Identification of protein aggregates in the aging vertebrate brain with prion-like and phase-separation properties

Abstract

Protein aggregation, which can sometimes spread in a prion-like manner, is a hallmark of neurodegenerative diseases. However, whether prion-like aggregates form during normal brain aging remains unknown. Here, we use quantitative proteomics in the African turquoise killifish to identify protein aggregates that accumulate in old vertebrate brains. These aggregates are enriched for prion-like RNA-binding proteins, notably the ATP-dependent RNA helicase DDX5. We validate that DDX5 forms aggregate-like puncta in the brains of old killifish and mice. Interestingly, DDX5's prion-like domain allows these aggregates to propagate across many generations in yeast. In vitro, DDX5 phase separates into condensates. Mutations that abolish DDX5 prion propagation also impair the protein's ability to phase separate. DDX5 condensates exhibit enhanced enzymatic activity, but they can mature into inactive, solid aggregates. Our findings suggest that protein aggregates with prion-like properties form during normal brain aging, which could have implications for the age-dependency of cognitive decline.

Keywords: CP: Cell biology; CP: Neuroscience; N. furzeri; aggregates; aging; killifish; phase separation.

Figure 1.. Quantitative identification of age-related protein aggregation in the vertebrate brain

(A) Quantitative proteomic analysis pipeline for the brain (and other organs) from young, old, and old TERT^Δ8/Δ8 male fish. For each organ, a tissue lysate (TL) and an aggregate (AGG) fraction were isolated, and tandem mass tag (TMT) labeling and mass spectrometry were performed. Workflow and tags: Table S1A. (B) Silver stain of SDS-PAGE gel with TL and AGG samples from young and old brains. Similar amounts (~1 μg) of TL and the AGG fraction from brain were used. Note that the protein quantity for the AGG fraction was underestimated. A version of the gel with additional organs is in Chen et al. (C) Principal-component analysis (PCA) on protein abundance in TL or AGG fractions from the brain of young (light blue square), old (dark blue circle), and old TERT^Δ8/Δ8 (gray triangles) fish. Each symbol represents an individual fish. There was an outlier for the old brain sample due to quantification prior to mass spectrometry analysis. Another version of this PCA is in Chen et al. (D) Ranked list of proteins with increased aggregation propensity in old versus young brains based on descending fold change (FC)—i.e., the relative abundance ratio of AGG over TL averaged across three fish in old versus young killifish brain (p < 0.05, Student’s t test). Predicted prion-like domain (PrD) scores (PLAAC) and fraction of disordered amino acids (aa) (DISOPRED3) are presented (Table S1D). Red: PrD score > 0 and fraction of disordered aa > 0.3 (30%) are considered to be prion-like proteins or intrinsically disordered proteins, respectively. Paralogs are indicated as “X of Y” (“2 of 3” means the second paralog out of three identified paralogs) (Table S1D). See also Table S1B. (E) Selected GO enrichment using gene set enrichment analysis (GSEA) for proteins with an increased aggregation propensity in old brains. Complete set in Figure S1A. Enrichment score, the score for enrichment/depletion of GO terms in this gene set. Counts, number of genes. p values were not false discovery rate (FDR) corrected. (F) Enrichment for PrDs in proteins with increased aggregation propensity with age in the brain and other tissues. Data represent the fraction of proteins with putative PrD (PrD score > 0). Red: all detected aggregated proteins; blue: proteins with increased aggregation propensity with aging (t ≥ 0.75, p < 0.05 from Student’s t test). Age-associated differences in aggregation propensity were assessed using Fisher’s exact test. Numbers in Table S3A.

Figure 2.. The RNA helicase DDX5 has a conserved PrD and shows increased aggregate-like puncta in old vertebrate brains

(A and B) Presence and location of predicted PrD (A) and intrinsically disordered region (IDR; B) in DDX5 in killifish and humans. Red line: PLAAC score. DDX5 amino acid positions are indicated at the bottom. Amino acid composition is color coded. Blue line: local disorder score (DISOPRED3). (C) Immunohistochemistry for DDX5 in brain sections from young (3.5 months) and old (7 months) male killifish. Green: DDX5; blue: DAPI (nuclei). Image representative of 3 individual fish for each age group. Staining was performed twice independently, ~7 sections per fish per experiment. Scale bar: 10 μm. Quantification of the subcellular localization of DDX5 puncta is in Figure S2F. (D) Immunohistochemistry for DDX5 in brain sections from young (4 months) and old (28 months) male mice. Green: DDX5; blue: DAPI (nuclei). Image representative of 2 mice per age group, 6 sections per mouse. Scale bar: 20 μm. (E) Generation of Tg(ddx5:DDX5-GFP) transgenic F0 founders (there was no germline transmission). (F) Immunohistochemistry for GFP in brain sections from old (~6–7 months) male Tg(ddx5:DDX5-GFP) F0 founders (which can have variable protein expression). Green: GFP; blue: DAPI (nuclei). Image representative of 3 individual fish, ~6 sections per fish. Scale bar: 10 μm. Immunocytochemistry in human 293T cells for untagged killifish DDX5 full length (red, antibody to killifish DDX5) or GFP-tagged killifish DDX5 (full length [FL] and two different truncation mutants [ΔIDR and ΔPrD] [green, antibody to GFP]). Left: representative images from two independent experiments, each performed in triplicate. White arrows indicate cells with puncta (determined by signal intensity over a pre-defined maximum). Scale bar: 10 mm. Right: mean ± SD of the percentage of transfected cells showing DDX5 puncta. Each dot represents the average of ~240 cells from 3 fields of view over two independent experiments. Circles: experiment #1; triangles: experiment #2. p values: Kruskal-Wallis with Dunn’s correction.

Figure 3.. DDX5 exhibits prion-like seeding properties

(A) “1-color” experimental design to assess protein aggregation in yeast by conditionally expressing C-terminally EGFP-tagged killifish protein candidates under a galactose-inducible promoter. Protein aggregation is scored by quantification of the EGFP fluorescent puncta. (B and C) Identification of killifish candidate proteins that do not aggregate (B) or that do aggregate (C) following galactose-inducible overexpression in yeast. White arrows: fluorescent puncta (aggregates). Scale bar: 5 μm. (D) “2-color” experimental design to assess prion-like aggregate propagation in yeast. This design tests whether brief overexpression (“induction”) of a killifish candidate (C-terminally tagged with mRuby3 and under the control of a galactose-inducible promoter) can induce a cross-generational prion-like propagation of a constitutively expressed candidate (C-terminally tagged with EGFP and under the control of the glyceraldehyde 3-phosphate dehydrogenase (GPD) promoter with a single-copy-number plasmid). The cross-generational effect was tested during “outgrowth” (i.e., 200-fold dilution), highlighting aggregates that can stably propagate for 7–8 generations in yeast cells. Protein aggregation is scored by quantification of mRuby3 (red) or EGFP (green) fluorescent puncta. (E) Using the 2-color system, killifish DDX5-EGFP aggregates were visible in both induction and outgrowth phases, indicating prion-like propagation. Representative of 4 independent experiments. White arrows: fluorescent puncta (aggregates). Scale bar: 5 μm. (F) Quantification of DDX5 aggregates and prion-like seeding potential by calculating the fraction of EGFP-positive cells with EGFP puncta during pre-induction, induction, and outgrowth. Black bar indicates mean values from 4–6 independent experiments (6 independent pre-induction measurements and 4 independent experiments with matched pre-induction, induction, and outgrowth measurements). Each dot represents the average value within each experiment (an average of 135 GFP-positive yeast cells were quantified for each strain under each experimental condition). p values: Student’s t test. (G) Quantification of DDX5 ΔPrD aggregation and prion-like seeding potential by calculating the fraction of EGFP-positive cells with EGFP puncta during pre-induction, induction with DDX5 FL-mRuby3, and outgrowth. Black bar indicates mean values from 4–5 independent experiments (5 independent pre-induction measurements and 4 independent experiments with matched pre-induction, induction, and outgrowth measurements). p values: Student’s t test. (H) Fluorescence recovery after photobleaching (FRAP) of killifish DDX5 (blue line) and yeast Sup35NM (a bona fide yeast prion, red line) puncta in yeast. Representative of 2 independent experiments. Mean ± SEM of the normalized intensity of the photobleached area at each time point (10 s interval with photobleaching performed at the 4^th frame).

Figure 4.. DDX5 phase separates and can form aggregates

(A) Phase diagrams for purified killifish DDX5 recombinant protein FL and truncation mutants (ΔPrD and ΔIDR) under variable NaCl (y axis) and protein concentrations (x axis). Empty squares: diffuse state; full circle: condensate state. (B) Left: phase separation temporal dynamics of DDX5 FL or truncation mutants (ΔPrD and ΔIDR) at various temperatures. Red: DyLight 549-labeled C-terminally SNAP-tagged recombinant protein. Representative of 2 independent experiments. Scale bar: 5 μm. Top right: Differential interference contrast (DIC) representative image of DDX5 condensates. (C) Experimental design to measure DDX5 ATPase activity in diffuse or condensate states. Empty squares: diffuse state; full circle: condensates. Bottom: western blot to verify DDX5 protein concentration. (D) Assessment of ATPase activity of the killifish DDX5 protein in diffuse (red circles) or condensate (blue circles, above 2 μM) forms. ATPase activity is measured as a function of the free phosphate (Pi) release rate (μM/min) at indicated DDX5 protein concentrations (μM). Values of 4 replicates from one representative experiment. Two independent experiments were performed (2^nd experiment in Figure S5B). DDX5 has no activity in the absence of RNA. (E) Formation of more disorganized DDX5 aggregates over time. SNAP-tagged recombinant DDX5 labeled with the DyLight 549 dye (left panels). Quantification of the form factor (sphericality of the condensates/aggregates) (right panel). p values: Student’s t test. (F) Comparison of ATPase activity of killifish DDX5: aggregate (red or blue bars), diffuse (green bar), and buffer only (purple bar). ATPase activity is measured as in (D). Representative of 2 independent experiments (mean ± SD from one experiment). (G) DDX5 aggregates upon seeding with aggregated DDX5 protein at indicated molar ratios (top) and time post-seeding (left). SNAP-tagged recombinant DDX5 labeled with DyLight 549 dye (Protein). His10- and SNAP-tagged recombinant DDX5 aggregates labeled with Alexa 488 (Seed). Scale bar: 25 μm (top) or 50 μm (bottom). Representative of 2 independent experiments.