The expression signature of genes for flies expressing Aβ_1–42 at day 3 (top part) and day 8 (bottom part) of adult life was determined and analysed for over-representation of genes belonging to particular molecular functions. Only functional groups that are significantly over-represented in the set of differentially regulated genes (P < 0.01) are depicted. The bars represent the ratio of enrichment (R), calculated as the ratio between the observed (O) number of genes belonging to a molecular function and the expected number of genes belonging to that function if selected at random. Molecular functions that are clearly redox-related are represented by shaded bars; other molecular functions are represented by open bars.

Flies expressing Aβ_1–42 are more sensitive to oxidative stress and have higher levels of oxidative damage. Supplementing the fly food with 10% v/v H₂O₂ for 72:00 h (a, shaded bars) was specifically toxic to flies expressing Aβ_1–42, significantly reducing their survival as compared with flies expressing Aβ_1–40 (a, elav-Gal4 UAS-Aβ_1–42 vs. elav-Gal4 UAS-Aβ_1–40, P < 0.05). H₂O₂ treatment did not reduce the survival of Aβ_1–40 as compared with control flies (a, elav-Gal4 w¹¹¹⁸ vs. elav-Gal4 UAS-Aβ_1–40). The differences in survival at 72:00 h were not due to toxicity of the Aβs because in the absence of H₂O₂ there was no difference in survival between any of the groups (open bars). Flies expressing Aβ_1–42, but not flies expressing Aβ_1–40 or control flies, exhibited markers of oxidative stress. Measurement of carbonyl groups (nm/mg head protein extract) demonstrated that flies expressing Aβ_1–42 had a significantly higher carbonyl load than control flies (b, elav-Gal4 w¹¹¹⁸ vs. elav-Gal4 UAS-Aβ_1–42, P < 0.05). In flies expressing Aβ_1–40 the carbonyl load was not significantly increased (b, elav-Gal4 UAS-Aβ_1–40 vs. elav-Gal4 w¹¹¹⁸) above control levels. The data are representative of three independent Aβ₄₂ transgenic lines. Independent Student’s t-test result (*P < 0.05, **P < 0.01).

Over-expressing oxidative stress-related genes modified the Aβ-induced locomotor phenotype. Flies expressing Arctic Aβ_1–42 (a, UAS-Arctic Aβ_1–42) exhibited a progressive decline in locomotor function becoming immobile by day 30. Control flies (elav-GAL4) showed a markedly slower decline in locomotor function (a–d). Co-expression of ferritin heavy (a, UAS-Arctic Aβ_1–42UAS-Fer1HC) and light (a, UAS-Arctic Aβ_1–42UAS-Fer2LC) chains significantly improved locomotor function. The locomotor effects of the canonical antioxidant genes were in accord with their effects on longevity; CAT (b, UAS-Arctic Aβ_1–42UAS-CAT) and mitSOD2 (b, UAS-Arctic Aβ_1–42UAS-mitSOD2) improved climbing, whereas SOD1 (b, UAS-Arctic Aβ_1–42UAS-SOD1) accelerated the decline. The antioxidant genes had similar effects when co-expressed with wild-type Aβ_1–42 with co-expression of ferritin heavy chain (c, UAS-Aβ_1–42UAS-Fer1HC) and CAT (c, UAS-Aβ_1–42UAS-CAT) improving locomotor function and SOD1 (c, UAS-Aβ_1–42UAS-SOD1) accelerating the decline in locomotor function. Co-expression of GST (d, UAS-Arctic Aβ_1–42UAS-GST) gave a weak rescue of the locomotor deficits and carbonyl reductase (d, UAS-Arctic Aβ_1–42UAS-SNI) did not significantly improve locomotor function. In control experiments the expression of the key modifiers of Aβ toxicity (FerHC, FerLC and CAT) in the absence of Aβ did not have a beneficial effect on locomotor function at any time-point (supporting Fig. S2) (independent Student’s t-test, *P < 0.05, **P < 0.01, ***P < 0.001).

Oxidative stress mediated by the Arctic Aβ_1–42 is reversed by ferritin despite an increase in the levels of Aβ_1–42. A reduction in carbonyl load was apparent with co-expression of the light chain of ferritin (a, UAS-FerLC elav-Gal4 UAS-Arctic Aβ_1–42, P < 0.05); however, the effect was more marked following over-expression of the heavy chain of ferritin (a, UAS-FerHC elav-Gal4 UAS-Arctic Aβ_1–42, P < 0.01). In contrast, co-expression of SOD1 (a, UAS-SOD1 elav-Gal4 UAS-Arctic Aβ_1–42, P < 0.01) significantly increased the oxidative damage as compared with flies expressing Arctic Aβ_1–42 alone (a, control elav-Gal4 UAS-Arctic Aβ_1–42) (**P < 0.01, ***P < 0.001). The concentration of Aβ_1–42 in the heads of flies was determined for three independent biological replicates for each line of flies. Co-expression of both the heavy and light chains of ferritin results in significantly increased levels of Aβ (b, UAS-FerLC and UAS-FerHC). In contrast, modification of SOD1 or CAT activity (b, UAS-SOD1, UAS-SOD1 RNAi and UAS-CAT) has no effect on Aβ levels. Error bars show the SEM.

The percentage increase in median survival in flies expressing Arctic Aβ_1–42 that was attributable to the GS element was calculated (filled bars). Control flies (n = 100) possessing elav-Gal4±the GS element but lacking the Arctic Aβ_1–42 transgene (n = 100) were also assessed. The percentage increase in median survival in these control flies that was attributable to the GS element was determined (empty bars). The significance of the greater increase in median survival in Arctic Aβ_1–42 vs. the effect on control flies was significant (P < 0.001) in all cases. The number of independent GS element inserts that were independently identified in the screen and analysed in this assay is indicated in white on the filled bars. Where three or more GS inserts were available in a particular class the error bars indicate the SD of the estimates. We assumed that the estimates varied normally and the significance was calculated by Student’s t-test [ferritin light (L) chain inserts, P < 0.05; ferritin heavy (H) chain inserts, P < 0.001; CG9432, P < 0.001]. DH, dehydrogenase.

Co-expression of antioxidative stress genes with Aβ increased the median survival of flies. Co-expression of both ferritin heavy (elav-Gal4 UAS-Fer1HC) and light (elav-Gal4 UAS-Fer2LC) chains prolonged the lifespan of Arctic Aβ_1–42 flies (a and b, elav-Gal4 UAS-Arctic Aβ₄₂). Flies co-expressing Arctic Aβ_1–42 and the heavy chain of ferritin exhibited survival that was similar to that of control flies (b, elav-GAL4). Similarly, co-expression of another gene from the GS screen, carbonyl reductase (a, elav-Gal4 UAS-SNI), yielded a small but significant prolongation of the lifespan of Arctic Aβ_1–42 flies. The expression of other canonical antioxidative enzymes also had significant effects on longevity. Both mitSOD2 (a, elav-Gal4 UAS-mitSOD2) and CAT (a and c, elav-Gal4 UAS-CAT) prolonged the lifespan of the flies. Surprisingly, SOD1 (a and c, elav-Gal4 UAS-SOD1) enhanced the toxicity of Arctic Aβ_1–42. In contrast, the knockdown of endogenous SOD1 protein by UAS-RNAi (a and c, elav-Gal4 UAS-IR.SOD1) protected the fly from Aβ toxicity. The enhancer effect of SOD1 appears to be mediated by its catalytic activity because a dominant negative mutant of SOD1 prolonged the lifespan of the flies expressing Arctic Aβ (a and c, SOD1ⁿ¹⁰⁸). In control experiments there was no prolongation of lifespan when SOD1, mitSOD2, CAT, carbonyl reductase, ferritin heavy chain and ferritin light chain were expressed using elav^c155-Gal4 in flies that did not carry the UAS-Arctic Aβ_1–42 transgene (d). Kaplan–Meier survival curves were plotted and statistical significance was assessed by the log rank test using the spss 11.0 statistical package. Differences shown were all statistically significant (P < 0.001). In control experiments elav-Gal4 flies had the same lifespan as the background w¹¹¹⁸ flies.

The iron-chelating compound clioquinol was added to the fly food at final concentrations of 2, 20 and 200 μm and its effects on the Aβ-associated longevity phenotype and the accumulation of iron in the fly brain was assessed. The median lifespan of flies expressing Aβ_1–40 (a, circles) was not significantly increased by clioquinol at any concentration; however, flies expressing Arctic Aβ_1–42 demonstrated a clear concentration-related increase in median survival (a, diamonds). Wild-type Aβ_1–42 flies exhibited an intermediate, but significant, response (a, triangles). Kaplan–Meier survival statistics with the log rank test were used to analyse the data (significance of difference from no-clioquinol control, **P < 0.01, ***P < 0.001, unlabelled P > 0.05). Flies expressing Arctic Aβ_1–42 (b, triangles) accumulated significantly more iron in their brains than control flies (b, circles) and this was only reversed by treatment with 200 μm clioquinol. Error bars show the SD (n = 3). The significance of the difference between clioquinol-treated and non-treated flies was calculated pairwise using the two-tailed Student’s t-test (**P < 0.01).

Model of Aβ-mediated oxidative stress. CuZn-SOD1 breaks down superoxide free radicals (O₂−) in the cytoplasm to produce H₂O₂ and molecular oxygen (O₂). In the presence of oxygen, Fe²⁺ cycles back to Fe³⁺ by the Aβ to produce H₂O₂. H₂O₂ is neutralized into water by CAT. The heavy chain of ferritin (Fer1HC) has ferroxidase activity, which catalyses the conversion of Fe²⁺ into Fe³⁺ ions. Fe³⁺ is subsequently stored by the light chain of ferritin (Fer2LC). Thus, ferritin has two effects: it prevents Fe²⁺ from interacting with the Aβ and producing H₂O₂ and it prevents Fe²⁺ from reacting with H₂O₂ and producing the free radical hydroxyl (OH^.) via the Fenton reaction. Hydroxyl radicals can also oxidize lipids to generate long-lived reactive aldehydes. Carbonyl reductase (CAR) and GST are downstream antioxidant defences that participate in the detoxification of the reactive aldehyde species.