AIP-1 ameliorates beta-amyloid peptide toxicity in a Caenorhabditis elegans Alzheimer's disease model

Abstract

Multiple neurodegenerative diseases are causally linked to aggregation-prone proteins. Cellular mechanisms involving protein turnover may be key defense mechanisms against aggregating protein disorders. We have used a transgenic Caenorhabditis elegans Alzheimer's disease model to identify cellular responses to proteotoxicity resulting from expression of the human beta amyloid peptide (Abeta). We show up-regulation of aip-1 in Abeta-expressing animals. Mammalian homologues of AIP-1 have been shown to associate with, and regulate the function of, the 26S proteasome, leading us to hypothesize that induction of AIP-1 may be a protective cellular response directed toward modulating proteasomal function in response to toxic protein aggregation. Using our transgenic model, we show that overexpression of AIP-1 protected against, while RNAi knockdown of AIP-1 exacerbated, Abeta toxicity. AIP-1 overexpression also reduced accumulation of Abeta in this model, which is consistent with AIP-1 enhancing protein degradation. Transgenic expression of one of the two human aip-1 homologues (AIRAPL), but not the other (AIRAP), suppressed Abeta toxicity in C. elegans, which advocates the biological relevance of the data to human biology. Interestingly, AIRAPL and AIP-1 contain a predicted farnesylation site, which is absent from AIRAP. This farnesylation site was shown by others to be essential for an AIP-1 prolongevity function. Consistent with this, we show that an AIP-1 mutant lacking the predicted farnesylation site failed to protect against Abeta toxicity. Our results implicate AIP-1 in the regulation of protein turnover and protection against Abeta toxicity and point at AIRAPL as the functional mammalian homologue of AIP-1.

Figure 1.

Expression of the human amyloid-β₄₂ leads to an increase in aip-1 transcript accumulation. (A) Induction of aip-1 shown using GeneChip microarrays (Affymetrix, Inc.). aip-1 levels in an Aβ-expressing strain, CL4176, were compared to a control strain expressing GFP. In these strains, Aβ₄₂ or GFP were expressed at low levels at 16°C, while moving the animals to 25°C induced high levels of transgene expression (see Materials and Methods for details of temperature inducibility). Worms were harvested at the time of temperature change (T₀) and every 4 h afterwards until 20 h later (T₄–T₂₀). Error bars represent the SEM. (B) Microarray data were confirmed using real-time RT–PCR to measure aip-1 levels at the T₁₆ time point. Data were plotted as a fold change compared with the GFP control strain. Error bars represent the SEM. (C) Aβ and aip-1 expression were correlated in an aip-1/GFP transcriptional reporter strain. GFP fluorescence was detected in the pharynx at 16 and at 25°C (narrow black arrows), which is consistent with constitutive pharyngeal expression. In body wall muscle, however, GFP fluorescence was only detectable at 25°C (wide white arrows), which is consistent with Aβ₄₂-stimulated induction of aip-1 promoter. Size bar = 100 µm.

Figure 2.

AIP-1 protects against Aβ toxicity in a C. elegans model. Graphs shown here depict the progression of paralysis of Aβ-expressing animals after temperature induction. Animals were grown for 36 h at 16°C to keep Aβ expression at low levels, not enough to induce paralysis. Animals were then moved to 25°C to induce high levels of Aβ expression and paralysis. The horizontal axis represents the number of hours the animals spent at 25°C. (A) The paralysis phenotype was delayed in animals overexpressing aip-1 from a transgenic extra-chromosomal array under the control of either myo-3 or (B) aip-1 promoter (red curves). Blue curves represent the progression of paralysis in animals that lost the extra-chromosomal array and are thus carrying only the chromosomal copies of aip-1. (C) A negative control expressing GFP under the control of an aip-1 promoter shows no effect on paralysis, which argues against a non-specific transgene effect. (D) The effect of aip-1 overexpression on the paralysis phenotype is reversed by aip-1-specific RNAi, which is consistent with an AIP-1-dependent effect. (E) aip-1 knockdown by RNAi exacerbates the paralysis phenotype. The control RNAi curve represents animals fed E. coli (strain: HT115) carrying pL4440 vector, which only contains the multiple-cloning site between the two convergent T7 polymerase promoters (11) and was used as a negative control. Error bars represent the SEM. The P-values shown were obtained using a two-way ANOVA.

Figure 3.

aip-1 overexpression results in decreased accumulation of Aβ₄₂ peptide. Aβ levels were compared between animals overexpressing aip-1 and animals expressing wild-type levels of aip-1 by western blotting using the anti-Aβ antibody 6E10. Migration of monomeric Aβ was demonstrated by a purified monomeric peptide (lane 1). Aβ-specific bands were determined by comparison to a control strain (N2) lacking Aβ transgene (lane 4). A dramatic decrease in Aβ levels is shown in animals overexpressing aip-1 compared with a control strain (lanes 2 and 3, respectively).

Figure 4.

AIP-1 reduces the accumulation of GFP::degron, an aggregation-prone variant of GFP, in body wall and vulval muscles. (A) Animals carrying the integrated array dvIs38 (worm strain CL2337), which contains GFP::degron under the control of a myo-3 promoter and a long 3′-UTR required for temperature inducibility, accumulate GFP aggregates in their body wall muscles and suffer a paralysis phenotype similar to that induced by Aβ. (B) Transgenic overexpression of aip-1 reduces GFP::degron accumulation. (C) Reduction of aip-1 expression by RNAi causes increased accumulation of GFP::degron compared with animals treated with control RNAi. (D) aip-1 RNAi partially reverses the decreased GFP::degron accumulation caused by transgenic overexpression of aip-1. Size bar = 150 µm.

Figure 5.

An AIP-1 human homologue, AIRAPL, has a protective effect against Aβ₄₂ toxicity and AIP-1 putative farnesylation site is essential for its protective effect against Aβ₄₂ toxicity. (A) The paralysis phenotype of Aβ₄₂-expressing animals is delayed by human AIRAPL overexpression. (B) Human AIRAP overexpression had no effect on the paralysis phenotype of Aβ₄₂-expressing animals. (C) Deleting the C-terminal two residues in AIP-1 disrupts its putative isoprenylation site and abolishes its protective effect. Error bars represent the SEM. The P-values shown were obtained using a two-way ANOVA.