Amyloid-DNA Composites of Bacterial Biofilms Stimulate Autoimmunity

Abstract

Research on the human microbiome has established that commensal and pathogenic bacteria can influence obesity, cancer, and autoimmunity through mechanisms mostly unknown. We found that a component of bacterial biofilms, the amyloid protein curli, irreversibly formed fibers with bacterial DNA during biofilm formation. This interaction accelerated amyloid polymerization and created potent immunogenic complexes that activated immune cells, including dendritic cells, to produce cytokines such as type I interferons, which are pathogenic in systemic lupus erythematosus (SLE). When given systemically, curli-DNA composites triggered immune activation and production of autoantibodies in lupus-prone and wild-type mice. We also found that the infection of lupus-prone mice with curli-producing bacteria triggered higher autoantibody titers compared to curli-deficient bacteria. These data provide a mechanism by which the microbiome and biofilm-producing enteric infections may contribute to the progression of SLE and point to a potential molecular target for treatment of autoimmunity.

Figure 1. Salmonella biofilms contain curli amyloids and extracellular DNA (eDNA)

(A) 3D projection (upper) and side view (lower) of Confocal Laser Scanning Microscopy images of GFP expressing S. Typhimurium (green) pellicle biofilms during 72-hour time course of growth. White bars represent 10 µm. (B) Curli expression was monitored by Flow cytometry (FC) for 72 hours in S. Typhimurium containing a P_csgBA::gfp reporter plasmid. (C) 3D projection of CLSM images of S. Typhimurium biofilm with P_csgBA::gfp reporter plasmid at 72 hours. Biofilm was stained with Congo Red (red). (D) FC analysis of cell death in Salmonella biofilms using live/dead staining with Syto 9 and PI. (E) 3D projection of CLSM images of a 72-Salmonella biofilm with P_csgBA::gfp reporter plasmid stained with PI (pseudo colored blue). (F) CLSM image of a 72-hour S. Typhimurium biofilm with P_csgBA::gfp reporter plasmid stained with PI. Red “corona” of extracellular DNA around the bacteria is visible. All data shown are representative samples of three independent experiments.

Figure 2. eDNA forms complexes with curli in biofilms accelerating the amyloid polymerization process and limiting its degradation by DNAse

(A) CLSM image of GFP expressing S. Typhimurium (green) biofilm grown for 72 hours and stained with Congo Red (red) and TOTO-1 Iodide (blue). (B) Crystal violet assay of S. Typhimurium biofilms grown in 96-well plates with or without 3 hour DNAse treatment. Data shown is average and SE from 5 wells. Experiments were completed in triplicate. (C) 3D reconstructions (upper) and side view (lower) CLSM images of GFP expressing Salmonella biofilms with and without DNAse treatment for 3 hours after biofilm was formed showing minimal change in biofilm thickness after treatment with DNAse. White bars on side views represent 10 µm. Images are representative of three independent experiments. (D) Microscopy of purified curli amyloids stained with Congo Red and visualized under polarized microscope (left panel), propodium iodide (center panel) and Hoescht 33258 (right panel) and visualized under fluorescent microscope. (E) Microscopy of purified curli fibers co-stained with Hoescht 33258 (left panel), ThT (center panel) and merged (right panel). (F) Calculated polymerization lag times of synthetic peptides consisting of the amino acid sequence of the fourth and fifth repeats of CsgA (CsgA R4-5) in the presence of increasing concentrations of CpG (left), genomic DNA from Salmonella (middle), and salmon sperm DNA (right). (G) Representative ThT amyloid polymerization assay curves for the polymerization of the CsgA R4-5 in the presence and absence of CpG (left), Salmonella genomic (middle), and salmon sperm (right) DNA. Three independent experiments were completed with duplicate wells for each concentration of DNA in each experiment. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 3. Curli-DNA composites are danger signals that activate dendritic cells

(A) GFP expressing cDCs (green) interact with S. Typhimurium biofilms stained with Congo Red (red). Image shown is representative of three independent experiments. (B) Cytokine production by cDCs incubated for 6 hours on 72 hour pre-formed biofilms from wild type, msbB, and ΔfliCfljB Salmonella. (C) 3D reconstructions of CLSM images of GFP expressing WT (Left Panel), msbB (Center Panel), and ΔfliCfljB (Right Panel) Salmonella biofilms showing equivalent growth of these biofilms in these conditions. (D) Proinflammatory cytokine production from cDCs stimulated with a dose titration of curli-DNA composites isolated from Salmonella msbB mutant biofilms. Cytokine production was compared to LPS stimulation (100ng/ml). Data in (B)(D) is representative of three independent experiments, performed with three independent bone marrow-derived cultures from 3 different mice and each contained quadruplicate wells for each bacterial strain. (E) Synthetic amyloid peptide CsgA R4-5 was polymerized in the presence or absence of Salmonella genomic DNA and used to stimulate cDCs. Cytokine production in response to 20µg/ml CsgA R4-5, 20 ng/µl genomic DNA, or 20µg/ml CsgA R4-5 fibrillized in the presence of 20ng/µl genomic DNA is shown as Average and SEM, n=5. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 4. Curli-DNA composites elicit strong type I interferon responses from cDCs in vitro

mRNA expression of interferon beta, Ifnβ, and two IFN-stimulated genes (ISGs), Irf7 and Isg15, from cDCs generated in vitro for 6–7 days from bone marrow of NZBxW/F1 mice (A) and C57BL/6 mice (B) and stimulated for 6 hours with curli/DNA composites isolated from Salmonella biofilms. All of the conditions were normalized against the control (untreated DCs in medium only) for each strain in each experiment. Results are average of three independent experiments, (C) ISG mRNA expression in NZBxW/F1 splenic CD11c+ DCs isolated by magnetic beads 20 hours post i.p. injection of 50µg of curli. *, P < 0.05.

Figure 5. Curli/DNA composites accelerate systemic autoimmunity in lupus-prone NZBxW/F1 mice

(A) Anti-dsDNA and anti-chromatin autoantibodies were measured by ELISA in sera from NZBxW/F1 mice injected with PBS thrice a week, curli once a week (1×), or curli thrice a week (3×). Optical density (O.D.) indicates ELISA color change and the presence of anti-dsDNA or anti-chromatin autoantibodies. Error bars indicate SEM, n=5 per group. The dotted horizontal line indicates cutoff for positivity, calculated as two standard deviations above the average of sera from naïve C57BL/6 mice. (B) Representative antinuclear antibody assay (ANA) of sera from mice as in Figure 5A at the indicated ages. Hep-2 cells were incubated with sera and FITC-conjugated anti-IgG Abs. Antibody isotype determinations of anti-chromatin (C) and anti-dsDNA (D) antibodies. Serum from 10 month old diseased Sle1,2,3 mouse was utilized as positive control and from naïve C57BL/6 mouse was used as negative control. *, P < 0.05.

Figure 6. Curli/DNA composites induce autoantibody production and immune cell activation in WT Mice

(A) Anti-dsDNA and anti-chromatin autoantibody ELISAs from C57BL/6 mice injected with PBS or curli thrice a week (3×). As in Fig. 5, optical density (O.D.) indicates ELISA color change and the presence of anti-dsDNA or anti-chromatin. Error bars indicate SEM, n=5 per group. (B) Flow cytometry analysis of the expression of activation markers in T-cells and B-Cells and (C) in DCs from spleens of C57BL/6 mice 20 hours post i.p. curli injection; (D) mRNA expression of two interferon-stimulated genes (ISGs), Irf7 and Isg15, from C57BL/6 splenic CD11c+ DCs isolated with magnetic beads from the same mice. Results are shown as averages and standard deviation, n=3 per group, *, P < 0.05.

Figure 7. Infection with curli-producing bacteria accelerates autoimmunity in lupus-prone mice

Anti-dsDNA and anti-chromatin autoantibody ELISAs from NZBxW/F1 mice injected with live E. coli or S. Typhimurium strains. (A) Wild type E. coli or its csgA mutant was injected i.p. into NZBxW/F1 mice at 10⁵ CFU for four times. (B) Live virulent S. Typhimurium or its isogenic csgBA mutant was injected i.p. into NZBxW/F1 mice at 10⁵ CFU for four times. (C) Autoantibody isotypes. The dotted horizontal line indicates cutoff for positivity, calculated as two standard deviations above the average of sera from naïve C57BL/6 mice. *, P < 0.05. (D) Two representative ANA stainings from each experimental group are shown. Magnified insets at the right corner of two pictures.