Proteins that mediate protein aggregation and cytotoxicity distinguish Alzheimer's hippocampus from normal controls

Abstract

Neurodegenerative diseases are distinguished by characteristic protein aggregates initiated by disease-specific 'seed' proteins; however, roles of other co-aggregated proteins remain largely unexplored. Compact hippocampal aggregates were purified from Alzheimer's and control-subject pools using magnetic-bead immunoaffinity pulldowns. Their components were fractionated by electrophoretic mobility and analyzed by high-resolution proteomics. Although total detergent-insoluble aggregates from Alzheimer's and controls had similar protein content, within the fractions isolated by tau or Aβ1-42 pulldown, the protein constituents of Alzheimer-derived aggregates were more abundant, diverse, and post-translationally modified than those from controls. Tau- and Aβ-containing aggregates were distinguished by multiple components, and yet shared >90% of their protein constituents, implying similar accretion mechanisms. Alzheimer-specific protein enrichment in tau-containing aggregates was corroborated for individuals by three analyses. Five proteins inferred to co-aggregate with tau were confirmed by precise in situ methods, including proximity ligation amplification that requires co-localization within 40 nm. Nematode orthologs of 21 proteins, which showed Alzheimer-specific enrichment in tau-containing aggregates, were assessed for aggregation-promoting roles in C. elegans by RNA-interference 'knockdown'. Fifteen knockdowns (71%) rescued paralysis of worms expressing muscle Aβ, and 12 (57%) rescued chemotaxis disrupted by neuronal Aβ expression. Proteins identified in compact human aggregates, bound by antibody to total tau, were thus shown to play causal roles in aggregation based on nematode models triggered by Aβ1-42 . These observations imply shared mechanisms driving both types of aggregation, and/or aggregate-mediated cross-talk between tau and Aβ. Knowledge of protein components that promote protein accrual in diverse aggregate types implicates common mechanisms and identifies novel targets for drug intervention.

Keywords: Abeta(1-42); Alzheimer (Disease); C. elegans; acetylation (protein); aggregation (protein); beta amyloid; microtubule-associated protein tau; neurodegeneration; neurotoxicity; oxidation (protein); phosphorylation (protein); proteomics.

Figure 1

Proteins in detergent‐insoluble AD‐derived aggregates are more abundant and diverse than those from controls. (A) Aggregates insoluble in 1% sarcosyl, isolated from hippocampi of normal controls (3 subjects pooled, lanes 1, 3, and 5) or AD (pool of 3, lanes 2, 4, 6). Lanes 1 and 2: pulldown with antibody to Aβ_1–42; lanes 3 and 4: pulldown with antibody to tau; lanes 5 and 6: total large insoluble aggregates. (B) Venn diagram of proteins found to be significantly enriched (blue font) or depleted (red) in aggregates from AD relative to controls. More than half of the proteins listed in Table 1 (59/100) were significantly more abundant in both Aβ and tau pulldowns from AD than from controls.

Figure 2

P‐tau coincides with dynactin in situ. Images show hippocampus sections from Alzheimer disease (A–C, E–G) or age‐matched controls (D, H). (A–D) Conventional immunohistochemistry, visualizing proteins with fluor‐tagged secondary antibodies, after incubation with antibodies to P‐tau (Ser202, Thr205; green) and dynactin/p50 (red); nuclei (blue) were stained with DAPI. Light green and yellow (arrows) indicate superposition of red and green fluors. (E–H) Proximity ligation amplification (PLA); red product indicates close proximity (<40 nm separation) of P‐tau (Ser202, Thr205) to dynactin/p50.

Figure 3

Total tau colocalizes with 14‐3‐3 and a subset of internexin in AD hippocampus. Sections were immunostained for total tau (A, E; green), or by proximity ligation amplification (PLA) for total tau in conjunction with 14‐3‐3 proteins (B; red) or with internexin (F; red). C and G are merged images of A+B and E+F, respectively. D and H are the corresponding merged images for AMC hippocampus. Note that in H, the tau (green) exposure was increased ~10‐fold to illustrate the diffuse distribution of normally phosphorylated tau in control tissue.

Figure 4

P‐tau proximity to sequestosome‐1 and Aβ_1–42/APP. Sections were immunostained for sequestosome‐1 (SQSTM1, p62; A) or neurofilament heavy chain (NHC, D) and for PLA product (red) showing P‐tau (Thr205) proximal to SQSTM1 (B) or to Aβ_1–42 (E). C and D are merged images of A+B and D+E, respectively.

Figure 5

P‐tau co‐localizes with Aβ_1–42 oligomers. Antibodies to P‐tau [Thr205] and oligomeric Aβ_1–42 (A–D) produce red PLA product if they are within 40 nm. Counterstaining with antibody to LC3B/ATG8, a marker of autophagosomes (B, D, F), appears as green. In E and F, PLA product (red) indicates P‐tau contiguous to Aβ_1–42 oligomers. AD hippocampus sections are shown in A, B, E, and F, while C and D show AMC hippocampus.

Figure 6

Knockdown of AD‐enriched aggregate proteins rescues aggregation‐induced traits. (A) Proteins shown are overrepresented (in LC‐MS spectral counts) in aggregates isolated from AD relative to NC, except microtubule‐associated protein tau, which is hyperphosphorylated but not more abundant in AD (Table 1). Blue arrows connecting diamond symbols show increased abundance of proteins in tau‐IP aggregates; black arrows connecting triangles show increased abundance in Aβ‐IP aggregates. (B) Of 21 protein types more abundant in AD than in NC aggregates, knockdown of 14 (67%) of their nematode orthologs significantly rescued paralysis of a C. elegans strain expressing human Aβ_1‐42 in muscle (blue bars), or rescued chemotaxis disrupted by neuronal expression of Aβ_1‐42 (red bars). Significance was assessed by 1‐tailed t‐tests (as lower cytotoxicity was predicted) comparing fractions of unparalyzed or chemo‐attracted worms in three independent assays per group (*P < 0.05; **P < 0.005; ***P < 0.0005). Significance by chi‐squared tests within assays was P < 5 × 10⁻⁵ to 10⁻¹⁶.