCTCCGCTTATTGATATGC-3’), 1 ng of template DNA and topped up with nuclease-free water. The thermocycler program consisted of an initial 5-min step at 95 °C, followed by 32 cycles of 95 °C for 30 s, 55 °C for 30 s, 72 °C for 1 min, and a final melt curve of 95 °C for 15 s, 60 °C for 1 min, 95 °C for 15 s and 60 °C for 15 s. All samples were run in duplicate, and concentration was calculated using a standard curve of known concentration of C. albicans gDNA.

16S rRNA amplicon sequencing

DNA samples were sent to Microbiome Insights (Vancouver, Canada) where PCR was used to amplify the V4 region of the 16S rRNA gene with 515F/806R primers⁸⁴ using Phusion Hot Start II DNA Polymerase (Thermo Scientific). Samples from each experiment (experiment 1: B, B + C and B + R; experiment 2: B and B + M) were sequenced in the same run. PCR reactions were cleaned up, normalized using the high-throughput SequalPrep Normalization Plate Kit (Applied Biosystems), and quantified accurately with the KAPA qPCR Library Quantification kit (Roche, Switzerland). This generated ready-to-pool, dual-indexed amplicon libraries⁸⁵. Controls without template DNA and mock communities with known amounts of selected bacteria or fungi were included in the PCR and downstream sequencing steps to control for microbial contamination. The pooled and indexed libraries were denatured, diluted, and sequenced in paired-end modus on an Illumina MiSeq (Illumina Inc., USA).

Amplicon sequence processing

Sequence processing was conducted in R v.4.1.1⁸⁶. Sequences were checked for quality, trimmed, merged, and checked for chimeras using the DADA2 v1.20.0 pipeline⁸⁷. Taxonomy was assigned as amplicon sequence variants (ASVs) using a custom build database, containing the 16S rRNA amplicon sequences of the Oligo-MM12 strains⁸⁸. Any ASVs that did not align with the Oligo-MM12 strains were assigned as “Other”. The ASV table was preprocessed using the Phyloseq package v.1.36.0⁸⁹. Overall, 191 unique ASVs were detected. Samples with less than 1000 sequencing reads were excluded and ASVs appearing in only one sample were removed. The remaining dataset was filtered for ASVs appearing at least three times in 20% of samples, leaving 15 unique ASVs. This dataset was used for all subsequent analyses.

Metabolomics processing

Fecal samples collected at the 12-week endpoint were processed for untargeted metabolomics. Samples from each experiment (experiment 1: B, B + C and B + R; experiment 2: B and B + M) were analyzed in the same run. Samples were homogenized in 5 volumes of 50% ice-cold methanol (mg:ml) at 30 Hz for 3 min, followed by a 30-min incubation on ice and spun at max speed for 10 min at 4 °C. Supernatant was removed and incubated at −80 °C for up to 1 week before diluting 10X in 50% methanol. Samples were frozen at −80 °C and delivered to the Calgary Metabolomics Research Facility. General metabolomics runs were performed on a Q Exactive HF Hybrid Quadrupole-Orbitrap Mass Spectrometer (Thermo Fisher) coupled to a Vanquish UHPLC System (Thermo Fisher). Chromatographic separation was achieved on a Syncronis HILIC UHPLC column (2.1 mm × 100 mm × 1.7 μm; Thermo Fisher) using a binary solvent system (A + B) at a flow rate of 600 μL/min. Solvent A, 20 mM ammonium formate pH 3.0 in mass spectrometry grade H₂O; Solvent B, mass spectrometry grade acetonitrile (Thermo Fisher) with 0.1% formic acid (%v/v). A sample injection volume of 2 μL was used. The mass spectrometer was run in negative full-scan mode at a resolution of 240,000 scanning from 50 to 750 m/z. Following data curation and annotation, 86 metabolites remained in the dataset for experiment 1 and 122 for experiment 2.

Fecal lipid analysis

Fecal lipid concentration was determined from samples collected at 9 weeks of age. Mouse fecal pellets were weighed and homogenized in 500 μl of sterile PBS (Corning) at 15 Hz for 2 min. Total lipid content was isolated from 100 μl of homogenate with Lipid Extraction Kit (Chloroform free; Cell Biolabs Inc., USA) and lipid concentration was quantified in duplicate with Lipid Quantification Kit (Fluorometric; Cell Biolabs Inc.) according to manufacturer instructions. Results were quantified with the SpectraMax i3x (Molecular Devices, USA) and normalized to sample weight.

Oral glucose tolerance test

At 12 weeks of age, food was removed from cages at 0530-0600, followed by export from gnotobiotic isolators. Following 6 h of fasting, mice were gavaged with 50% dextrose (Sigma-Aldrich) at 2 g/kg body weight and blood glucose measurements were performed on tail blood at 15, 30, 60, 90 and 120 min with the OneTouch Ultra 2 glucometer (Lifescan, Switzerland). Body weight and fasted blood glucose were measured within 30 min of initiation of the test.

Adipose tissue histology and analysis

Peri-gonadal adipose tissue was dissected and placed in Dulbecco’s Modified Eagle Medium without D-Glucose, L-Glutamine, Phenol Red or Sodium Pyruvate (DMEM; Gibco, USA) supplemented with 10% inactivated horse serum (HS; Sigma-Aldrich) and 12.5 mM HEPES (Gibco) for up to 1 h before a ~1 cm piece of tissue was placed in 10% formalin. After 48 h, tissues were transferred to 70% ethanol for up to 2 weeks. Tissues were paraffin-embedded, sectioned at 4 μm and H&E stained by Alberta Precision Laboratories (APL). Images were acquired on the Aperio AT2 slide scanning microscope (Leica Biosystems, Germany) at 20X magnification. Adipocyte size was quantified using the Adiposoft plugin⁹⁰ in ImageJ with exclusion on edges. Three to four representative image sections were analyzed per sample.

Liver histology and analysis

Following wet weight measurement of the whole liver, the left lateral lobe was dissected and placed in 10% formalin. After 24 h, tissues were transferred to 70% ethanol for up to 2 weeks. Tissues were paraffin-embedded, sectioned at 4 μm and H&E stained by APL. Images were acquired on the Aperio AT2 slide scanning microscope (Leica Biosystems) at ×20 magnification. Lipid droplets were quantified in ImageJ with the Analyze Particles function with 0.9 circularity. Three to four representative image sections were analyzed per sample.

Plasma lipid, hormone and cytokine quantification

Following completion of the OGTT, animals were anesthetized with isoflurane and blood was drawn by cardiac puncture with a heparinized needle. Blood was combined with Dipeptidyl peptidase 4 inhibitor (Sigma-Aldrich) and Aprotinin (Sigma-Aldrich) and placed on ice for up to 2 h. Samples were spun at 2000 × g for 10 min and plasma was separated and stored at −80 °C until further processing. Plasma lipids were quantified by APL. Metabolic hormones and cytokines were quantified with U-PLEX Metabolic Group 1 Mouse Assay kit on the MESO QuickPlex SQ 120 (Meso Scale Discovery, USA). Plasma was diluted 4X in PBS and the assay was carried out following the manufacturer’s instructions.

Preparation of fecal filtrates

Fecal samples were collected from mice at 6–13 weeks of age (B, B + C, B + R, B + M) and processed immediately. Approximately 200 mg of feces was weighed and diluted 1:5 with sterile PBS (Gibco). Samples were homogenized for 2 min at 15 Hz with the TissueLyser II bead beater (Qiagen) and centrifuged at 4000 × g at 4 °C for 5 min. The supernatant was filter sterilized with a 0.22 μm sterile syringe filter and stored at −80 °C for up to 1 week.

Enteroid culture and treatment

Murine jejunal enteroids were generated from intestinal crypts isolated from conventionally housed C57Bl/6J male mice at 20 weeks of age⁹¹, resuspended in Matrigel (Corning), and subsequently cultured in D-Mouse Complete Medium (50% WRN conditioned media and the other 50% comprised of advanced DMEM/F12 (Gibco) supplemented with GlutaMAX (2 mM), HEPES (10 mM), Penicillin-Streptomycin solution (Sigma), N2 supplement (1X), B27 supplement (1X), in addition to hEGF (50 ng/mL), N-Acetylcysteine (1 mM), Primocin (100 μg/mL; InvivoGen)) + Y27632 (10 μM). Enteroids-derived monolayers were seeded between passages 7–12 at 1 × 10⁶ cells per Matrigel-coated 3 μm transwell insert. Monolayers were grown in 2D-Mouse complete media (50% comprised of WRN conditioned media and the other 50% advanced DMEM/F12 (Gibco) supplemented with glutamax (200 mM), HEPES (1 M), Penicillin-Streptomycin solution (Sigma), N2 supplement (1X), B27 supplement (1X), in addition to hEGF (50 ng/mL), Primocin (100 μg/mL; InvivoGen)) + Y27632 (10 μM) and media was changed the day after seeding and every other day after that. Once transepithelial electrical resistance reached 1000Ω, media containing 10% fecal filtrates was added to the transwell. 2D-Mouse complete media containing 20 ng/ml of IL-17 (R&D Systems) was added to the media outside of the transwell. Following 24 h of treatment, monolayers were collected in Trizol and stored at −80 °C.

RNA isolation and qPCR

RNA was isolated from 4 pooled monolayers with the RNeasy Mini kit (Qiagen). Reverse transcription was performed with the All-in-One 5X RT MasterMix (Applied Biological Materials Inc., Canada) and cDNA quality was assessed using the NanoDrop spectrophotometer (Thermo Fisher). PCR reactions were performed with iQ SYBR Green Supermix (BioRad) on the QuantStudio 3 Real-Time PCR System (Applied Biosystems). Twenty μL reactions were run in triplicate with the following cycling protocol: initial 2-min step at 94 °C, followed by 40 cycles of 94 °C for 15 s and 50 °C for 1 min and a final melt curve of 95 °C for 15 s, 60 °C for 1 min, 95 °C for 15 s and 60 °C for 15 s. Primers were as follows: Actb (forward, 5’- GTGACGTTGACATCCGTAAAGA-3’; reverse, 5’-GCCGGACTCATCGTACTCC-3’), Cd36 (forward, 5’-ATGGGCTGTGATCGGAACTG-3’; reverse, 5’-GTCTTCCCAATAAGCATGTCTCC-3’); Slc5a1 (forward, 5’- TGGTGTACGGATCAGGTCATTG-3’; reverse, 5’- TTCAGATAGCCACACAGGGTACAG-3’). Fold gene expression was calculated with the 2^−ΔΔCt method with Actb serving as the housekeeping gene.

Adipose tissue immune cell isolation

Peri-gonadal adipose tissue was dissected and placed in DMEM (Gibco) supplemented with 10% HS (Sigma-Aldrich) and 12.5 mM HEPES (Gibco) on ice for up to 1 h. The wet weight of the tissue was obtained, and samples were mechanically minced and placed back in the media. One mg/mL of collagenase Type II (Sigma-Aldrich) and 5KU/mL of DNase I (Sigma-Aldrich) were added to the media and samples were incubated at 100 rpm 37 °C for 20 min with manual shaking every 5 min. Samples were supplemented with 0.1 M EDTA and incubated at 37 °C with 10 rpm shaking for another 5–10 min. Once digestion was complete, samples were filtered through 100 μm mesh and topped up with 25 mL DMEM solution. Samples were spun at 300 × g for 5 min at 4 °C. Supernatant was aspirated, and red blood cell lysis was performed with 2 mL ACK Lysing Buffer (Quality Biological) at room temperature for 2 or 7 min for samples from SD or HFHS mice, respectively. Samples were topped up with 8 mL cold PBS supplemented with 2% HS (Sigma-Aldrich), filtered through 250 μm mesh, and spun at 300 × g for 5 min at 4 °C. Supernatant was aspirated, and the cell pellet was resuspended in RPMI Medium 1640 (Gibco) supplemented with 10% HS.

Splenocytes isolation

Spleens were dissected and placed in RPMI (Gibco) supplemented with 5% HS (Sigma-Aldrich) on ice for up to 2 h. The wet weight of the tissue was obtained, and samples were mechanically minced. One mg/mL of Collagenase Type IA (Sigma-Aldrich) and 30 U/mL of DNase I were added to the media and samples were incubated at 220 rpm 37 °C for 25 min with mechanical shaking halfway through. Samples were filtered through a 100 μm cell strainer, topped up with 5 mL PBS supplemented with 2% HS and spun at 400 × g, 5 min 4 °C. Supernatant was removed and red blood cell lysis was performed with 5 mL ACK Lysing Buffer (Quality Biological) at room temperature for 7 min. Samples were topped up with 15 mL cold PBS supplemented with 2% HS (Sigma-Aldrich) and spun at 400 × g for 5 min at 4 °C. Supernatant was removed, and the pellet was resuspended in RPMI (Gibco) supplemented with 10% HS.

Immunolabelling and flow cytometry

Isolated immune cells from adipose tissue and splenocytes were stained with Fixable Viability Stain (eF506; eBioscience) and stained for extracellular immune cell markers. Cells were then fixed and permeabilized with the Foxp3/Transcription Factor Staining Buffer Set (eBioscience, USA) and stained for intracellular immune cell markers. Samples were stained with a lymphoid and myeloid panel. The following antibodies were included in the lymphoid panel: anti-CD45 AF532 (clone 30-F11; eBioscience), anti-TCRb BV750 (clone H57-597; BD), anti-γδTCR PE-CF594 (clone GL3; BD), anti-CD4 BV786 (clone RM4-5; BD), anti-CD8 BV570 (clone 53-6.7; BioLegend), anti-CD19 PerCP-Cy5.5 (clone 1D3; BD), anti-NK1.1 BUV395 (clone PK136; BD), anti-CD90.2 BUV496 (clone 30-H12; BD), anti-T-bet BV421 (clone 4B10; BioLegend), anti-GATA3 BV711 (clone L50-823; BD), anti-FOXP3 AF488 (clone FJK-16s; eBioscience), anti-RORγt PE (clone B2D; eBioscience), anti-Eomes PE-Cy5 (clone Dan11mag; eBioscience), anti-IL-5 eF450 (clone TRFK5; eBioscience), anti-IL-10 BV605 (clone JES5-16E3; BD), anti-IFN-γ BV650 (clone XMG1.2; BD), anti-IL-17A PE-Cy7 (clone eBio17B7; eBioscience). The following antibodies were included in the myeloid panel: anti-CD45 AF532 (clone 30-F11; eBioscience), anti-CD11b BUV395 (clone M1/70; BD), anti-SigF BV786 (clone E50-2440; BD), anti-Ly6G BV750 (clone 1A8; BD), anti-CD64 BV650 (clone X54-5/7.1; BD), anti-CD11c AF647 (clone N4/18; BioLegend), anti-MHCII eF450 (clone M5/114.15.2; eBioscience), anti-CD103 PE-Dazzle594 (clone M290; BioLegend), anti-Ly6C BV570 (clone HK1.4; BioLegend), anti-CX3CR1 BV421 (clone SA011F11; BioLegend), anti-Tim4 PE-Cy7 (clone RMT4-54; Biolegend); anti-CD206 AF700 (clone C068C2; Biolegend), anti-CD9 FITC (clone MZ3; BioLegend). Samples were run on the Cytek Aurora (Cytek Biosciences, USA). The same gating strategy was used for both tissue types, except for macrophage populations, due to low CD64⁺ population in spleens and our interest in WAT-specific macrophage populations.

Statistics and reproducibility

Sample size calculations were not performed. We aimed to achieve n = 8–10 animals/group based on the literature for DIO studies in mice^92,93,94, though some limitations in this were due the distribution of mouse sex among the litters. Mouse litters were assigned to the different diet groups in an attempt to evenly distribute litter sizes. This was dependent on breeding rates and birth. Mice were housed in different isolators based on colonization status (i.e., B, B + C, B + R, B + M). The Investigators were not blinded to allocation during experiments and outcome assessment as there was no subjective component. Data was excluded from certain analyses if they were identified as outliers by ROUT (Q = 1%). This is specified in the figure captions. Given the low expression levels of Cd36 and Slc5a1 in jejunal enteroids, samples were excluded from qPCR analysis if triplicates differed by greater than 2 cycles⁹⁵.

Bacterial alpha-diversity was assessed with the Shannon diversity index using vegan v2.6-4⁹⁶. Data was assessed for normality with the Shapiro–Wilk test, and differences between fungal colonized groups and their respective “B” control groups (sequenced in the same run) were assessed by unpaired t-test (parametric) or Mann–Whitney (non-parametric) test accordingly. Boxplots display the median and hinges, with the lower and upper hinges displaying the 25th and 75th percentile, respectively. Whiskers extend from the hinge to the lowest/largest value below 1.5X the inter-quartile range. Beta-diversity was assessed with the Bray–Curtis dissimilarity index with variance-stabilizing transformation and visualized with principal coordinate analysis (PCoA), where ellipses represent the 95% confidence interval. Differences between fungal colonized groups and their respective B-only group, along with sex differences, were assessed by permutational ANOVA at 3 weeks, with the addition of a diet effect at 12 weeks using vegan v2.6-4⁹⁶. The relative abundance of the 12 bacterial strains was compared between fungal colonized groups and their respective B-only group by unpaired t-test or Mann–Whitney test depending on the normality of the data, as assessed by the Shapiro–Wilk test.

For metabolic data collected at a single time point, data was first assessed by Three-way ANOVA to determine the presence of sex effect. If present, males and females were analyzed separately; otherwise, males and females were pooled for analysis. Outliers were identified with the ROUT method (Q = 1%) and excluded as indicated in the figure legends. The effect of diet and colonization was then analyzed by Two-way ANOVA, where fungal colonized groups were compared to B-only groups pooled from all experiments within each diet with Dunnett’s multiple comparison test. Longitudinal measures (i.e., body weight, OGTT), were first analyzed by RM-ANOVA (MANOVA.RM v0.5.4⁹⁷), where colonization, diet and sex were included as the between subject factors and time was the within subject factor. If the effect of sex was significant, males and females were analyzed separately. Individual time points in either sex for body weight, during the OGTT and trapezoidal AUC were compared using Two-way ANOVA with Dunnett’s multiple comparison test. Immune data for either sex on either diet were assessed for normality by Shapiro–Wilk test, and fungal colonized groups were compared to B-only groups by One-way ANOVA with Dunnett’s multiple comparison test or Kruskal–Wallis test with Dunn’s multiple comparison test, accordingly. Bar and line plots display the mean with standard error of the mean (SEM).

Metabolomics data was processed with Metaboanalyst v5.0⁹⁸. Metabolite concentrations were normalized by the median, square root transformed and subject to Pareto scaling. Euclidean distance calculations from PCA scores and permutational ANOVA were performed using vegan v2.6-4⁹⁶. Fungal groups were compared to the “B” group from the same experiment that was processed in the same batch. Direction of comparison for fold change analysis was specified as fungal groups/B-only and test parameters were defined as non-parametric with unequal variance and false discovery rate (p < 0.05) correction. MIMOSA2 v2.0.0⁴⁴ was performed using a customized AGORA database that included only the Oligo-MM12 strains and the same ASVs identified with DADA2 were mapped to the database with a similarity threshold of 0.99. Metabolite data was log-transformed, and the regression model was fit using rank-based estimation with the significance threshold set to 0.05.

For Spearman correlation analysis, study variables were grouped as colonization, metabolism, bacterial abundance or WAT immunity. Data underwent mean normalization and correlations between variables from the different groups were identified using Hmisc v5.1.1⁹⁹ package. Significant correlations (coefficient > |0.6|) were plotted with circlize v0.4.15¹⁰⁰.

Flow cytometry data was gated using FlowJo (v10.8.1). Graphs were made using either RStudio (v2022.12.0 + 353) or GraphPad Prism v9.5.1.

Reporting summary

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