Mice

The following mice on the C57BL/6 background were used in this study: CD45.2 wild-type C57BL/6 (The Jackson Laboratory), CD45.1 wild-type C57BL/6J (The Jackson Laboratory, 002014), Cx3cr1GFP/+ (ref. 45; The Jackson Laboratory, 005582), Tmem119Cre (see below), Rosa26LSL-EYFP (ref. 23; The Jackson Laboratory, 006148), Cx3cr1LSL-DTR/+ (ref. 25; The Jackson Laboratory, 025629), Ccr2−/− (ref.46; The Jackson Laboratory, 004999), Nr4a1−/−(ref.47 ;The Jackson Laboratory, 006187), Maffl/fl (ref. 48; kindly provided by F. Andris), Mafbfl/fl (generated by D.P. and the GIGA Mouse facility and Transgenics Platform, Liège University, Belgium, see below), Lyz2Cre (ref. 49; The Jackson Laboratory, 004781) and Ms4a3Cre (ref. 9). Myeloid-restriced Maf or Mafb depletion was achieved by crossing Maffl/fl or Mafbfl/fl mice with Lyz2Cre or Ms4a3Cre mice.

C57BL/6 Tmem119Cre knock-in mice were generated using CRISPR/Cas-mediated genome engineering by Cyagen Bioscience. In brief, the Tmem119 targeting vector was designed by cloning a genomic fragment encompassing exon 2 of the Tmem119 gene from BAC clones RP23-187D5 and RP23-126P3. A Cre-polyA cassette was introduced in the Tmem119 targeting vector upstream of the ATG start codon between a 2.1-kb 5′ homology arm and a 2.1-kb 3′ homology arm. Tmem119-gRNA (protospacer, CAGGGGACCATGTTGAGCTATGG), Cas9 mRNA and Tmem119 targeting vector were co-injected into pronuclei of C57BL/6J one-cell-stage zygotes, followed by implantation of the zygotes into surrogate mothers to obtain targeted knock-in offspring. F0 knock-in founder animals were identified by PCR followed by sequence analysis. Tmem119Cre/+ mice were then back-crossed to CD45.2 or CD45.1 C57BL/6J mice for at least four generations. Tmem119Cre mice were genotyped by PCR using the following primers: PCR primers 1 for mutant allele (annealing temperature, 60.0 °C): forward primer: 5′- TCCGTAACCTGGATAGTGAAACAG-3′; reverse primer: 5′-ATATGTCCTTCCGAGTGAGAGAC-3′; product size: 270 bp (mutant). PCR primers 2 for wild-type allele (annealing temperature, 60.0 °C): forward primer: 5′-ACCGAGGACAGAAATGAATAAGATG-3′; reverse primer: 5′-AGGGAACGAGGATGGGTAGTAG-3′; product size: 643 bp (wild type).

C57BL/6 Mafbfl/fl mice were generated using recombination-mediated genetic engineering. Briefly, the genomic segment covering the Mafb single exon was retrieved to PL253 vector using BAC recombineering. The loxP-EM7-Neo-loxP cassette was cloned by PCR from PL452 plasmid and ligated to the Mafb 5′ segment (PL253/Mafb/Neo 5′) and the cassette was ‘popped out’ by electroporating to SW106 cells expressing Cre and 5′ loxP left in the construct. The FRT-Neo-FRT-loxP cassette was cloned from PL451 plasmid and ligated to the Mafb 3′ segment. The purified plasmid was electroporated into mouse embryonic stem cells and the cells were selected under G418 treatment for 1 week. The bona fide clones with successful homologous recombination were screened by Southern blot. Successfully recombined clones were injected into blastocysts to make Mafbfl-Neo mice. These mice were crossed to an FLP-expressing line to remove the Pgk-Neo cassette and generate Mafbfl mice. Mafbfl mice were genotyped by PCR using the following primers: forward primer: 5′- TCCATCCATCTTGGGAAAAG-3′; reverse primer: 5′-TCAGGACTGGGCTGCTAGTT-3′; product size: 320 bp (Mutant), 220 bp (wild type).

Tmem119Cre and Rosa26LSL-EYFP mice were crossed to create Tmem119CreRosa26LSL-EYFP mice. Tmem119Cre and Cx3cr1LSL-DTR/+ mice were crossed to create Tmem119CreCx3cr1LSL-DTR/+ mice, referred to as ‘IMDTR’ mice. Since we observed some YFP labeling in CD45− cells in the testis and ovaries of Tmem119CreRosa26LSL-EYFP mice, we did not use Tmem119CreRosa26LSL-EYFP or Tmem119CreCx3cr1LSL-DTR mice as breeders to avoid any issues arising from germline recombination. CD45.1/CD45.2 IMDTR mice were generated by crossing CD45.1 Tmem119Cre with CD45.2 Cx3cr1LSL-DTR mice.

No sex-specific differences were observed in pilot experiments. A mix of male and female mice between 6 and 10 weeks of age were used for each experiment, except for chimera experiments where mice between 11 and 15 weeks of age were used. The mice were bred and housed under specific-pathogen-free conditions at the GIGA Institute (Liège University, Belgium), maintained in a 12-h light-dark cycle, and had access to normal diet chow and water ad libitum. Mice were identified according to genotype and all experiments were performed with age-matched and sex-matched littermates. For Csf1r-blocking experiments, mice were randomly assigned to vehicle or isotype antibodies and anti-Csf1r treatments. For experiments using IMDTR mice that were treated or not with DT, mice were randomly allocated to DT treatment or not. Investigators were not blinded during the collection and analysis of the data, except for the quantification of microscopy lung sections, where investigators were blinded.

All animal experiments described in this study were reviewed and approved by the Institutional Animal Care and Use Committee of the University of Liège (ethical approval no. DE1956). The ‘Guide for the Care and Use of Laboratory Animals,’ prepared by the Institute of Laboratory Animal Resources, National Research Council, and published by the National Academy Press, as well as European and local legislations, was followed carefully. Accordingly, the temperature and relative humidity were 21 °C and 45–60%, respectively.

Reagents and antibodies

A complete list of the reagents, antibodies and commercial assays used in this paper can be found in Supplementary Table 1.

In vivo treatments with chemicals and antibodies

For DT-induced depletion of IM, IMDTR mice were injected i.p. with a single dose of 50 ng DT (List Biological Labs, 150), unless otherwise stated. Control mice were either untreated IMDTR mice, or Tmem119Cre/+ littermate control mice injected with DT. For EdU incorporation experiments, IMDTR mice were injected i.p. with 1 mg EdU (Santa Cruz Biotechnology, sc-284628) in 200 µl PBS 4 h before killing, unless otherwise stated. For all experiments involving EdU incorporation, 1 µg of PerCP-Cy5.5-conjugated anti-mouse CD45 (clone 104, BD Biosciences, 552950) was i.v. injected 10 min before killing to distinguish blood circulating (CD45-PerCP-Cy5.5+) and tissue leukocytes (CD45-PerCP-Cy5.5−). For Csf1r-blocking experiments, 250 µg of anti-mouse Csf1r-blocking antibody (clone AFS98, Bio X Cell, BE0213) or isotype control (clone 2A3, Bio X Cell, BE0089) was injected i.v. 6 and 28 h after DT injection. For experiments with Csf1r inhibitors, 100 mg per kg body weight of pexidartinib (PLX3397; MedChemExpress, HY-16749) was injected i.p. 24 and 48 h after DT injection.

Bone marrow, blood and tissue leukocyte isolation

Blood was collected by retro-orbital plexus bleeding of terminally anesthetized mice. Mice were then euthanized by cervical dislocation. Peritoneal lavage was obtained by injecting 10 ml HBSS (Lonza, BE10-508F) into the peritoneal cavity and collecting the washout. Mice were then perfused with 10 ml PBS via the left ventricle, and lungs, brain, liver, spleen, intestine and colon were dissected.

For BM cells, femurs and tibias were dissected and cleaned of soft adhering tissue. Distal and proximal ends were opened, and BM cells were flushed out. After centrifugation, cell pellets were resuspended in ice-cold PBS (Thermo Fisher, 14190094) containing 10 mM EDTA (Merck Millipore, 1084181000) and cell suspensions were filtered using a cell strainer (70 µm, Corning, 352350) to obtain a single-cell suspension.

Lungs, brains, liver and spleen were cut into small pieces with razor blades, and digested for 1 h at 37 °C in HBSS containing 5% vol/vol FBS (Thermo Fisher, 10270098), 1 mg ml−1 collagenase A (Sigma, 14190094) and 0.05 mg ml−1 DNase I (Sigma, 11284932001). After 45 min of digestion, the suspension was flushed using a 18-gauge needle to dissociate aggregates. Ice-cold PBS (Thermo Fisher, 14190094) containing 10 mM EDTA (Merck Millipore, 1084181000) was added to stop the digestion process and cell suspensions were filtered using a cell strainer (70 µm, Corning, 352350). Mononuclear leukocytes from lungs and livers were enriched using a Percoll density gradient (GE Healthcare, 17089101) and by harvesting cells from the 1.080:1.038 g ml−1 interface.

For the isolation of leukocytes from the small intestines and colons, small intestines and colons were dissected from the pylorus and the rectum, were separated from the mesenteric tissue from Peyer’s patches and from fat and were placed in ice-cold HBSS with 2% FBS. Intestinal content was removed with PBS, and the small intestines and colons were opened by a longitudinal cut and washed three times in ice-cold HBSS with 2% FBS. To remove mucus and epithelial cells, small intestines and colons were incubated with HBSS with 2% FBS and 1 mM 1,4 dithiothreitol (Sigma, 10197777001) for 20 min with constant shaking followed by an incubation with HBSS containing 2% FBS and 1.3 mM EDTA for 40 min. Tissue pieces were then cut into small pieces and incubated for 1 h at 37 °C with RPMI containing 2% FBS, 2 mg ml−1 collagenase IV (Thermo Fisher, 17104019) and 40 U ml−1 DNase I. At the end of incubation, the suspension was homogenized with a 19-gauge syringe and filtered through a 70-µm strainer.

Generation of bone marrow (competitive) chimeras

Eighteen-week-old CD45.2 or CD45.1/CD45.2 IMDTR mice were anesthetized by i.p. injection of 200 µl PBS containing ketamine (75 mg per kg body weight; Dechra, 804132) and xylazine (10 mg per kg body weight; Bayer, 0076901). The thoracic cavity was protected with a 0.6-cm-thick lead cover and mice were lethally irradiated with two doses of 6 Gy 15 min apart. Once recovered from the anesthesia, mice were reconstituted by i.v. administration of 107 BM cells from congenic CD45.1 wild-type mice. For mixed BM chimeras, mice were injected i.v. with 107 BM cells consisting of a 1:1 mix of cells obtained from CD45.1 wild-type and CD45.2 Ccr2−/−, Nr4a1−/− or Ms4a3CreMafbfl/fl mice. From the day of irradiation, mice were treated for 4 weeks with 0.05 mg ml−1 of enrofloxacin (Baytril, Bayer) in drinking water. Chimerism was assessed by flow cytometry in the blood 4 weeks after irradiation.

Adoptive transfer of bone marrow monocytes

BM Ly6C+ monocytes were isolated from congenic CD45.1 wild-type mice using the Monocyte Isolation Kit (Miltenyi Biotec, 130-100-629). Around 2 × 106 BM Ly6C+ monocytes were administered i.v. into CD45.1/CD45.2 IMDTR mice that were injected i.p. with 50 ng DT 24 h before monocyte transfer to deplete endogenous IMs.

Flow cytometry

Cells (0.5–5 × 106) were pre-incubated with Mouse BD Fc Block (BD Biosciences, 553142) to avoid unspecific binding to Fc receptors and stained with appropriate antibodies at 4 °C in the dark for 30 min. For EdU staining, extracellular-stained cells were permeabilized and stained using Click-iT EdU Alexa Fluor 488 Flow Cytometry Assay Kit (Thermo Fisher, 10632), according to the manufacturer’s instructions. For DAPI cell cycle analyses, extracellular-stained cells were permeabilized and stained with 1 µg ml−1 DAPI (BioLegend, 422801) in the dark for 30 min at room temperature (RT). For Ki67 stainings, extracellular-stained cells were permeabilized and stained using either FITC Mouse Anti-Ki67 Set (BD Biosciences, 556026) or PerCP-eFluor710 Mouse Anti-Ki67 (Thermo Fisher, 46-5698-80). Cell viability was assessed using LIVE/DEAD Fixable Near-IR (775) stain (Thermo Fisher, L34976) and the cell suspensions were analyzed with an LSRFortessa (BD Biosciences). Results were analyzed using FlowJo software (Tree Star). For scRNA-seq and bulk RNA-seq, lung myeloid cells were sorted using a FACSAria III (BD Biosciences). The full list of antibodies used can be found in the Supplementary Table 1.

MCP-1/Ccl2 quantification

IMDTR and littermate control mice were euthanized at indicated time points after DT administration. Blood was collected and lungs were perfused through the right ventricle with 10 ml PBS and isolated. Blood samples were left undisturbed for 30–45 min at RT to allow clot formation. The serum was separated from the blood clot by centrifugation for 10 min at 2,000g at 4 °C. Serum was stored at −80 °C. Dissected lungs were snap frozen and homogenized in 360 µl ice-cold lysis buffer (40 mM Tris-HCl (pH 7.4), 150 mM NaCl, 10% glycerol and cOmplete Protease Inhibitor Cocktail (Sigma, 11697498001) using a tissue homogenizer (IKA) with the addition of 1% NP-40 (Sigma, 74385) after homogenization. Samples were then rotated for 20 min at 4 °C, followed by a centrifugation to pellet debris. Protein concentration of cleared lysates was determined using Pierce BCA Protein Assay Kit (Thermo Fisher), according to the manufacturer’s instructions. Cleared lysates were stored at −80 °C. Ccl2 levels in serum and lung homogenates were determined using MCP-1/Ccl2 Mouse Uncoated ELISA Kit (Thermo Fisher), according to the manufacturer’s instructions.

Bulk RNA-seq: sample preparation and analysis

Native IM subsets, cMo and AMs were isolated from uninjected IMDTR mice, while repopulated IM subsets were isolated from IMDTR mice that had been treated i.p. with 50 ng DT 14 d earlier. Cell populations were FACS sorted using the gating strategy shown in Fig. 1c into TRIzol reagent (Thermo Fisher, 10296010). Total RNA was extracted with the standard TRIzol RNA extraction protocol. RNA quality and quantity were evaluated using a 2100 bioanalyzer (Agilent) and the Quant-iT RiboGreen RNA Assay Kit (Thermo Fisher, R11490). One hundred nanograms of RNA was used to generate the libraries using the TruSeq Stranded mRNA kit (Illumina, 20020594). These libraries were sequenced on an Illumina NovaSeq sequencer on an SP flow cell. Sequence alignment with the mouse genome (GRCm38), sequence counting and quality control were performed using the nf-core/rnaseq pipeline. RNA-seq data were analyzed using R Bioconductor (3.5.1) and DESeq2 package (version 1.26.0)50.

scRNA-seq

To compare lung monocytes and IMs from untreated IMDTR mice (group ‘no treatment’) with those from IMDTR mice treated with 50 ng DT i.p. 96 h before (group ‘DT96h’), five mice from each group were killed and lung single-cell suspensions were obtained after enzymatic digestion. CD11b+ cells were enriched by MACS using CD11b MicroBeads (Miltenyi Biotec, 130-049-601). Lung monocytes and IMs were then FACS sorted separately as CD45+SSCloCD11b+F4/80+ CD64− and CD64+ cells, respectively (Fig. 1c), and the 10x Genomics platform (Single Cell 3′ Solution) was used for scRNA-seq. The IM pool was then enriched in the final single-cell suspension to reach a monocyte/IM ratio of 1:1. For each sample, an aliquot of Trypan blue-treated cells was examined under the microscope for counting, viability and aggregate assessment following FACS sorting. Viability was above 90% for all samples and no aggregates were observed. Cell preparations were resuspended in calcium-free and magnesium-free PBS containing 0.4 mg ml−1 of UltraPure BSA (Thermo Fisher Scientific, AM2616).

To analyze lung monocytes (CD45+SSCloCD11b+F4/80+CD64−) and IMs (CD45+SSCloCD11b+F4/80+CD64+) from IMDTR mice treated 12 h (group ‘DT12h’), 24 h (group ‘DT24’) and 48 h (group ‘DT48h’) before with 50 ng DT i.p., and to analyze lung monocytes (CD45+SSCloCD11b+F4/80+CD64−) and CD64+ cells (CD45+SSCloCD11b+F4/80+CD64+) from Lyz2CreMafbfl/f/l (group ‘Mafb-KO’), Lyz2CreMaffl/fl (group ‘cMAF-KO’) and littermate control (group ‘control’) mice, a similar protocol was applied, but cells from each group were barcoded with different anti-mouse Hashtag antibodies (BioLegend) before being pooled for encapsulation and library construction. To obtain a higher resolution in analyzing lung myeloid cells in myeloid-restricted Mafb-deficient and Maf-deficient mice, the pooled Mafb-KO/cMAF-KO/control samples were composed of a ratio of monocytes:CD64+ cells of 3:7 instead of 1:1.

For library preparation, approximately 3,000 cells per sample (for ‘DT96h’ and ‘no treatment’), or 20,000 cells for pooled hashtag-labeled samples, were loaded into the Chromium Controller, in which they were partitioned, and their polyA RNAs captured and barcoded using Chromium Single Cell 3′ GEM, Library & Gel Bead Kit v3 (10x Genomics). The cDNAs were amplified and libraries compatible with Illumina sequencers were generated using Chromium Single Cell 3′ GEM, Library & Gel Bead Kit v3 (10x Genomics). For Hash Tag Oligonucleotide (HTO) library, 1 µl HTO additive primer v2 (0.2 µM stock) were added to the mix at the cDNA amplification step. The libraries were sequenced on an Illumina NovaSeq sequencer on an SP100 cell flow (read 1, 28 cy; read 2, 76 cy; index 1, 10 cy; index 2, 10 cy) at a depth of 50,000 reads per cell.

The Cell Ranger (v3.0.2) application (10x Genomics) was then used to demultiplex the BCL files into FASTQ files (cellranger mkfastq), to perform alignment (to Cell Ranger human genome references 3.0.2 GRCm38/build 97), filtering and unique molecular identifier counting and to produce gene-barcode matrices (cellranger count).

Filtered matrix files were used for further scRNA-seq analyses with R Bioconductor (3.12) and Seurat (3.2.1)51. The cells from pooled hashtag-labeled samples were demultiplexed with the barcode detected in each cell.

Filtered matrices containing cell IDs and feature names in each sample were used to build a Seurat object. We performed quality control by filtering out the cells with less than 200 detected genes, the genes detected in less than three cells and the cells exhibiting more than 10% of mitochondrial genes. Gene counts in each sample were normalized separately by default method ‘LogNormalize’ with a scale factor of 10,000 and log transformation. Two thousand highly variable features were identified with the ‘vst’ method.

After merging cells from all samples, cell contaminants were removed based on the expression of specific genes. Four clusters were identified in the remaining cells using the FindClusters function and the DEGs were calculated using the FindAllMarkers function (Seurat package).

Single-cell RNA velocity estimation

The counts for unspliced and ambiguous transcripts were calculated from CellRanger output using the velocyto command-line tool (http://velocyto.org/)52 and saved in loom files. The single-cell RNA velocities were estimated using scVelo toolkit (https://scvelo.readthedocs.io/)53. Briefly, the loom files were used as input for scVelo analysis. Genes with a minimum of 20 of both unspliced and spliced counts and on the top list of 2,000 genes were filtered, normalized and log transformed (scv.pp.filter_and_normalize with default parameters). Thirty principal components and 30 neighbors obtained from Euclidean distances in principal-component analysis space were used for computing first-order and second-order moments for each cell. We used generalized dynamical modeling to recover the full splicing kinetics of spliced genes, and the single-cell RNA velocities were plotted with the same cluster labels and embedding as in Fig. 4a.

Gene Ontology enrichment analysis with differentially expressed gene signatures

The DEG lists for enrichment analyses were calculated using Seurat function FindMarkers with only.pos = TRUE to output only positively regulated genes. Thresholds logfc.threshold of 0.2 and adjusted P value of 0.01 were applied to filter the gene lists. Gene Ontology (GO) enrichment analyses were made using enrichGO functions from clusterProfiler package54 with default arguments. Only biology process terms of ontology were shown in the final results.

Immunofluorescence

For lung immunofluorescence stainings, lungs were perfused with 10 ml PBS via the left ventricle and lungs were collected. Lungs were fixed for 4 h in 4% paraformaldehyde (Thermo Fisher, F/1501/PB15) at 4 °C. Fixed lungs were then cryoprotected in 30% sucrose (VWR, Avantor, 57-50-1) in PBS for 4 h at 4 °C, followed by embedding in optimal cutting temperature compound (OCT; Tissue-Tek, 4583) at −80 °C overnight, and lung OCT sections were cut (7-µm-thick sections) and blocked in methanol 100% (Merck, 67-56-1) at −20 °C for 20 min. Samples were stained in blocking buffer (PBS with 0.3% Triton X-100 (Merck, 648466) and 2% donkey serum (Sigma Aldrich, D9663)) with rat anti-mouse antibodies directed against MHC class II (I-A/I-E; 1:100 dilution in blocking buffer; clone M5/114.15.2, eBioscience, 14-5321-82) overnight at 4 °C. After washing samples with PBS, secondary donkey anti-rat IgG antibodies conjugated with Alexa Fluor 594 (1:1,000 dilution in blocking buffer; Thermo Fisher, A21209) were added in blocking buffer and incubated for 1 h in the dark at RT. Samples were washed with PBS and incubated with Alexa Fluor 488-conjugated rat anti-mouse antibodies directed against CD206 (clone C068C2, BioLegend, 141710; 1:50 dilution in blocking buffer), eFluor 570-conjugated rat anti-mouse antibodies directed against Ki67 (1:200 dilution in blocking buffer; clone SolA15, eBioscience, 41-5698-82), APC-conjugated rat anti-mouse antibodies directed against CD11b (1:50 dilution in blocking buffer; clone M1/70, eBioscience, 17-0112-83) in blocking buffer for 6 h at 4 °C. Finally, samples were washed one last time with PBS and were mounted with 10 μl ProLong Antifade reagent (Invitrogen, P36961) containing 0.1% Sytox blue nucleic acid stain (Invitrogen, S11348) on glass slides and stored at RT in the dark overnight.

All samples were analyzed by spectral fluorescence microscopy. Images of full lung sections were acquired on an LSM 980 with Airyscan 2 inverted confocal microscope (Zeiss) using a LD C-Apochromat ×40/1.1 W objective and Zen Black software. All fluorophores were excited simultaneously, and the emission spectra were collected with a spectral detector 32-channel GaAsP photomultiplier tube in lambda mode at 8.8-nm bins from 411 to 694 nm. A spectral unmixing was performed based on the monospectral spectra. Images were processed with the Zen Blue software. For quantification, the numbers of CD11b+CD206loMHC-IIhi IMs (CD206− IMs), CD11b+CD206hiMHC-IIlo/int IMs (CD206+ IM) and CD11b+CD206−MHC-IIloKi67hi cells were counted blindly and manually on a total surface of 12–16 mm2 per mouse section. The results were expressed in cell number per mm2 of lung section.

Single-cell regulatory network inference and clustering analysis

To predict the potential active transcription factors, Ly6C+ cMo, Tr-Mo, CD206− IMs and CD206+ IMs were subjected to SCENIC analysis using the SCENIC package33. The normalized counts, nFeature_RNA and nCount_RNA in merged Seurat object were used for the initial SCENIC analysis. The genes expressed with a value of 3 in 0.5% of the cells and detected in 1% of the cells were kept for following SCENIC analysis. Coexpression network analysis was made with GENIE3 in the SCENIC package. To represent the SCENIC results, the results of the ‘3.4_regulonAUC’ output were added to the metadata of Seurat object so that regulon AUC scores could be plotted using the FeaturePlot function. The top 50 regulons with highest variance are shown with their z-scores in the heat map.

Monocle, tradeSeq and pseudotime analysis during interstitial macrophage development

To evaluate trajectory-based differential expression analysis during IM development in IMDTR mice, Ly6C+ cMo, Tr-Mo, CD206− IMs and CD206+ IMs were subjected to Monocle28 analysis. The Monocle CDS object was built with counts and metadata from Seurat object and converted using SeuratWrappers package. Cells were clustered with the cluster_cells function using calculated UMAP coordinates and a resolution of 0.51 × 10−3. The trajectories along pseudotime were built using learn_graph and order_cells functions. The DEGs across trajectories were calculated using Moran’s I test (graph_test function) and only the genes with a q value of 0 and Morans’s I value over 0.25 were kept as significant DEGs and subjected to further analyses.

To compare the expression patterns of DEGs across pseudotime, the counts matrix, pseudotime and cell weights calculated above were then used as input in fitGAM function (tradeSeq package)29. The association of average expression of each gene with pseudotime was tested using associationTest and the DEGs between CD206− IM and CD206+ IM trajectories were calculated with the diffEndTest function. The value of the estimated smoother on a grid of pseudotimes was estimated for each DEG using predictSmooth. The DEG with waldStat > 70 and |log fold change| > 2 were annotated as ‘changed genes’, meaning that their expression patterns were different in CD206− and CD206+ IM trajectories, while the rest of the DEGs were considered as ‘unchanged genes’, meaning that the expression patterns were similar in both trajectories. Finally, the scaled estimated smoothers calculated by predictSmooth were used to build heat maps with the ComplexHeatmap package55.

Interstitial macrophage and monocyte signature scoring

The IM-specific, cMo-specific and CD16.2+ Mo-specific gene signatures were calculated using previously published scRNA-seq data20 by comparing IM, cMo or CD16.2+ Mo populations to all other cell types in the dataset using the FindMarker function (Seurat). The genes with |log fold change| > 1 and only positively regulated ones were considered as the IM, cMo or CD16.2+ Mo signature. The signatures were then used to calculate the scores for each cell using the VISION package56 (Fig. 8i) or with AddModuleScore function (Seurat; Extended Data Fig. 4e). The scores were stored in Seurat object and plotted with Seurat package.

Statistical analysis

Graphs were prepared with Prism 9 (GraphPad) or R Bioconductor (3.5.1)57, and ggplot2 for data in Fig. 3b. No statistical methods were used to predetermine sample sizes, but our sample sizes are similar to those reported in previous publications18,20,24,58. Data distribution was assumed to be normal when parametric tests were performed, but this was not formally tested. Data from independent experiments were pooled for analysis in each data panel, unless otherwise indicated. No data were excluded from the analyses. Statistical analyses were performed with Prism 9 (GraphPad), and with R Bioconductor (3.5.1)57 and DESeq2 (ref. 50) or Seurat (3.2.1.)51 for bulk and scRNA-seq data, respectively. The statistical analyses performed for each experiment are indicated in the respective figure legends. We considered a P value lower than 0.05 to be significant (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

Reporting summary

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