Synovial fibroblast-neutrophil interactions promote pathogenic adaptive immunity in rheumatoid arthritis

Abstract

Rheumatoid arthritis (RA) is characterized by synovial joint inflammation and by development of pathogenic humoral and cellular autoimmunity to citrullinated proteins. Neutrophil extracellular traps (NETs) are a source of citrullinated autoantigens and activate RA synovial fibroblasts (FLS), cells crucial in joint damage. We investigated the molecular mechanisms by which NETs promote proinflammatory phenotypes in FLS, and whether these interactions generate pathogenic anti-citrulline adaptive immune responses. NETs containing citrullinated peptides are internalized by FLS through a RAGE-TLR9 pathway promoting FLS inflammatory phenotype and their upregulation of MHC class II. Once internalized, arthritogenic NET-peptides are loaded into FLS MHC class II and presented to Ag-specific T cells. HLADRB1*0401 transgenic mice immunized with mouse FLS loaded with NETs develop antibodies specific to citrullinated forms of relevant RA autoantigens implicated in RA pathogenesis as well as cartilage damage. These results implicate FLS as mediators in RA pathogenesis, through the internalization and presentation of NET citrullinated peptides to the adaptive immune system leading to pathogenic autoimmunity and cartilage damage.

Conflict of interest statement

Competing interests: PRT has a patent on Rh-PG probe and is a consultant to Padlock Therapeutics, a wholly owned subsidiary of Bristol Myers Squibb

Figures

Figure 1. ACPAs recognize multiple citrullinated peptides in NETs and induce NETosis

A. Rhodamine-Phenylglyoxal probe against citrulline was utilized to detect specific citrullinated proteins in purified NETs generated by stimulating control neutrophils with IgM rheumatoid factor. B. ACPAs differentially recognize citrullinated autoantigens in NETs when compared to control IgG. Spontaneously generated NETs from peripheral blood neutrophils from two RA patients (RA01 and RA02) were isolated and resolved in SDS-PAGE. Western blot was performed using ACPAs or control IgG. C. ACPAs bind to NETs and D. enhance NETosis. Red represents control IgG (ctrl-IgG) or ACPAs, and blue is Hoechst. E. Representative PEAKS’ histogram of citrullinated peptides detected in NETs. F. Mass Spectrometry analysis demonstrates multiple citrullinated peptides in NETs. G–H ELISA analysis of synovial fluid from OA (n=10) and RA (n=17) patients to detect autoantibodies recognizing citrullinated or non-citrullinated forms of MPO and neutrophil elastase, respectively. I–J ELISA analysis of IgGs isolated from synovial fluid (SF) of OA (n=6) and RA (n=11) patients showing recognition of native or citrullinated MPO or elastase. A, B and C are representative of three independent experiments. Scale bars, 10μm. Results are the mean ± s.e.m. of n= 5–6. For statistical analyses Mann-Whitney U test was used.

Figure 2. NETs are internalized by FLS into early endosomal antigen 1 (EEA1)-positive compartments

RA-NETs were incubated with OA or RA FLS or skin fibroblasts for 2h. A. Internalized NETs colocalize with EEA1 compartments in FLS. Red represents MPO, green is EEA1 and blue is DNA. Results are representative of three independent experiments performed with confocal microscope, Scale bars, 10 μm. B. Western blot analysis of fibroblasts incubated in presence or absence of RA NETs shows that MPO bound to NETs is internalized by OA and RA FLS, but minimally by healthy control skin fibroblasts. Results are representative of three independent experiments C. Percentage of MPO positive cells (FLS that internalized NETs) decreased after chloroquine (CQ), but not cytochalasin D (cytoD) exposure. D. Representative confocal images after treatment with cytoD and CQ. Red represents MPO or F-actin, green is MPO and blue is DNA, Scale bars, 10 μm. E. Western blot analysis confirms that MPO internalization is impaired in FLS preincubated with CQ, but not in FLS pre-incubated with cytoD. F. IL-6 release by FLS is dependent on NET internalization. Results are the mean ± s.e.m. of n=4. For statistical analyses Mann-Whitney U test was used; ns= non-significant.

Figure 3. RAGE-TLR9 axis mediates internalization of NETs by FLS

OA and RA FLS were incubated in absence or presence of TLR9 inhibitors or control oligos and NETs for 72h. A. Western blot analysis shows intracellular MPO when FLS were incubated in presence or absence of TLR9 inhibitor. B. IL-6 and C. IL-8 quantification on FLS supernatants. D. qPCR analysis displays TLR9 mRNA expression in OA and RA FLS and skin fibroblasts in the presence or absence of NETs. Results for B–D are the mean ± s.e.m of four independent experiments. E. FLS were pre-treated with or without 2–4 μM RAGE inhibitor, and incubated with NETs for 1 hour. Co-immunoprecipitation was performed against RAGE and TLR9 was detected by Western blot. Anti-GFP was used as negative control. F. Plasma membrane (top panel) and intracellular (lower panel) detection of RAGE and TLR9 was performed on FLS pre-treated with or without RAGE inhibitor; red represents TLR9, green represents RAGE and blue is DNA of three independent experiments; white arrows highlight areas of colocalization of RAGE and TLR9. Scale bars, 5 μm. Mann-Whitney U test was used.

Figure 4. IL-17B present in NETs upregulates MHCII in OA and RA FLS

OA and RA FLS were incubated in presence or absence of spontaneously generated RA-NETs or 1000 U/mL IFN-γ A. Detection of MHCII in FLS by immunofluorescence; green represents MHCII and blue is DNA. Results are representative of three independent experiments, scale bars: 10μm. B. Plasma membrane and intracellular MHCI and MHCII were quantified by flow cytometry in RA FLS treated with NETs, IFN-γ or IFN-α for 5 days. C. IL-17B (red) is externalized in control (Ctrl) NETs generated with 1μg/mL of LPS and in spontaneously generated-RA NETs. D. IL-17B is detected in isolated NETs by Western Blot analysis. Each lane depicts independent NET isolation per group (Ctrl and RA). Mann-Whitney U test was used. Results are mean ± s.e.m. of 10 independent experiments,*p<0.05. E. OA FLS were incubated with RA-NETs in presence or absence of 1μg/mL IL-17B neutralizing Ab for 48h. Quantification of HLADR-A and DR-B mRNA was performed by real-time PCR. Bars mean ± s.e.m. of four independent experiments F. RA FLS were incubated with 100 ng of human recombinant IL-17A or IL-17B for 72h. qPCR and immunofluorescence analyses assessed MHCII mRNA expression and protein localization, respectively. ANOVA Bonferroni’s test was performed for B, E–F. Results are the mean ± s.e.m. of four independent experiments.

Figure 5. Arthritogenic peptides contained in NETs internalized by FLS are loaded into their MHCII compartment and presented to Ag-specific CD4+ T-cells

A. Internalized NET-bound proteins colocalize with MHCII compartments in OA and RA FLS. B. Plasma membrane detection of MHCII and NET proteins in unpermeabilized FLS assessed by immunofluorescence. Red represents MPO or HC-gp39, green is MHCII and blue is DNA. C. Detection of mouse IL-2 after DRB1*04:01 RA FLS (with and without NETs) were incubated in the presence or absence of HC-gp39-specific CD4+ T-cell hybridomas for 5 days. Peptide is HC-gp39 263–275 (RSFTLASSETGVG). D Detection of various cytokines after DRB1*04:04 RA-FLS (with or without NETs) were incubated with haplotype matched cit-vimentin-specific CD4+ T-cells in the presence or absence of neutralizing antibodies against CD28 or MHCII. Mann-Whitney U test was used. Results are mean ± s.e.m. of four-six independent experiments. Scale bars, 10μm.

Figure 6. DRB1*04:01 humanized mice that receive intraarticular injections of mouse FLS loaded with RA-NETs develop ACPAs and cartilage damage

DRB1*04:01 FLS were isolated and incubated with human RA-NETs for 3 days prior to intra-articular injection. A. Internalization of NETs by FLS was assessed by immunofluorescence using an antibody against MPO (red). B. Serum ACPA levels at various time-points in animals immunized with FLS alone or FLS loaded with NETs (n=5/group). Results represent mean ± s.e.m. Mann-Whitney U test was used C. Sera from 3 mice immunized for 7 weeks with FLS or with FLS loaded with NETs was analyzed by dot-blot against citrullinated proteins. D. Western blot analysis to detect serum autoantibodies recognizing human NET proteins (arrows) in DRB1*04:01 animals that received FLS alone or FLS+NETs; each lane depicts two independent NET isolations. E. IL-2 synthesis by DRB1*04:01 mouse splenocytes incubated with a cocktail of native (pep; H3, H4, MPO and vimentin) or citrullinated peptides (cit pep; a cocktail of cit-H3, cit-H4, cit-MPO and cit-vimentin), when comparing animals immunized with FLS loaded with NETs (DR4 + NETs) with animals immunized with FLS alone (DR4). PHA is used as positive control. Results represent mean ± s.e.m of six independent experiments. Mann-Whitney U test was used. F. CD4+ T cell-depleted animals immunized with FLS loaded with NETs demonstrate significantly lower titers of ACPAs (n=4), as measured by ELISA, when compared with non-T cell depleted mice. Results represent mean ± s.e.m. Mann-Whitney U test was used. G. Percentage of FLS that migrate to the spleen and lymph nodes (LNs) after intraarticular injection of FLS loaded with NETs or untreated mice (n= 4–5). Results represent % mean ± s.e.m of FLS positive for CellTraceViolet (%CTV). Mann-Whitney U test was used. H. Epitope Chip analysis to quantify antibodies recognizing specific epitopes of RA-relevant autoantigens in animals immunized with FLS loaded with NETs when compared with animals immunized with FLS alone. Heat map represents the average of mean fluorescent units (MFI) of 5 animals/group. I. Sagittal sections of cartilage of the injected tibiofemoral compartment were stained with Safranin O (Saf-O) demonstrating impaired cartilage integrity (arrowhead) in animals immunized with FLS loaded with NETs when compared to animals immunized with FLS alone. J. Percentage of cartilage loss of the femurs and tibias of animals immunized with FLS or FLS loaded with NETs. Results represent mean ± s.e.m of 2–3 independent experiments. Mann-Whitney U test was used.