Macrophages in the synovial lining niche initiate neutrophil recruitment and articular inflammation

Abstract

The first immune-activating changes within joint resident cells that lead to pathogenic leukocyte recruitment during articular inflammation remain largely unknown. In this study, we employ state-of-the-art confocal microscopy and image analysis in a systemic, whole-organ, and quantitative way to present evidence that synovial inflammation begins with the activation of lining macrophages. We show that lining, but not sublining macrophages phagocytose immune complexes containing the model antigen. Using the antigen-induced arthritis (AIA) model, we demonstrate that on recognition of antigen-antibody complexes, lining macrophages undergo significant activation, which is dependent on interferon regulatory factor 5 (IRF5), and produce chemokines, most notably CXCL1. Consequently, at the onset of inflammation, neutrophils are preferentially recruited in the vicinity of antigen-laden macrophages in the synovial lining niche. As inflammation progresses, neutrophils disperse across the whole synovium and form swarms in synovial sublining during resolution. Our study alters the paradigm of lining macrophages as immunosuppressive cells to important instigators of synovial inflammation.

Figure 1.

Lining macrophages phagocytose ICs and model antigen-mBSA. (A) Quantification of PE fluorescence intensity by FACS in synovial resident macrophages 3 h after IA injections. Significance values by mixed-effects analysis with Šídák multiple comparisons post hoc: *, P < 0.05. GMOF, geometric mean of fluorescence. (B) Synovial neutrophil counts as a result of IC and PE challenge by FACS at 3 h after IA injections. Significance values by one-way ANOVA with Šídák multiple comparisons post hoc: *, P < 0.05; **, P < 0.01; data representative of two experiments. (C) Confocal fluorescence microscopy of the whole mouse knee in sagittal plane 4 h after injection of fluorescently labeled mBSA (large panel with white arrowheads pointing out lining and cartilage localization of the antigen, whereas red arrowhead indicates leak in the posterior part; pan-macrophage CD68, cyan; lining macrophage VSIG4, magenta; mBSA, yellow; nuclei, gray; colocalization of VSIG4 and mBSA, white; scale bar = 100 µm). (D) Zoomed region of the lining shows preferential uptake of mBSA by the lining macrophages and not the sublining ones (three right panels, scale bar = 20 µm). (E) Quantification of mBSA uptake by the lining and sublining macrophages (N = 2; n = 6) in a scatter plot; each symbol indicates one section, and lines connect macrophages of the same section. N indicates the number of mice, n the number of quantified sections; data representative of two independent experiments. Significance values by Wilcoxon matched-pairs signed rank test: *, P < 0.05. FP, fat pad; M, meniscus; PT, patellar tendon; SC, synovial cavity.

Figure S1.

Synovial lining macrophages can be identified and isolated by the expression of VSIG4. (A) Confocal image of the naive knee from CX3CR1-eGFP Ly6G-tdTomato double reporter mouse containing resident macrophages and no neutrophils; VISG4 and CX3CR1 colocalize in the lining (pan-macrophage CD68, cyan; lining macrophage VSIG4, magenta; CX3CR1, orange; nuclei, gray; sagittal plane; scale bar = 30 µm). (B) VISG4 marks lining macrophages by reanalysis of publicly available small bulk RNA sequencing dataset of murine paws in steady state (Culemann et al., 2019). (C) Gating strategy used to isolate lining and sublining macrophages by FACS for small bulk RNA sequencing; pseudo-color plots of contralateral knee at 4 h p.c. created in FlowJo v10. F, femur; FP, fat pad; SC, synovial cavity; T, tibia.

Figure S2.

Activated transcriptional profile in lining macrophages at the onset of AIA and CXCL1 expression in the lining. (A) GO analysis of differentially expressed genes in lining and sublining macrophages at 4 h after mBSA challenge (n = 3). FDR, false discovery rate. (B) GO analysis of differentially expressed genes in lining and sublining macrophages from PBS control knees (n = 3). (C) Confocal images of the knee immunostained for CXCL1 at 4 h p.c. (large panel; VSIG4, magenta; CXCL1, green; CD68, cyan; CD31, red; nuclei, gray; scale bar = 100 µm). Zoomed region of the lining shows expression of CXCL1 in various cells (middle panel, scale bar = 10 µm), whereas further enlargement demonstrates colocalization of CXCL1 and VSIG4 (right panel, scale bar = 3 µm). (D) Quantification of CXCL1 distribution in the synovium in additional mice from two independent experiments; Each violin plot represents one to two knee sections of a single mouse with median and interquartile range indicated. F, femur; FP, fat pad; M, meniscus; T, tibia.

Figure 2.

Activation of lining macrophages at the onset of AIA. (A) Confocal fluorescence microscopy of mouse knee demonstrates CXCL1 immunostaining localizing in the lining at 4 h after AIA induction (large panel; VSIG4, magenta; CXCL1, green; CD68, blue; nuclei, gray; scale bar = 200 µm) with representative quantification of CXCL1 expression in regards to the synovial lining of the image in A; each point is one CXCL1 “surface” rendered by Imaris software with median and interquartile range indicated. Zoomed region of the lining shows expression of CXCL1 in the cells of the lining niche (bottom left panel, scale bar = 7 µm), whereas further enlargement demonstrates colocalization of CXCL1 and VSIG4 indicated with a white arrowhead, as well as CXCL1 expression in VSIG4⁻ lining cell indicated with a red arrowhead (bottom right panel, scale bar = 2 µm). F, femur; FP, fat pad; M, meniscus; PB, patellar bursa; T, tibia. Images representative of 10 sections from five different mice and two independent experiments. (B) Heatmap of genes within the GO term “Neutrophil recruitment” expressed in synovial macrophages shows higher expression of chemokines in lining cells; CXCL1 indicated with an arrowhead. (C) Confocal image of sub-patellar synovial region demonstrating fluorescent mBSA and CXCL1 coincidence in the lining area (left panel; VSIG4, magenta; CXCL1, green; CD68, blue; mBSA, yellow; nuclei, gray; scale bar = 40 µm). Enlarged region shows lining macrophages with intracellular mBSA and CXCL1 expression (right panel, arrowheads, scale bar = 7 µm). F, femur; P, patella; PB, patellar bursa; PT, patellar tendon.

Figure 3.

Topography of recruited neutrophils during AIA: Onset-specific clustering in the synovial lining and interaction with mBSA-laden lining macrophages. (A and B) Confocal images of murine synovia during AIA in CX3CR1-eGFP Ly6G-tdTomato double reporter mice show that neutrophils cluster in the synovial lining niche at the onset: 6 h large overview (A) and 4 h anterior synovium ROI (B) have uniform distribution at day 2 (peak of inflammation, middle panel) and form localized swarms at day 7 (resolution of inflammation, bottom panel). CD68, cyan; CX3CR1, orange; VSIG4, magenta; Ly6G-tdTomato, green; nuclei, gray; scale bar = 300 and 50 µm in the entire mouse knee cross-section and ROIs, respectively. (C) Respective quantification of neutrophil position within the anterior synovium of the whole knee cross-sections as in A: neutrophil distance to the lining at the onset (6 h), peak (day 2), and resolution (day 7) of AIA presented in violin plots. Each violin plot represents three mice combined with two to three sections at different positions in the knee quantified, median and interquartile range indicated. Total quantifications: N = 3; n = 2–3; N indicates mouse group size, n the number of sections per mouse; samples from four independent experiments. P values by Kolmogorov–Smirnov comparison: ****, P < 0.0001. CL, cruciate ligaments; F, femur; FP, fat pad; M, meniscus; SC, synovial cavity. (D) Lining macrophages (L.MΦ) exhibiting higher mBSA uptake interact with more neutrophils (PMN) compared with the macrophages with lower mBSA uptake (whole anterior lining quantified on two different sections of a single mouse at 4 h p.c.; P value by Wilcoxon rank–sum test: *, P < 0.05). (E) Confocal image of the anterior synovium 4 h after injection of fluorescently labeled mBSA shows recruited neutrophils interacting with mBSA-laden lining macrophages (CD68, cyan; VSIG4, magenta; mBSA, yellow; neutrophil CD177, green; colocalization of VSIG4 and mBSA, white; scale bar = 10 µm). FP, fat pad; SC, synovial cavity.

Figure 4.

Activation of microvasculature in the synovial lining niche at the onset of AIA. (A) Representative confocal image of the synovium demonstrating higher number of transferred CXCR2 OE T cells compared with the control T cells and the closer proximity of OE T cells to the lining (CD4⁺; CD8⁺ T cells, green; VSIG4, magenta; control T cells, cyan; CXCR2 OE T cells, orange; nuclei, gray; scale bar = 30 µm); single experiment with two animals. (B) Numbers ratio of CXCR2 OE T cells to control T cells measured by FACS in the blood and the synovium of mice 6 h p.c. (5 h after adoptive T cell transfer) shows preferential recruitment of CXCR2 OE T cells to the synovium. Significance value by paired t test: *, P < 0.05. Experiment performed two times. (C) CXCR2 OE T cells in the synovial cavity interacting with the lining macrophage (CD177, green; VSIG4, magenta; CD31, blue; control T cells, cyan; CXCR2 OE T cells, orange; nuclei, gray; scale bar = 10 µm). (D) Confocal image of the whole mouse knee cross-section immunostained for endothelial activation marker E-selectin (CD62E) reveals distinct position of activated vasculature close to the lining (CD62E, green; CD31, magenta; CD68, cyan; lining fibroblast Prg4, orange; nuclei, gray; scale bar = 200 µm). (E) Segmentation of the vasculature in the anterior synovium of the image in A using Imaris to classify highly activated (CD62E^high) and non-activated (CD62E^−/low) blood vessels. Inset shows representative distribution of activated microvasculature in regards to the synovial lining of the image mask in E; each point is one blood vessel surface of the respective image; median and interquartile range indicated. (F) Enlarged region of the synovial lining with CD62E^high blood vessel is shown in the left panel, whereas the right panel shows patellar region with CD62E^high blood vessels in direct contact with lining macrophages (arrowheads), as well as an activated larger-diameter blood vessel situated a few cell layers deeper (scale bar = 20 µm). B, bone; F, femur; FP, fat pad; M, meniscus; SC, synovial cavity; T, tibia. Images and quantification representative of four animals from two independent experiments.

Figure S3.

Visualizing activated endothelium and IRF5-dependent activation of the synovial lining. (A) Distance of activated and non-activated vasculature to the lining at 4 h p.c. in three different mice from two experiments. Each violin plot represents two to three knee sections of a single mouse with median and interquartile range indicated. P values by Wilcoxon rank–sum test: ****, P < 0.0001. (B) Staining of IRF5 demonstrates expression in VISG4⁺ lining macrophages that is higher compared to inflammatory monocytes and neutrophils by FACS. (C) Representative confocal images of the synovia from WT and IRF5KO mice show fewer neutrophils in IRF5KO mice, in line with FACS data (CD68, cyan; VSIG4, magenta; CD177, green; nuclei, gray; scale bar = 50 µm). B, bone; SC, synovial cavity. (D) Representative images of Fast Green and Safranin O–stained knee sections from a single experiment in WT and CX3CR^CreER IRF5^fl/fl at day 7 p.c. (upper panels, articular surface region; lower panels, patellar region; asterisks indicate bone erosion, arrows synovial hyperplasia, and arrowheads fat pad infiltration; coronal orientation; scale bar = 300 µm). (E) Clinical scoring of respective sections for synovial hyperplasia, bone erosion, and BM proliferation shows improved pathology in lining macrophage–specific IRF5KO mice.

Figure 5.

IRF5 deficiency impairs the activation of lining macrophages and neutrophil recruitment at the onset of AIA. (A) FACS assessment of synovial neutrophil numbers at the onset of AIA (6 h) demonstrates diminished recruitment in both global and CX3CR1 driven IRF5KO mice. Data were pooled from three independent experiments with four to six mice per group. P values by two-way ANOVA with Šídák's multiple comparisons post hoc: *, P < 0.05; **, P < 0.01. (B) Reduced synovial neutrophil counts at 6 h p.c. by FACS in lining macrophage–specific IRF5KO (CX3CR1^CreERIRF5^fl/fl) achieved by tamoxifen administration 4 wk prior to induction of AIA. P values from a single experiment by two-way ANOVA with Šídák's multiple comparisons post hoc: **, P < 0.01. (C) GO comparison of WT and IRF5KO lining macrophages shows impaired activation of IRF5KO cells in response to mBSA 4 h p.c. (D) Differentially expressed genes between WT (wt) and IRF5KO lining macrophages at 4 h after mBSA challenge, with CXCL1 expression markedly reduced in IRF5KO cells. (E) Representative confocal images of the synovial lining in WT and IRF5KO mice receiving adoptive transfer of WT neutrophils reveals that transferred neutrophils interact more with lining macrophages in WT mice compared to IRF5-deficient mice at 4 h p.c. (transferred neutrophils [tPMN], cyan; VSIG4, magenta; nuclei, gray; scale bar = 30 µm). (F) Quantification of the contact between lining macrophages and transferred neutrophils from a single experiment (N = 2–3; n = 13–18). N indicates the number of mice per group, n the number of total sections quantified. P value by Wilcoxon rank–sum test: *, P < 0.05. B, bone; SC, synovial cavity.