Mast cells link immune sensing to antigen-avoidance behaviour – Nature

Mice

BALB/c Cpa3^Cre/+ (ref. ⁹), C57BL/6 Cpa3^Cre/+ (ref. ⁹), BALB/c Mcpt8–Cre⁵¹, BALB/c Cpa3^−/− (ref. ⁵²), BALB/c Cpa3^Y356L,E378A (ref. ⁵³), BALB/c Hdc^−/− (ref. ⁵⁴), BALB/c Mcpt6^−/− (ref. ⁵⁵) and Nr4a1–GFP⁵⁶ mice on C57BL/6 or (BALB/c × C57BL/6)F1 background were maintained in the mouse breeding facilities of the DKFZ Heidelberg. BALB/c Igh-7^−/− (ref. ¹¹) mice were bred at Imperial College London and used at the Interdisciplinary Neurobehavioral Core, Heidelberg University, for experiments. BALB/c wild-type mice and C57BL/6 Wnt1|GCaMP3 (Wnt1-Cre;R26R-GCaMP3)^57,58 mice were maintained in the mouse breeding facilities of the KU Leuven. Mice were housed with a 12-h day–night cycle in a controlled environment at 20–24 °C and 45–65% humidity. All behavioural tests were performed on adult (>7 weeks old) female and male mice. Controls for Cpa3^Cre/+, Mcpt8–Cre, Cpa3^Y356L,E378A, Cpa3^−/− and Mcpt6^−/− mice were gender-matched and age-matched wild-type littermates. Controls for Igh-7^−/− and Hdc^−/− mice were gender-matched and age-matched wild-type mice housed in the same animal room. All animal experiments were performed in accordance with institutional and governmental regulations. Experiments in Heidelberg were approved by the Regierungspräsidium Karlsruhe, Germany. Experiments in Leuven were approved by the Animal Care and Animal Experiments Committee of the KU Leuven, Leuven, Belgium.

Immunizations

Mice were actively immunized by intraperitoneal injection with 25 μg OVA (Sigma-Aldrich) complexed with 2 mg Al(OH)₃ (alum) (InvivoGen) on days 0 and 14, and avoidance tests started on day 20. For Th2 cytokine treatment, mice were injected intraperitoneally with a cocktail composed of 2 μg IL-3 (Peprotech), 12 μg anti-mouse IL-3 (MP2-8F8, BioLegend), 2 μg IL-4 (Peprotech), 12 μg anti-mouse IL-4 (11B11, BioLegend) and 0.4 μg IL-9 (Peprotech) on days 2 and 16. Mixing IL-3 with the anti-IL-3 antibody MP2-8F8 and IL-4 with the anti-IL-4 antibody 11B11 generates cytokine–antibody complexes that show increased activity in vivo⁵⁹, which we exploited here to increase the magnitude and duration of cytokine effects in vivo. No such effects have been described for IL-9; hence no anti-IL-9 antibody was used. For passive sensitization, mice were injected intraperitoneally on day 0 and intravenously on day 9 with 10 μg anti-OVA IgE monoclonal antibody (E-C1, Chondrex), and intragastric challenge (anaphylaxis) or avoidance test started on day 10.

IntelliCage

Keeping mice in IntelliCages prevents stress induced by handling and thus enables observation of natural behaviour⁶⁰. For systemic immunization, mice received two injections of OVA-alum on days 0 and 14 as described above. On day 17, mice received a subcutaneous implant of a unique radio-frequency identification (RFID) transponder into the nape under isoflurane anaesthesia and were placed in an IntelliCage apparatus (TSE Systems) in groups of 13–16 animals. Mice were kept on a 12-h light–dark cycle with ad libidum access to chow and water. After acclimatization (1–3 days), the water in four of eight bottles was exchanged for 20% (v/v) egg white water containing either 0.25% (w/v), 1% (w/v) (C57BL/6) or 8% (w/v) (BALB/c) sucrose. Drinking behaviour was analysed for up to 14 days. Data were collected using the IntelliCage Plus software (NewBehavior AG).

For passive sensitization, mice received two injections of monoclonal anti-OVA-IgE on days 0 and 9 as described above. On day 11, mice received a subcutaneous implant of a unique RFID transponder into the nape under isoflurane anaesthesia and were placed in the IntelliCage apparatus. Mice were kept on a 12-h light–dark cycle with ad libidum access to chow and water. After acclimatization, the water in four of eight bottles was exchanged for 20% (v/v) egg white water containing 8% (w/v) sucrose. Drinking behaviour was analysed for up to 14 days. Data were collected using the IntelliCage Plus software (NewBehavior AG).

Egg white preparation

Egg white (20% v/v) was prepared as follows: chicken egg white was separated from its yolk, diluted in water and strained through filter paper (grade 3 hw; 65 g m⁻²; Ahlstrom Munksjö). The 20% egg white solutions used for experiments contained on average 0.78 ± 0.17 endotoxin units per millilitre, which is substantially lower than that of standard mouse chow⁶¹.

Two-bottle test

One week after the second immunization, BALB/c Cpa3^+/+ and BALB/c Cpa3^Cre/+ or knockout mice and their corresponding littermates were individually housed. Cages were equipped with two identical bottles, one containing water and the other containing 20% egg white water with 8% sucrose. According to the results of sodium dodecyl sulfate polyacrylamide gel electrophoresis, the 20% egg white water solution contained approximately 40 mg ml⁻¹ OVA (not shown). Every 24 h, bottles were weighed, and the positions of the bottles were changed to control for a side preference in drinking behaviour. The avoidance tests were run for up to 7 days.

Multiple OVA gavage treatments

Mice received 50 mg OVA by intragastric gavage every 2–3 days (for 7 days treatment, 4× gavage; for 11 days, 6× gavage; for 16 days, 8× gavage). Mice were inspected for diarrhoea 1 h after each gavage. By comparison to this OVA gavage, voluntary consumption by mast-cell-deficient BALB/c Cpa3^Cre/+ mice in IntelliCage experiments (Fig. 1) was on average 586 mg OVA per day. Hence, the amount of antigen given by gavage was not higher than the amount of antigen consumed voluntarily by mast-cell-deficient mice.

Pharmacological inhibition of FLAP and 5-HTR3

Mice were immunized with OVA-alum as described above. On day 20, mice were deprived of drinking water overnight. Palonosetron (Sigma-Aldrich) was injected intraperitoneally at a dose of 0.5 mg kg⁻¹ 12 h before the avoidance test. MK-886 (Abcam) was administered by intragastric gavage at a dose of 10 mg kg⁻¹ 1 h before the avoidance test. As repeated dosing with MK-886 can alter mouse behaviour⁶², we only gave the mice a single dose, followed by a 1-day observation. IntelliCage (MK-886) and two-bottle tests (MK-886; palonosetron) were run for 24 h.

Analysis of mast cell activation in Nr4a1-GFP mice

Mice were immunized with OVA-alum as described above. On day 20, Nr4a1-GFP mice were housed individually. After a 12-h period of water deprivation, cages were equipped with a bottle containing 25% OVA (Sigma-Aldrich) in water. Control mice were given a bottle containing 25% bovine serum albumin (BSA, Roth) in water. After 3 h, the bottles were weighed to check for consumption from the bottles, and mice were euthanized for further analysis. Activation of purified mast cells was monitored by flow cytometry for GFP expression.

For analysis of anaphylaxis in Nr4a1-GFP mice, mice were immunized with OVA-alum as described above. On day 21, mice received intragastric gavage with 50 mg OVA (Sigma) or 50 mg BSA (Sigma), and their body temperature was monitored using a rectal thermometer. After 3 h, mice were euthanized for further analysis. Activation of purified mast cells was monitored as described above.

Pharmacological ablation of Trpv1-positive sensory neurons

Mice were immunized with OVA-alum as described above. Between the OVA immunizations, starting on day 10, resiniferatoxin (Cayman Chemicals) was injected subcutaneously into the flank in three escalating doses (30, 70 and 100 μg kg⁻¹) on consecutive days. Control mice were treated with vehicle (dimethyl sulfoxide in phosphate-buffered saline (PBS)). On day 20, Trpv1 denervation was confirmed by prolonged withdrawal latency of mice to noxious heat (52 °C) applied to the tail (tail flick test; data not shown). Beginning on the next day, OVA avoidance was analysed by the two-bottle test.

Tissue digestion

Gingival single-cell suspensions were prepared as previously described⁶³. In brief, the palate and mandible were isolated, and tissues were digested for 1 h at 37 °C in RPMI supplemented with 10% fetal calf serum (FCS; Sigma-Aldrich), 0.15 μg DNase I and 3.2 mg ml⁻¹ collagenase IV (all enzymes from Sigma-Aldrich). Then, 0.5 M EDTA (Roth) was added during the last 5 min, and supernatant was filtered through a 70-μm cell strainer (ThermoFisher). Undigested gingiva tissue was peeled from the palate and mandible and mashed through the same filter to yield the gingiva cell suspension.

Tongue single-cell suspensions were prepared by finely mincing the tongue and digesting the tissue for three rounds of 15 min at 37 °C in RPMI supplemented with 0.1 mg ml⁻¹ Liberase (Sigma-Aldrich) and 2.5 μg ml⁻¹ DNase I (Sigma-Aldrich). After each round of digestion, the cell suspensions were filtered through a 70-μm cell strainer (ThermoFisher), and new enzyme solution was added to the tissue. All fractions were combined to yield the tongue single-cell suspension.

For isolation of stomach intraepithelial leucocytes, the stomach was cut open and food remnants were removed. Stomachs were incubated for 15 min at 37 °C in HBSS supplemented with 20 mM EDTA (Roth) to release epithelial layers from the connective tissue. The cell suspension was applied to a spin column (ThermoFisher) packed with 100-μm zirconia beads (Roth). After centrifugation, the flowthrough was collected, yielding an intraepithelial cell suspension containing mucosal stomach mast cells.

For preparation of small intestine cell suspensions, small intestines were cut open and food remnants were removed. Intestines were incubated for 15 min at 37 °C in HBSS supplemented with 2% FCS (Sigma-Aldrich), 5 mM EDTA (Roth), 1 mM DTT (Merck) and 10 mM HEPES (Life Technologies) to release epithelial layers from the connective tissue²⁷. The cells in the soluble fraction (containing intraepithelial mast cells) were filtered through a 70-μm cell strainer (ThermoFisher). The remaining intestine tissue was washed in PBS and transferred into RPMI supplemented with 2% FCS (Sigma-Aldrich), 20 mM HEPES (Life Technologies), 0.2 mg ml⁻¹ collagenase IV (Sigma-Aldrich), 0.5 mg ml⁻¹ hyaluronidase I (Sigma-Aldrich) and 0.1 mg ml⁻¹ DNase I (Sigma-Aldrich). Digestion was carried out for 30 min at 37 °C, and digested tissue was filtered through a 100-μm cell strainer (ThermoFisher), yielding the lamina propria fraction (containing lamina propria mast cells).

Blood was drawn by cardiac puncture, followed by red blood cell lysis according to the manufacturer’s protocol (RBC Lysis Buffer, BioLegend).

Flow cytometry

Single-cell suspensions were centrifuged and incubated for 15 min with 200 μg ml⁻¹ mouse IgG (Jackson ImmunoResearch Laboratories) to block Fcγ receptors. After washing with PBS supplemented with 5% FCS (Sigma-Aldrich), cells were stained with fluorochrome-coupled antibodies (see list in the Antibodies section) for 20 min on ice and protected from light. Cells were washed and incubated with 100 nM SytoxBlue (Life Technologies) for dead cell exclusion. For absolute quantitation of cells, a defined number of 123 count eBeads (Life Technologies) were added to the samples before analysis with a BD LSRFortessa (Becton Dickinson). Data were analysed using FlowJo software (Treestar), using the gating strategies shown in Supplementary Fig. 1. Mast cells in the tongue and gingiva were gated as live CD45⁺MHCII⁻CD11b⁻CD117⁺FcεRI⁺ cells. Stomach mast cells were gated as live CD45⁺CD117⁺FcεRI⁺/IgE⁺ cells. Intestinal mast cells were gated as live CD45⁺CD3⁻CD11b⁻CD19⁻Gr-1⁻Ter119⁻CD117⁺FcεRI⁺ cells. Neutrophils were gated as live CD45⁺CD11b⁺Siglec-F⁻ Gr-1⁺/Ly6G⁺ cells. Basophils were identified as live CD45⁺CD90.2⁻CD11c⁻Gr-1⁻Siglec-F⁻MHCII⁻B220⁻CD49b⁺IgE⁺ cells. For reagents, see list in the Antibodies section.

Intracellular Ca²⁺ measurement of mast cells

For analysis of Ca²⁺ flux in stomach mast cells, stomach intraepithelial leucocytes were spun down and resuspended in calcium imaging buffer (125 mM NaCl, 3 mM KCl, 2.5 mM CaCl₂, 0.6 mM MgCl₂, 10 mM HEPES, 20 mM glucose, 1.2 mM NaHCO₃ and 20 mM sucrose, brought to pH 7.4 with NaOH) supplemented with 0.1% BSA (Roth), 2.5 mM probenecid (Biotinum), 0.01% Pluronic-F127 (Sigma-Aldrich) and 200 μg ml⁻¹ mouse IgG (Jackson ImmunoResearch Laboratories) to block Fcγ receptors. After washing with calcium imaging buffer, cells were stained with 4 μM Fluo-4 (Thermo Fisher Scientific) and CD45 BV421, CD117 PE and FcεRI APC antibodies for 30 min at room temperature in the dark. Cells were then washed with calcium imaging buffer supplemented with 0.1% BSA (Roth) and 100 nM SytoxBlue (Life Technologies). Cells were kept at 37 °C during measurements and analysed on a BD LSRFortessa (Becton Dickinson). After 30 s of baseline measurements, OVA (Sigma-Aldrich) was added to a final concentration of 1.25 mg ml⁻¹, and Fluo-4 fluorescence was acquired for 90 s. As a positive control, ionomycin (Sigma-Aldrich) was added to a final concentration of 16.4 mmol ml⁻¹ for the last 30 s of the measurement. Data were analysed using FlowJo software (BD Bioscience).

Intracellular serotonin staining

Stomach intraepithelial leucocytes were prepared as described above. Cells were incubated with ZombieFITC (1:500, BioLegend) for dead/live discrimination and 10 μg ml⁻¹ anti-CD16/32 antibodies (39, BioLegend) to block Fcγ receptors for 15 min at room temperature. After washing with PBS supplemented with 5% FCS (Sigma-Aldrich), cells were stained with CD45 BV785, CD117 APC and IgE BV421 (R35-72, BD Bioscience) for 20 min on ice and protected from light. After washing and centrifugation, cells were fixed and permeabilized using a FoxP3-intracellular staining kit (BioLegend) according to the manufacturer’s instructions. Cells were stained with 0.11 μg ml⁻¹ anti-5-HT (5HT-H209, Dako) or isotype control antibodies for 30 min, washed in PBS and stained with anti-mouse-IgG1 (RMG1-1, BioLegend) antibodies for 30 min before analysis with a BD LSRFortessa (Becton Dickinson).

Serological analysis

OVA-specific IgE and IgG1 were measured by enzyme-linked immunosorbent assay as previously described⁶⁴. Anti-OVA IgG1 (L71, Biozol) and anti-OVA IgE (2C6, Invitrogen) were used as standards. Rat anti-mouse IgG1-HRP (1:2000, X56, BD Pharmingen) and rat anti-mouse IgE-HRP (1:2000, 23G3, SouthernBiotech) were used for detection. Serum samples were diluted 1:40000 (IgG1) and 1:10 (IgE). IL-4 and IL-6 were measured by LEGENDplex Mouse Th Cytokine (BioLegend) assay according to the manufacturer’s instructions.

Ca²⁺ imaging of full-thickness small intestine preparations

For ex vivo Ca²⁺ imaging, the ileum of immunized adult Wnt1|GCaMP3 mice or immunized and AAV9-transduced (pENN.AAV.CamKII.GCaMP6f.WPRE.SV40, Addgene) Cpa3^+/+ and Cpa3^Cre/+ mice was isolated. Tissues were opened along the mesenteric border and pinned flat in a Sylgard-lined dish containing Krebs solution (120.9 mM NaCl, 5.9 mM KCl, 1.2 mM MgCl₂, 1.2 mM NaH₂PO₄, 14.4 mM NaHCO₃, 11.5 mM glucose and 2.5 mM CaCl₂), bubbled with 95% O₂/5% CO₂ at room temperature. The luminal contents were cleared away with Krebs washes. Tissues were mounted over an inox ring and stabilized using a matched rubber O-ring⁶⁵. Ring preparations were placed on a glass-bottomed dish and imaged using a ×20 objective on an inverted Zeiss Axiovert 200M microscope equipped with a monochromator (Poly V) and a cooled CCD camera (Imago QE) (TILL Photonics). Preparations were constantly superfused with carbogenated Krebs solution at room temperature using a local gravity-fed (±1 ml min⁻¹) perfusion pipette. Krebs only, BSA (1% in Krebs) and OVA (1% in Krebs) were sequentially applied for 5 min each on to the mucosal surface using a perfusion pipette positioned above the imaged myenteric or submucosal plexus. Custom-written routines in Igor Pro (Wavemetrics)⁶⁶ were used for analysis. Heatmaps display normalized fluorescence (F_i/F₀) traces, and each row shows the fluorescence signal of an individual neuron under the control Krebs condition, followed by BSA (1%) and OVA (1%) mucosal perfusion. The signal of each neuron was normalized to baseline fluorescence under the Krebs condition. Traces depicted in the heatmaps are sorted top-down by the maximum amplitude of the signals detected. For percentages of activated neurons, regions of interest were drawn over each GCaMP-expressing neuron to calculate the average Ca²⁺ signal intensity normalized to the baseline (displayed as F_i/F₀). Background subtraction was performed on some recordings where changes in background intensity were apparent. Neurons were considered to be active if at least one neuronal Ca²⁺ peak was detected during each 5 min recording period.

Vagotomy

Vagotomy was performed as previously described⁶⁷. In brief, both vagal trunks were transected below the diaphragm. To ensure transection of all small vagal branches, neural and connective tissue surrounding the oesophagus was removed. Pyloroplasty was performed to avoid gastric dilatation due to vagotomy. Control mice underwent a sham operation, in which vagal trunks were exposed but not cut and pyloroplasty was performed.

RNA isolation from stomach and small intestine

Naive, immunized and challenged (days 5, 7 or 11 of the two-bottle test; or 4×, 6× or 8× OVA gavage) BALB/c Cpa3^+/+ and Cpa3^Cre/+ mice were euthanized by CO₂ asphyxiation. Stomach and small intestine pieces (4 cm proximal to the duodenum) were cut open, and food remnants, fat and Peyer’s plaques were removed. Tissues were immediately frozen in liquid nitrogen and ground in a mortar with a pestle. RNA isolation was performed according to the instructions of the PureLink RNA Mini kit (Invitrogen). Total RNA quality was determined by the RNA integrity number provided by the BioAnalyzer system (RNA 6000 Pico Kit, Agilent). Isolated RNA was stored at −80 °C until use.

RNA sequencing

Library preparation was performed with a TruSeq Stranded RNA Kit (Illumina) according to the manufacturer’s instructions. After library preparation, indexed samples were pooled and diluted to 2 nM with 2% PhiX spike in. The multiplexed library was then paired-end sequenced using NextSeq 1000/2000 P2 Reagents (200 Cycles) on the NextSeq 1000/2000 platform (conditions: Read1/Read2: 111 Cycles; Index1: 8 Cycles; Index2: 8 Cycles). Data were mapped using STAR aligner (v.2.5.2b)⁶⁸, and reads were annotated using the FeatureCounts algorithm from the subread package (v.1.5.1)⁶⁹. Both mapping and annotation were performed on Genome Reference Consortium Mouse Build 38 (GRCm38)⁷⁰. Count data normalization and differential expression analysis were performed using DESeq2 (ref. ⁷¹), comparing immunized unchallenged mice with immunized mice of the same genotype challenged by either the two-bottle test or OVA gavage.

Principal component analysis was performed on read counts normalized by the variant stabilizing transformation included in the DESeq2 package, based on the top 500 most variable genes. For subsequent data analysis and visualization, shrinkage towards zero of log₂ fold changes (lfcshrink) was computed using the apeglm implementation⁷². P values were adjusted using the Benjamini–Hochberg algorithm, and results were accepted as significant if adjusted P < 0.05.

Gene set enrichment analysis (GSEA) for each experimental group was performed on the complete dataset ranked with lfcshrink using ClusterProfiler⁷³. All gene ontology (GO) terms related to ‘biological processes’ in the org.Mm.eg.db database⁷⁴ were considered. Significantly (Benjamini–Hochberg adjusted P < 0.05) enriched GO terms were accepted for further processing. Based on the individual descriptions, enriched GO terms were manually annotated into four immune-related subgroups: innate immunity, adaptive immunity, mucosal immunity and chemotaxis (annotation of individual GO terms can be found in Supplementary Table 1). From subgrouped GO terms, core enrichment genes (genes that contribute most to the enrichment results for a GO term; GSEA documentation on the Broad website) were extracted and filtered (log₂ fold change <−1 or log₂ fold change >1.5 in at least one comparison). Fold changes of these genes were compared between experimental groups (Fig. 3a,c and Extended Data Fig. 7b,c).

Genes contained in the ‘hallmark inflammatory response’ gene set were retrieved from the Molecular Signatures Database (hallmark gene set collection: M5932, MSigDB)¹⁴, and their fold changes in wild-type mice under avoidance and non-avoidance conditions were compared.

In the volcano plots, horizontal lines indicate the significance threshold and vertical lines indicate the log₂ fold change thresholds of −3 and 3, respectively. GSEA core enrichment genes are indicated in red, and manually curated mast-cell-related genes are depicted in blue (Fig. 3c–f). DEG with log₂ fold change >3 in the stomach (Fig. 3c,d) and intestine (Fig. 3e,f) were filtered and displayed as a heatmap (Fig. 3g). In this heatmap, genes not significantly differentially expressed in other comparisons are depicted as white squares. Genes with significant (P < 0.05) changes in expression have coloured squares. The colour scale indicates log₂ fold change.

Analysis of anxiety-like behaviour in Cpa3
^Cre/+ mice

Anxiety-like behaviour was tested using an elevated plus maze, open field test and light–dark test. Mice are averse to brightly lit open areas. However, they have a natural drive to explore a perceived threatening stimulus. Low levels of anxiety lead to increased exploratory behaviour, whereas high levels of anxiety lead to less locomotion and more time spent in enclosed areas. Tests for anxiety-like behaviour were performed between 09:00 and 13:00. C57BL/6 Cpa3^+/+ and Cpa3^Cre/+ mice were brought into the behavioural room 30 min before behavioural testing began. All experimental setups were cleaned with soap and 70% ethanol at the end of the measurements.

The elevated plus maze consisted of an opaque-grey plastic apparatus with four arms (6 cm wide and 35 cm long), two open (illuminated with 100 lux) and two closed (20 lux), set in a cross from a neutral central intersection (6 × 6 cm) and elevated 70 cm above the floor. The mice were placed in the centre of the maze, and 5-min test sessions were digitally recorded and analysed with Sygnis Tracker software (Sygnis).

For the open field test, mice were placed in the centre of a bright open arena (40 cm wide, 40 cm long and 40 cm high; 290 lux), and their behaviour was monitored with a digital camera and ANY-maze video tracking system (Stoelting Co.) for 10 min.

The behavioural test box used for the light–dark test consisted of two compartments, a 29 × 21-cm (21 cm high; 300 lux illumination) lit compartment and a 15 × 21-cm (21 cm high; 10 lux illumination) dark compartment, connected by an opening at floor level allowing transition between the compartments. Mice were placed in the dark compartment, and their behaviour was monitored with a digital camera and Sygnis Tracker software for 10 min.

Analysis of general behaviour of Cpa3
^Cre/+ mice

The LABORAS home cage observation system (Metris B.V.) consists of an adapted home cage placed on a carbon fibre platform that automatically detects behaviour-specific vibration patterns produced by the animal⁷⁵. The LABORAS software (v.2.6.) processes the vibrations into various validated behaviours (climbing, grooming, locomotion immobility) and tracking information (distance travelled and speed). These behavioural parameters are automatically calculated as time duration or frequency counts. C57BL/6 Cpa3^+/+ and Cpa3^Cre/+ mice were individually placed in these calibrated cages under standard housing conditions with free access to food and water for a 24-h period. Mice were not habituated to the LABORAS cages before the experiments were started.

Antibodies

The following antibodies were used in flow cytometry: B220 FITC 1:50 (BD Pharmingen, RA3-6B2), CD3 BV421 1:200 (17A2, BioLegend), CD3 FITC 1:50 (17A2, BD Pharmingen), CD3 PE-Cy-7 1:25 (145-2C11, BD Pharmingen), CD11b PerCP-Cy5.5 1:400 (M1/70, eBioscience), CD11b BV421 1:400 (M1/70, BioLegend), CD11b PE-Cy-7 1:400 (M1/70, eBioscience), CD11c BV421 1:100 (N418, BioLegend), CD16/32 unconjugated 10 μg ml⁻¹ (93, BioLegend), CD19 BV421 (6D5, BioLegend), CD19 APC 1:400 (1D3, BD Pharmingen), CD45 BV421 1:400 (30-F11, BioLegend), CD45 BV785 1:400 (30-F11, BioLegend), CD49b APC 1:100 (DX5, BD Pharmingen), CD90.2 APC-Cy7 1:400 (30-H12, BioLegend), CD117 PE 1:800 (2B8, eBioscience), CD117 APC 1:800 (2B8, BD Pharmingen), CD117 BV711 1:800 (2B8, BioLegend), FcεRI APC 1:200 (MAR-1, eBioscience), Gr-1 BV421 1:800 (RB6-8C5, BioLegend), Gr-1 BV605 1:200 (RB6-8C5, BioLegend), IgE PE 1:100 (RME1, BioLegend), IgE BV786 1:100 (RME-1, BD Pharmingen), IgE BV421 1:100 (RME-1, BD Pharmingen), Ly6G PerCP-Cy5.5 1:100 (1A8, BD Pharmingen), MHCII A700 1:100 (M5/114.15.2, eBioscience), Siglec-F BV421 1:100 (E50-2440, BD Pharmingen), Siglec-F PE 1:100 (E50-2440, BD Pharmingen), Ter119 BV421 1:200 (Ter119, BioLegend), 5-HT unconjugated 0.11 μg ml⁻¹ (5HT-H209, Dako) and mouse-IgG1 PE 1:100 (RMG1-1, BioLegend).

Schematics

Schematics shown in Figs. 4a and 5 and Extended Data Fig. 1a,e,j were made in Adobe Illustrator (v.25.0.1) using BioRender with permission to publish. The photograph of the IntelliCage was supplied by TSE Systems with permission to publish (Fig. 1b).

Reporting summary

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