Proteins in aggregates functionally impact multiple neurodegenerative disease models by forming proteasome-blocking complexes

Abstract

Age-dependent neurodegenerative diseases progressively form aggregates containing both shared components (e.g., TDP-43, phosphorylated tau) and proteins specific to each disease. We investigated whether diverse neuropathies might have additional aggregation-prone proteins in common, discoverable by proteomics. Caenorhabditis elegans expressing unc-54p/Q40::YFP, a model of polyglutamine array diseases such as Huntington's, accrues aggregates in muscle 2-6 days posthatch. These foci, isolated on antibody-coupled magnetic beads, were characterized by high-resolution mass spectrometry. Three Q40::YFP-associated proteins were inferred to promote aggregation and cytotoxicity, traits reduced or delayed by their RNA interference knockdown. These RNAi treatments also retarded aggregation/cytotoxicity in Alzheimer's disease models, nematodes with muscle or pan-neuronal Aβ₁₋₄₂ expression and behavioral phenotypes. The most abundant aggregated proteins are glutamine/asparagine-rich, favoring hydrophobic interactions with other random-coil domains. A particularly potent modulator of aggregation, CRAM-1/HYPK, contributed < 1% of protein aggregate peptides, yet its knockdown reduced Q40::YFP aggregates 72-86% (P < 10(-6) ). In worms expressing Aβ₁₋₄₂, knockdown of cram-1 reduced β-amyloid 60% (P < 0.002) and slowed age-dependent paralysis > 30% (P < 10(-6)). In wild-type worms, cram-1 knockdown reduced aggregation and extended lifespan, but impaired early reproduction. Protection against seeded aggregates requires proteasome function, implying that normal CRAM-1 levels promote aggregation by interfering with proteasomal degradation of misfolded proteins. Molecular dynamic modeling predicts spontaneous and stable interactions of CRAM-1 (or human orthologs) with ubiquitin, and we verified that CRAM-1 reduces degradation of a tagged-ubiquitin reporter. We propose that CRAM-1 exemplifies a class of primitive chaperones that are initially protective and highly beneficial for early reproduction, but ultimately impair aggregate clearance and limit longevity.

Keywords: (protein) aggregation; Alzheimer (disease); C. elegans; Huntington (disease); neurodegeneration; proteasome.

Figure 1

Aggregated proteins from Caenorhabditis elegans adults expressing Q40::YFP in muscle. (A–H): center areas of 2D gels, stained with SYPRO Ruby, resolving proteins from aggregates pulled down with antibody to GFP. Worms were grown from hatch on ‘FV’ bacteria without RNAi (A–D) or expressing dsRNA targeting cram-1 (E, F) or pqn-53 (G, H). Aggregates, isolated from strain AM141 at 3 d_PH (A, B) or 5 d_PH (C–H), were partitioned into those soluble (A, C, E, G) or insoluble (B, D, F, H) in 1% sarcosyl. Lanes contain equal worm equivalents of aggregated proteins, dissolved in Laemmli buffer at 95 °C. Material at ∼40 kDa binds antibody to GFP and thus may comprise modified/degraded fragments of Q40::YFP. Sarcosyl-soluble and sarcosyl-insoluble aggregate proteins from wild-type/Bristol-N2 at 2 d_PH (L4 larvae) or 5 d_PH (adults) were resolved by 1D electrophoresis (I) and quantified (J). *Two-tailed t-test P < 0.003; **P < 0.0003.

Figure 2

Aggregates and associated traits are lessened by RNAi. (A–D) For 3 days from the L4/adult molt, worms were fed E. coli with empty vector (FV) or expressing dsRNA targeting cram-1 or pqn-53. Worms were imaged on 5 d_PH (adult day 2.5). (A–C), fluorescence images; (D) aggregate count (blue bars, left scale) and intensity/aggregate (orange bars, right scale), ± SEM, for N = 12–14 worms/group. Each knockdown was compared to control by two-tailed t-tests. (E) Fluorescence recovery after photobleaching was assessed as described (van Ham et al., 2009). For times ≥ 4 min, each group differed from either other group by two-tailed t-test, P < 2E-6. (F) CL4176 worms were exposed to RNAi or MG132 from the L3/L4 molt; myo-3p/Aß_1–42 was induced 48 h later by upshift to 25 °C, assaying paralysis 32 h later. (G, H), paralysis 28 h postinduction, as (F) except treatments began at hatching, with Aß_1–42 induced 48 h later. (F–H), MG132 was added at 20 μm, oleuropein (Ole) at 80 μg mL⁻¹; N > 150/group. (I) Uninduced CL4176 (at 20 °C) undergo age-dependent paralysis, delayed ∼30% by cram-1 RNAi fed from the L4/adult molt (P < 10⁻⁶; N = 34–35/group). (J) CL4176 worms, fed dual RNAi from early L4, had Aß_1–42 induced at 48 h; paralysis and amyloid were measured 32 h later. The cram-1/FV RNAi mix (blue bar) reduced paralysis below other groups (each P < 0.001); in two repeats, each P < 0.05, 0.002. (K) CL4176 adults were stained with thioflavin T (ThT) after 3 days on FV or a 1:1 mixture of FV and cram-1 RNAi. (L) ß-amyloid staining with ThT fell ∼60% with cram-1 RNAi (P < 0.001, one-tailed t-test). (M) Impaired chemotaxis to n-butanol, in CL2355 worms expressing pan-neuronal Aß_1–42. Induced worms (M, N, P), RNAi-treated from the L3/L4 molt, were upshifted 48 h later. Chemotaxis (%) was scored after 0.5 or 2 h. Uninduced worms (O) were fed RNAi from hatch. Chemotaxis declined between d5 (solid bars) and d7 (hatched bars). (P) Worms were fed dual RNAi bacteria (1:1) as indicated. (F–H, J, M–P) Error bars show standard errors of proportions. Key and legend: unadjusted chi-squared P values are shown, N = 50–200/group. Similar results were obtained in repeats for each panel.

Figure 3

Immunofluorescence detection of cram-1 RNAi effects on wild-type worms and worms expressing Q40::YFP or Aβ_1–42 in muscle. (A–D) Adult Caenorhabditis elegans; scale bars are 0.1 mm. (A) N2 (wild-type) worms were fed cram-1 RNAi or FV, from hatch. Adults (4 d_PH) were immunostained with primary antibody to ubiquitin or CRAM-1. (B, C) AM141 (5 day) worms were imaged by YFP fluorescence and fluor-tagged antibody to ubiquitin (B) or proteasomes (C). (D) CL4176 worms, induced at 2 d_PH, were imaged 40 h later with antibodies to Aβ_1–42 and proteasomes. Channel crossover was < 5%; fluorescence was reduced > 85% in controls w/o primary antibody.

Figure 4

Western blot analyses of aggregated proteins from worms expressing Aβ_1–42 or Q40::YFP. (A–D) AM141 worms, expressing Q40::YFP in muscle, were fed cram-1 dsRNA or FV bacteria. Aggregates were isolated and fractions resolved in gradient gel lanes (8–12%, Invitrogen). Proteins were stained with SYPRO Ruby (A), or electro-blotted to nylon membranes and probed with antibodies raised to ubiquitin (B), CRAM-1 synthetic peptides (C), or GFP/YFP (D), followed by biotinylated 2nd antibody to IgG. HRP-streptavidin was bound to biotin, and imaged by chemiluminescence (Western Blot Kit, Pierce). (E–G) CL4176 worms, expressing human Aβ_1–42 in muscle, were fed cram-1RNAi or FV as above. Aggregate proteins were analyzed as above. (E) SYPRO Ruby-stained proteins or fractions, separated on 4–20% gradient gels (Bio-Rad). Small proteins such as Aβ_1–42 were resolved on 16% gels (panels F, G). After blotting, nylon filters were probed with antibodies to ubiquitin (F) or Aβ_1–42 (G) and detected as above. Labeled size standards are shown in panels F and G. Triangles (right) indicate expected positions for di-, tri-, tetra-, and penta-ubiquitinated Aβ_1–42.

Figure 5

Molecular dynamic modeling of CRAM-1 interactions with other proteins. (A) Ribbon structure model of CRAM-1 alone or interacting with di-ubiquitin created with modeller 9.12 and viewed with vmd software. Interaction energy change ΔE (B) and binding energy change ΔG (C) were calculated from molecular dynamic simulations of ≥ 10 ns, under gromacs, for mono-, di-, tri-, and tetra-ubiquitin interacting with CRAM-1 (blue bars) or its human orthologs SERF1 and 2 (green bars). (D) (Top panels) Radius of gyration (Rg) calculated for tetra-ubiquitin (UBQ4) interacting with p62 (sequestosome-1) or with p62+CRAM-1. Blue ovals show regions of stable behavior for the first simulation, which becomes chaotic when CRAM-1 is added. (Lower panels) Root-mean-square distance between molecular centers of mass (RMSD, an index of structural dispersion), for UBQ4 interacting with p62, or p62+CRAM-1. Gold arrows show intermolecular distance narrowing for UBQ4+p62 interaction, but increasing for UBQ4+p62+CRAM-1.

Figure 6

RNAi to cram-1 extends lifespan and lowers fecundity. (A) Images of worms expressing mCherry::ubiquitin (upper panels) and Q82::GFP (lower) in body wall muscle, at 3 d_PH (adult day 1). Worms were fed FV bacteria (left panels), or cram-1 siRNA bacteria (right panels). (B) Red fluorescence (mCherry-ubiquitin) was quantitated by ImageJ (http://fiji.sc/Fiji), from images as in A, 10–12 worms per group, at 3–4 d_PH (adult days 1–2). *Each two-tailed t-test P < 6E-6. The mean decline in fluorescence, at 3 d_PH (two repeats), was 46%. (C) Lifespan data for wild-type Caenorhabditis elegans fed from the L4/adult molt on FV or cram-1-dsRNA bacteria. Worms were transferred to fresh plates daily for 8 day to remove progeny and scored at 1- or 2-day intervals for movement (spontaneous or after gentle prodding). In two independent experiments (35 worms/group), cram-1 RNAi extended mean lifespan 11–12% (each P < 0.001, Gehans–Wilcoxon test). (D) Worm dimensions were measured at the indicated stages (larval, L1–L4; young adult, YA) and times (x-axis, hours after egg isolation), using WormSizer, a Fiji plug-in (http://fiji.sc/Fiji). Lengths (shown) and widths (not shown) did not differ significantly at any time, between worms receiving cram-1 dsRNA or FV control bacteria, in three experiments. Error bars, often smaller than symbols, show ± SEM. (E) Fertility data for worms maintained as in (D). Parents were moved to fresh plates at 24-h intervals; L2 larvae were counted 24 h later for 10–14 plates per group. Similar results were obtained in replicate experiments. Significance by two-tailed t-tests, cram-1 KD vs. FV control: *P < 2E-3; **P < 4E-4; ***P < 2E-5.