The ASC inflammasome adapter governs SAA-derived protein aggregation in inflammatory amyloidosis

Abstract

Extracellularly released molecular inflammasome assemblies -ASC specks- cross-seed Aβ amyloid in Alzheimer's disease. Here we show that ASC governs the extent of inflammation-induced amyloid A (AA) amyloidosis, a systemic disease caused by the aggregation and peripheral deposition of the acute-phase reactant serum amyloid A (SAA) in chronic inflammatory conditions. Using super-resolution microscopy, we found that ASC colocalized tightly with SAA in human AA amyloidosis. Recombinant ASC specks accelerated SAA fibril formation and mass spectrometry after limited proteolysis showed that ASC interacts with SAA via its pyrin domain (PYD). In a murine model of inflammatory AA amyloidosis, splenic amyloid load was conspicuously decreased in Pycard^-/- mice which lack ASC. Treatment with anti-ASC^PYD antibodies decreased amyloid loads in wild-type mice suffering from AA amyloidosis. The prevalence of natural anti-ASC IgG (-logEC₅₀ ≥ 2) in 19,334 hospital patients was <0.01%, suggesting that anti-ASC antibody treatment modalities would not be confounded by natural autoimmunity. These findings expand the role played by ASC and IL-1 independent inflammasome employments to extraneural proteinopathies and suggest that anti-ASC immunotherapy may contribute to resolving such diseases.

Keywords: ASC; Amyloidosis; Inflammation; Innate Immunity; Serum Amyloid A (SAA).

Figure 1. Co-aggregation of ASC and SAA in human inflammation-induced amyloid A amyloidosis.

(A, B) ASC and Congo-red-stained myocard from patients without (A) and with AA amyloidosis (B; 68 and 82-year old males, respectively). (C–E) Individual and merged confocal images of cardiac blood vessels (labeled with PAI1) from post-mortem tissue of a patient with AA amyloidosis, presenting high level of SAA (labeled with MAS41676 and DM003) and ASC (labeled with AL177 and MAB/MY6745). (F, G) Individual and merged confocal images of cardiac blood vessels (labeled with PAI1) from post-mortem tissue, presenting low level of SAA (labeled with MAS41676 and DM003) and ASC (labeled with AL177). (H) Graphical representation of the quantification of 100X confocal images for SAA and ASC. Each point represents one field of view, bars represent median. Integrated density was normalized by the mean value control. Data from one patient in each group are shown. (I) Confocal images from post-mortem tissue of a patient with AA amyloidosis showing accumulation of both SAA (MAS41676, magenta) and ASC (AL177, yellow) within the blood vessel border (labeled with PAI1) and in adjacent muscle cells. (J) STED imaging of co-aggregation SAA (magenta) and ASC (yellow). Imaris 3D model shows the intricacy of SAA and ASC within these aggregates. (K) Confocal images from post-mortem tissue of patient with AA amyloidosis showing accumulation of both SAA (DM003, magenta) and ASC (MAB-ASC (MY6745), yellow) within the blood vessel border (labeled with PAI1) and in adjacent muscle cells. (L) STED imaging of co-aggregation of SAA (magenta) and ASC (yellow). Imaris 3D model shows the intricacy of SAA and ASC within these aggregates. Source data are available online for this figure.

Figure 2. ASC contacts specific SAA epitopes and accelerates SAA fibrillation.

(A) In vitro SAA aggregation in the presence or absence of ASC specks over 150 h. Each curve represents a technical replicate. A logistic regression was performed on replicates of all conditions and the fitted curve is shown. (B) Latency to half-maximal fluorescence signal (t_1/2) for 0, 200k and 500k ASC specks. Standard boxplots: median and quartiles (Wilcoxon rank-sum test with Holm correction for multiple comparisons). Each dot represents one measurement. *P < 0.05 (P values: 0.048; 0.048; 0.013). (C) Overview of LiP-MS. (D–F) Volcano plots (top) and ASC 3D structures (bottom) showing the LiP-MS hits and affected ASC epitopes (red) and respective amino acid sequences after mixing monomeric recombinant human ASC with monomeric recombinant human SAA1 (D), monomeric recombinant human ASC with recombinant human SAA1 fibrils (E), or human ASC specks with monomeric recombinant human SAA1 (F). Ordinate; FDR-adjusted (D, E) and unadjusted P values (F); unpaired two-tailed t test. (D–F) The colors denote the following: red dots—significant changes in the ASC protein. Light blue dots—significant changes other than in the ASC proteins originating from co-purified proteins. Gray dots—nonsignificant changes. Source data are available online for this figure.

Figure 3. Decreased splenic amyloid load in the absence of ASC.

(A) Experimental design. Experimental groups consisted of total n = 46 AA⁺ (n = 24 Pycard^−/−/n = 22 Pycard^+/+) and n = 27 AA⁻ (n = 14 Pycard^−/−/n = 13 Pycard^+/+) animals (P values: 0.003; 0.008; 0.03). (B) SAA serum concentrations (mean ± SEM; triplicates; Wilcoxon rank-sum test) in Pycard^+/+ and Pycard^−/− mice after injection of AgNO₃, AEF, or AgNO₃ + AEF (AA amyloid induction). (C) AA⁺ (Pycard^+/+ and Pycard^−/−; P30) spleens stained with LCP (green). Arrows: red pulp invasion of amyloid. Asterisks: follicles. (D) Quantification of splenic LCP signal. Dots: individual mice. Scatterplots: mean ± SEM (error bars); Wilcoxon rank-sum test with Holms correction for multiple comparisons (P values: 0.002; 0.001; 0.0004; 0.002). (E) Top: Western blot of monomeric SAA in spleen homogenates of Pycard^+/+ and Pycard^−/− mice. To ascertain SAA antibody functionality, mouse spleen homogenate from independently obtained and Congo red-confirmed AA⁺ tissue served as positive, whereas non-induced (AA⁻) spleen tissue served as negative technical controls. Bottom: Quantification of Western blot data, from four mice per group. mean ± SEM; Kruskal–Wallis test (P value: 0.029). (F) Spleen sections stained with HE and ASC IHC from Pycard^+/+ (wt) and Pycard^−/− AA⁺ mice. Arrows: amorphous amyloid structures. Asterisks: splenic follicles. ASC with brown stain in IHC. Boxes are magnified in the right panels. (G) Spleen sections stained with HE and ASC from Pycard^+/+ AA⁻ mice. (H) Phagocytic activity (OD) from unstimulated and mSAA-stimulated Pycard^+/+ and Pycard^−/− BMDMs specimens (P values; top: 0.002; bottom: 0.002). Each assay was carried out in duplicates. Cyt D: 10 µM Cytochalasin D. Mean ± SEM. Wilcoxon rank-sum test with Holms correction for multiple comparisons. Dots: individual bone marrow samples. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns: not significant. Source data are available online for this figure.

Figure 4. Anti-ASC^PYD immunotherapy reduces peripheral amyloid load.

(A) Scheme of anti-ASC immunotherapy. Total n = 15 experimental AA⁺ Pycard^+/+ (wt) animals (n = 5 per treatment group). (B) Anti-ASC antibody single-injection pharmacokinetics. N = 3 per treatment group, technical duplicates; mean ± SEM. (C) Western blot of SAA in spleen homogenate of Pycard^+/+ AA⁺ mice treated with anti-ASC or isotype antibody, or untreated. Absence of amyloidosis induction as control (AA⁻). Actin and spleen homogenate from a C57BL/6 wt mouse served as negative control. Scatterplots show data from three mice/group and two independently performed immunoblots resulting in 6 samples and data points (technical duplicates; mean ± SEM, Kruskal–Wallis test, P value: 0.005). (D) HE-stained spleen sections of Pycard^+/+ AA⁺ mice. Arrows: amyloid. Asterisks: white pulp. (E) Spleen sectiond stained with ASC antibodies (brown). (F) Quantification of ASC-positive areas in spleen sections. Two fields of view, each representing a different section for each mouse (3 mice/group) were quantified. Scatterplots: mean ± SEM. Statistics: Kruskal–Wallis test, P value: 0.007. (G) Spleen sections stained with LCP. (H) Integrated density expressed as arbitrary units/µm² in LCP-stained spleen sections. Each dot represents a quantified field of view. Three field of views for each animal were analyzed (nine fields of view from three mice per group). Scatterplots: mean ± SEM; Kruskal–Wallis test, P value: 0.0001. **P ≤ 0.01, ****P ≤ 0.0001. Source data are available online for this figure.

Figure 5. Population-wide interrogation of autoantibodies against ASC in a large unselected hospital cohort advocates stringent immune tolerance against ASC.

(A) Sample provenance by various hospital units. The contributions are depicted as percentage. (B) Age distribution of subjects. (C) Raincloud plots which includes standard boxplots (median and quartiles) displaying jittered p(EC₅₀) values for all patients screened (All), for the fraction of patients characterized by ICD-10 code E85 (Amyloidosis), AA amyloidosis, or Alzheimer’s disease. The dashed red line at p(EC₅₀) = 2: reactivity cutoff. Blue dots represent the hits, characterized by p(EC₅₀) ≥ 2 and a mean squared residual error < 20% of the actual p(EC₅₀). None of the groups were significantly different (Kruskal–Wallis test, P value = 0.079, α = 0.01; Wilcoxon rank-sum test P value after Holm correction for multiple comparisons: 0.43 (All vs. Amyloidosis), 0.47 (All vs. AA Amyloidosis), 0.47 (Amyloidosis vs. AA Amyloidosis), 0.47 (All vs. Alzheimer), 0.28 (Amyloidosis vs. Alzheimer), 0.43 (AA Amyloidosis vs. Alzheimer), α = 0.01. (D) Heatmap showing p(EC₅₀) values of seropositive samples from screen and controls assayed against an array of antigens. Only the plasma of patient #5 exceeded the cutoff value and was a confirmed binder of human ASC. All samples and controls were assayed as duplicates. (E) Patient #5 (purple), along with multiple control antibodies, was assayed against the full-length ASC protein (ASC), the PYD of ASC, the CARD of ASC, and against LAG3 and TIM3. All proteins used contained a his-tag. All samples and controls were assayed as duplicates. (D, E) Two scales are given: (1) for plasma samples, the p(EC₅₀) of the respective plasma dilution is used. (2) For monoclonal antibodies of known concentration, the p(EC₅₀) of the concentrations (in µg/ml) are shown. Anti-ASC, Anti-PrP, and anti-Tau441-positive control (PC) as well as CARD-ab, PYD-ab, and His-ab are monoclonal antibodies. Source data are available online for this figure.