Research ArticleFIBROSIS

Interleukin-36γ–producing macrophages drive IL-17–mediated fibrosis

See allHide authors and affiliations

Science Immunology  11 Oct 2019:
Vol. 4, Issue 40, eaax4783
DOI: 10.1126/sciimmunol.aax4783
  • Fig. 1 Single-cell characterization of macrophages in fibrotic and regenerative microenvironments.

    (A) Experimental overview. A virtual aggregate of macrophages in fibrosis and regeneration generated from scRNAseq after sorting of F4/80hi+CD64+. Cells were isolated from murine volumetric muscle injuries at 1 week, treatment with biomaterials UBM (regenerative, IL-4 tissue environment), synthetic (fibrotic, IL-17–rich environment), or saline (wound control). (B) Heatmap of differentially expressed genes. Up to 200 cells per cluster are shown, ordered by cluster, with the top 10 differentially expressed genes. Functionally relevant genes from terminal clusters are annotated. (C) Dimensional reduction projection of cells onto two dimensions using UMAP. Cells are colored by experimental biomaterial condition (top) and computationally determined cluster (bottom). (D) Summary of cluster differentiation trajectories, composition by experimental origin, markers, and biological functions generated by bioinformatics analysis. (E) Flow cytometry strategy informed by computationally determined markers including CD9, CD301b, and MHCII differentiating in vivo macrophage subsets from UBM or synthetic biomaterial. Subsets are colored equivalent to computational clusters, back-gated into tSNE projection.

  • Fig. 2 scRNAseq reveals surface markers that discriminate diverse regenerative and fibrotic macrophage subsets.

    (A) Histogram of CD86 and CD206 expression of bulk macrophages from regenerative, fibrotic, and saline microenvironments (green, red, and blue) by flow cytometry. (B) Feature plots of scRNAseq expression levels of Cd86 (bottom) and Cd206 (Mrc1, top) expression superimposed on UMAP plots of cells from scRNAseq. Circled borders mark regions enriched for cells from the regenerative or fibrotic experimental condition. (C) Cluster expression of canonical M1 (red) and M2 (green) markers. Expression levels are given as cluster averages normalized to the maximum value per gene. (D) Gene expression (UMI count) of M1 and M2 markers per single cell. Cells are ordered and colored by condition (regenerative as green, saline as blue, and fibrotic as red). Violin plots of cluster gene expression of surface markers identified by differential expression analysis. (E) In vivo flow cytometry strategy using CD9 and CD301b differentiates fibrotic (F1 and F2) and regenerative (R1 and R2) macrophage subsets in vivo. A.U., arbitrary units. (F) Mean fluorescence values indicate subset-specific expression of activation markers CD86 and CD206 (n = 4, **P < 0.005, ****P < 0.0001). (G) Comparison of scRNAseq (SC) and NanoString (NS) expression profiles after FACS of macrophage subsets. Genes are shown as either differentially expressed in both methods with agreement in fold change direction (SC and NS), found differentially expressed in one dataset (SC or NS), not found in either dataset (Not found), or found in both with disagreement in direction of fold change (Disagreement).

  • Fig. 3 Regenerative-associated clusters reveal distinct macrophage phenotypes including inflammatory R1 and phagocytotic R2.

    (A) Predicted lineage schematic of regenerative subsets from Slingshot pseudotime analysis and subset surface markers. (B) Slingshot pseudotime trajectory of regenerative-associated clusters shown on a principal component plot (PC1 versus PC2). Cells are colored by cluster. (C) Heatmap of top 20 differentially expressed genes in a comparison of R1 and R2. nES, normalized enrichment score. (D) Violin plots for differentially expressed genes comparing R1 and R2. (E) Gene set enrichment comparing R1 and R2. Plots with higher peaks (red) indicate enrichment of gene sets in R2, whereas plots with negative peaks (blue) indicate enrichment of gene sets in R1. (F) Gene network plots of R1 and R2 generated. Nodes represent genes with connections generated by STRING metadata analysis. Sets of genes associated with specific functions are annotated. (G) Flow cytometry gating scheme validates in vivo protein marker combination for R1 and R2. Macrophages defined as F4/80hi+ from live, CD45+. (H) One- to 6-week time course of R1 and R2 subsets in UCM, PCL, and saline microenvironments (n = 4, biologically independent). Two-way analysis of variance (ANOVA) with subsequent multiple testing P values is presented. ****P < 0.0001. n.s., not significant. (I) Immunofluorescence histology of defect region for CD301b (red) and F4/80 (green) at 1 week post-VML with UBM (scale bars, 25 μm; arrows indicate colocalization).

  • Fig. 4 Fibrosis-associated macrophages include distinct subsets F1 (MHCIIhi+) and F2 (CD9hi+IL-36γ+).

    (A) Lineage schematic of fibrotic subsets from Slingshot pseudotime analysis and trajectory including descriptive marker combination. Pseudotime trajectory is shown in a principal component plot (PC1 versus PC2). (B) Fibrotic subsets are distinguished by specific marker. (C) Heatmap of gene set enrichment scores normalized across clusters for gene sets found up-regulated in F1 and running gene set enrichment plots for the IFNγ and IFNα responses. (D) Gene network representation for relationships of differentially expressed genes in F1 (top) and F2 (bottom) by STRING metadata scores. (E) Flow cytometry gating strategy specific to F1 and (F2 + FP1) from F4/80hi+ macrophages using CD9, MHCII, and CD301b. (F) Time course of F1 and (F2 + FP1) subsets in UBM, PCL, and saline microenvironments (n = 4, biologically independent). Two-way ANOVA P values are presented. *P < 0.05; ***P < 0.0001. (G) Immunofluorescence histology for IL-36γ (violet) and F4/80 (green) at 1-week VML with synthetic PCL material. Low power indicates location of PCL material and muscle defect; arrows detail colocalization (scale bars, 500 and 50 μm).

  • Fig. 5 Profibrotic CD9hi+IL-36γ+ macrophages are dependent on IL-17 signaling, and terminal clusters are relevant in human pathologies.

    (A) Immunofluorescence staining for mouse macrophage marker F4/80 and CD9 in WT, Il17ra−/−, and Il17a−/− mice 12 weeks after implantation with PCL (scale bars, 50 μm). (B) Il36γ gene expression in WT, Il17ra−/−, and Il17a−/− mice with PCL normalized to saline controls (n = 4, biologically independent; ANOVA with multiple comparison, ***P < 0.001). (C) Time course of Il36γ signaling and fibrosis-related gene expression over 6 weeks, normalized to saline controls (n = 4, biologically independent; ANOVA with multiple comparison, ***P < 0.001). (D) Immunofluorescent staining for CD64-, CD9-, and IL-36γ–positive macrophages in human breast implant tissue capsules (scale bars, 50 μm), juvenile xanthogranuloma, and Langerhans cell histiocytosis (scale bars, 200 μm). (E) Gene expression correlations of human IL17RA with IL36γ, IL17RA, and CD9 with MSR1 in human breast implant fibrotic capsules. (F) Network diagrams and similarity heatmaps for terminal fibrotic and regenerative macrophage clusters to clusters from repository single-cell RNA datasets for murine models of cancer (sarcoma ± immunotherapies aCTLA-4, aPD-1), lung fibrosis (± bleomycin induction), and human liver. Circles represent percent compositions of clusters by condition.

Supplementary Materials

  • immunology.sciencemag.org/cgi/content/full/4/40/eaax4783/DC1

    Fig. S1. Quality control of scRNAseq data.

    Fig. S2. Differential expression and quality control metrics show clusters grouped on nondesirable traits.

    Fig. S3. Precursor clusters show similarities across condition.

    Fig. S4. scRNAseq reveals that canonical M1 and M2 markers do not correlate with single-cell clusters.

    Fig. S5. Macrophage gating scheme for regenerative subsets.

    Fig. S6. Macrophage gating scheme for fibrotic subsets.

    Fig. S7. Murine macrophage CD9 and CD301b profiles using flow cytometry.

    Fig. S8. Differential gene expression heatmap for RP3 and R2.

    Fig. S9. Cell composition of clusters by experimental condition.

    Fig. S10. Expression of inflammatory genes in fibrotic cluster F2.

    Fig. S11. Quantification of expression levels of CD11c and CD206.

    Fig. S12. RNA velocity analysis.

    Fig. S13. Regenerative subsets R1 and R2 phagocytosis assay.

    Fig. S14. Immunofluorescence microscopy of F4/80+CD301b+ macrophages.

    Fig. S15. Cluster R3 expresses tissue-specific genes.

    Fig. S16. Expression of inflammatory genes in fibrotic cluster F2.

    Fig. S17. Immunofluorescence microscopy on F4/80+CD9+ macrophages.

    Fig. S18. In vitro TH17 dependence of Il36γ expression in macrophages and IL-36γ induction of fibrosis phenotypes in cultured fibroblasts.

    Fig. S19. Analysis pipeline for publicly available datasets.

    Fig. S20. Scheme to generate for tSNE projection for virtual aggregate flow cytometry data.

    Fig. S21. Scaling metric for the scRNAseq raw data.

    Fig. S22. Variance by principle component.

    Fig. S23. Expression of fibroblast signature genes in small contaminant cluster.

    Table S1. Quality control metrics for 10× scRNAseq.

    Table S2. Macrophage FACS panel.

    Table S3. Macrophage subtype panel.

    Table S4. Murine TaqMan gene expression assay probes.

    Table S5. Human TaqMan gene expression assay probes.

    Table S6. Sorted cell populations for NanoString Gene Expression Profiling.

    Table S7. Differentially expressed genes for scRNAseq clusters.

  • Supplementary Materials

    The PDF file includes:

    • Fig. S1. Quality control of scRNAseq data.
    • Fig. S2. Differential expression and quality control metrics show clusters grouped on nondesirable traits.
    • Fig. S3. Precursor clusters show similarities across condition.
    • Fig. S4. scRNAseq reveals that canonical M1 and M2 markers do not correlate with single-cell clusters.
    • Fig. S5. Macrophage gating scheme for regenerative subsets.
    • Fig. S6. Macrophage gating scheme for fibrotic subsets.
    • Fig. S7. Murine macrophage CD9 and CD301b profiles using flow cytometry.
    • Fig. S8. Differential gene expression heatmap for RP3 and R2.
    • Fig. S9. Cell composition of clusters by experimental condition.
    • Fig. S10. Expression of inflammatory genes in fibrotic cluster F2.
    • Fig. S11. Quantification of expression levels of CD11c and CD206.
    • Fig. S12. RNA velocity analysis.
    • Fig. S13. Regenerative subsets R1 and R2 phagocytosis assay.
    • Fig. S14. Immunofluorescence microscopy of F4/80+CD301b+ macrophages.
    • Fig. S15. Cluster R3 expresses tissue-specific genes.
    • Fig. S16. Expression of inflammatory genes in fibrotic cluster F2.
    • Fig. S17. Immunofluorescence microscopy on F4/80+CD9+ macrophages.
    • Fig. S18. In vitro TH17 dependence of Il36γ expression in macrophages and IL-36γ induction of fibrosis phenotypes in cultured fibroblasts.
    • Fig. S19. Analysis pipeline for publicly available datasets.
    • Fig. S20. Scheme to generate for tSNE projection for virtual aggregate flow cytometry data.
    • Fig. S21. Scaling metric for the scRNAseq raw data.
    • Fig. S22. Variance by principle component.
    • Fig. S23. Expression of fibroblast signature genes in small contaminant cluster.
    • Table S1. Quality control metrics for 10× scRNAseq.
    • Table S2. Macrophage FACS panel.
    • Table S3. Macrophage subtype panel.
    • Table S4. Murine TaqMan gene expression assay probes.
    • Table S5. Human TaqMan gene expression assay probes.
    • Table S6. Sorted cell populations for NanoString Gene Expression Profiling.
    • Legend for table S7

    Download PDF

    Other Supplementary Material for this manuscript includes the following:

    • Table S7 (.csv format). Differentially expressed genes for scRNAseq clusters.

    Files in this Data Supplement:

Stay Connected to Science Immunology

Navigate This Article