Curcumin Promotes A-beta Fibrillation and Reduces Neurotoxicity in Transgenic Drosophila

The pathology of Alzheimer's disease (AD) is characterized by the presence of extracellular deposits of misfolded and aggregated amyloid-β (Aβ) peptide and intraneuronal accumulation of tangles comprised of hyperphosphorylated Tau protein. For several years, the natural compound curcumin has been proposed to be a candidate for enhanced clearance of toxic Aβ amyloid. In this study we have studied the potency of feeding curcumin as a drug candidate to alleviate Aβ toxicity in transgenic Drosophila. The longevity as well as the locomotor activity of five different AD model genotypes, measured relative to a control line, showed up to 75% improved lifespan and activity for curcumin fed flies. In contrast to the majority of studies of curcumin effects on amyloid we did not observe any decrease in the amount of Aβ deposition following curcumin treatment. Conformation-dependent spectra from p-FTAA, a luminescent conjugated oligothiophene bound to Aβ deposits in different Drosophila genotypes over time, indicated accelerated pre-fibrillar to fibril conversion of Aβ1–42 in curcumin treated flies. This finding was supported by in vitro fibrillation assays of recombinant Aβ1–42. Our study shows that curcumin promotes amyloid fibril conversion by reducing the pre-fibrillar/oligomeric species of Aβ, resulting in a reduced neurotoxicity in Drosophila.


Introduction
Drosophila melanogaster (Drosophila) has during the past decade emerged as a promising model system for Alzheimer's disease (AD) research [1,2,3]. Expression of various AD-related human proteins e.g., amyloid-b (Ab) and Tau, in the Drosophila model system results in animals displaying many of the histological hallmarks of AD seen in humans [4], as well as a correlation of lifespan to aggregation propensity of the protein or peptide expressed [5] which surpasses that of corresponding rodent models. Drosophila Ab models have also been used as platforms for pharmacological treatment assays by the putative aggregation inhibitor Congo red [3], designed native state stabilizers [6], and by genetic manipulations co-expressing molecular chaperones [7].
For several years, curcumin has been proposed to be a candidate drug for enhanced clearance of the toxic amyloids generated by the Ab peptide. Curcumin has also been reported to have anti-inflammatory and anti-oxidative activity thus preventing tissue damage [8,9,10,11], to reduce Ab-induced toxicity [12], as well as to reduce microglia activation [13]. Many studies, in vitro and in vivo, have shown inhibition of Ab oligomer [9] and fibril formation [9,14,15,16] in the presence of curcumin, depending on the used assay. Curcumin appears to be a promiscuous AD-drug as it has also been shown to block Ab toxicity in vivo through inhibition of Tau phosphorylation [11,17], as well as regulating AbPP and BACE-1 transcription by interfering with copper ions [18] in cell cultures. Because of these promising results curcumin has been tested in humans as a drug candidate for AD [19]. The dominating view is that curcumin is an aggregation inhibitor, but recent studies of curcumin have shown that while it inhibits the oligomeric forms of Ab, it accelerates the formation of Ab fibrils [20]. Curcumin is a potent binder of amyloid deposits, making it a molecular candidate for histological staining in pathology [9,21,22], and it has been suggested as a useful derivate in combination with near-infrared imaging in living subjects [23]. In addition to these results, curcumin has also shown effects, mainly due to its anti-inflammatory properties, on other diseases such as rheumatoid arthritis, pancreatitis, cancer, osteoarthritis, and in some ocular as well as gastrointestinal conditions, such as ulcerative colitis [24]. Current drug development strategies include development of curcumin analogues with similar biological activity as curcumin, but with improved pharmacokinetic characteristics, including increased bioavailability and water solubility [25]. Amyloidogenic proteins are known to be able to assemble into multiple forms of amyloid-like structures in vitro in a process known as ''amyloid fibril polymorphism'' [26,27] suggesting that several mechanisms may be involved in amyloid formation. Multiple fibrillation pathways will probably result in multiple forms of pre-fibrillary species, responsible for toxicity. The connection between these complex folding processes and curcumin is not clear.
Here, we have addressed the potency of curcumin in alleviating AD-like symptoms in transgenic Drosophila models of AD. To this end we used four different Ab expressing Drosophila lines (Ab 1-40 single transgene, Ab 1-42 single transgene, Ab 1-42 double transgene, and Ab 1-42 E22G single transgene) [3], and one human Tau expressing Drosophila line [1], all expressed in the Drosophila central nervous system (CNS) and eyes by the Gal4/UAS system [28]. The longevity and the locomotor activity of the different genotypes were measured relative to a control line. To detect amyloid formation ex vivo, we combined antibody staining with a luminescent conjugated oligothiophene, p-FTAA [29], as a marker for amyloid, and collected structural dependent spectra from the probe for different time points as the aggregation accelerates in the tissue. These results, combined with in vitro fibrillation of recombinant Ab and quantification of the soluble and insoluble Ab produced in the flies in absence or presence of curcumin, shows that curcumin does not inhibit amyloid formation. On the contrary, curcumin rather accelerates amyloid fibril conversion, and hence reduce the pre-fibrillary species of Ab. This suggests that the reduction of pre-fibrillar species underlies the observed mitigated neurotoxicity in Drosophila.

Drosophila longevity
To address the effects of curcumin upon AD-related symptoms in Drosophila, we initially fed various concentrations of curcumin (0-0.01% w/w in yeast paste) to the control flies. Rather unexpectedly, this however rendered a reduced lifespan. The reduced viability was curcumin concentration dependent, resulting in a decreased median survival time (T 1/2 ) ( Figure 1A). The Ab 1-40 expressing flies showed a decreased T 1/2 compared to the control flies, and was only affected by the higher curcumin concentrations, also here resulting in a decreased life span. For the low concentration of curcumin, the lifespan was unaffected ( Figure 1B). Curcumin feeding of the single inserted Ab 1-42 expressing flies showed a positive effect of the T 1/2 at low and intermediate curcumin concentrations and no effect of the T 1/2 on the highest curcumin concentration ( Figure 1C). Curcumin feeding of the double inserted Ab 1-42 expressing flies showed a positive effect at low and intermediate concentrations of curcumin treatments. The lifespan for high curcumin concentration treated flies was not apparently different when compared to the untreated flies ( Figure 1D). All curcumin treatments of the Ab 1-42 E22G expressing flies showed a substantial positive effect upon curcumin treatment. The greatest observed effect was found on 0.001% curcumin treatment of the Ab 1-42 E22G expressing flies, which increased the T 1/2 by 75% compared to untreated flies ( Figure 1E). Curcumin feeding of the Tau expressing flies rendered no effect on survival at low concentrations, but a toxic effect on high concentrations of curcumin ( Figure 1F). The median survival time ( Figure 1G) of all transgenes and curcumin concentrations displayed a clear effect upon curcumin treatment for genotypes having the strongest phenotype. The toxic effect on curcumin treatment was balanced with the rescuing effect of genotypes having a mild phenotype. The T 1/2 for all genotypes and curcumin concentrations are summarized in Table S1. The results from the longevity assay were essentially mirrored in a conventional climbing assay with a reduced climbing shifted of 1-2 days prior to death ( Figure S1) as reported in previous studies [30].

Drosophila locomotor activity
The DAM2 system automatically counts the number of beam breaks for flies walking in a horizontal tube over a period of 24 hours [31]. This setup allowed for characterization of the locomotor and behavior rhythms of Drosophila. Comparing the locomotor activity of the flies without curcumin treatment at day 5 (c.f. Figure 2A, E, I, M, Q, U), there were obvious differences in the activity and circadian rhythm between the different genotypes. The DAM2 system hence appeared more sensitive for assaying early signs of neurological impairment compared to the conventional climbing assay (cf. Figure 2 with Figure S1). Studies of continuous curcumin treatment of flies at the intermediate concentration (0.001%) was performed at different days of aging, with 5 days increments. Control flies showed slight activity deterioration upon curcumin treatment. Overall, the locomotor activity of the flies decreased with increasing age, but the number of beam breaks per hour was almost consistent during the first hours of the assay. The decreased total number of beam brakes upon increased age was caused by the shortened number of active hours (Figure 2A-D). Ab 1-40 expressing flies showed no activity improvement upon curcumin treatment ( Figure 2E-H). Single insert Ab 1-42 expressing flies showed a higher number of beam breaks and a continuation of activity during a larger number of hours upon curcumin treatment for flies at 5 and 10 days ( Figure 2I-L). The effect of curcumin showed a tendency of decreasing with age ( Figure 2L). Double insert Ab 1-42 expressing flies showed an activity enhancement upon curcumin treatment for flies of all ages ( Figure 2M-P). Also here the effect of curcumin showed a tendency of declining with age ( Figure 2P). The Ab 1-42 E22G expressing flies showed a severely decreased locomotor activity already at day 5 compared to control flies (c.f. Figure 2A and 2Q). Curcumin treated Ab 1-42 E22G expressing flies at day 5 showed an increased number of beam breaks per hour during the first hours, but the number of active hours was not significantly enhanced ( Figure 2Q). A small but significant increase in activity was observed for 10 days old Ab 1-42 E22G expressing flies treated with curcumin ( Figure 2R). The activity assay was performed with the unusual population of the Ab 1-42 E22G expressing flies surviving two days longer than T 1/2 of the untreated Ab 1-42 E22G expressing flies. No activity assay was performed for Ab 1-42 E22G expressing flies at ages beyond 10 days, due to their short lifespan. Tau expressing flies appear severely affected in their locomotor activity, and interestingly showed a large enhanced activity during the first hours upon curcumin treatment ( Figure 2U-X). The activity enhancement was sustained with increasing age, which was different from the Ab transgenes ( Figure 2X). Notably, the Tau expressing flies showed the greatest increase in locomotor activity of all genotypes upon curcumin treatment, but as described above, these flies did not show any significant effect in the lifespan or in climbing behavior upon curcumin treatment (Figure 1 and. Figure S1).
Histological analysis of amyloid deposits in the Drosophila brain over time Drosophila brains were investigated by fluorescence microscopy using combined antibody and amyloid specific luminescent conjugated oligothiophene (LCO), p-FTAA, for the presence of amyloid deposition [29]. Images shown in Figure 3 and Figure S2 represent parts of brains where protein deposition was obvious. The histological analysis described below was based on visual inspection of 20 flies of each genotype and is summarized in Table 1 before the median life span (T 1/2 ) for each genotype. Control flies with and without curcumin, displayed no antibody or LCO binding in the brain tissue ( Figure 3A-E), but unspecific staining was seen in the retina and exoskeletal head capsule, as described before [32]. However, the double inserted Ab 1-42 expressing flies showed extensive amyloid staining, with several long extended fibrillar aggregates found already in newly eclosed flies ( Figure 3F). The amounts of aggregates visible by LCO staining increased with increasing age ( Figure 3F, G, I). Some staining differences was observed in ten day old flies, where the untreated flies displayed stronger antibody staining and decreased LCO staining compared to treated flies ( Figure 3G, H, Table 1). The staining pattern from ten-day-old curcumin treated flies did not differ from twenty-day-old flies, independent of treatment ( Figure 3G-J). DAPI staining from regions with a high load of LCO-positive aggregates was decreased, suggesting neuronal loss. The Ab 1-42 E22G expressing flies displayed a spot-like staining from both LCO and Ab antibody independent of age and treatment, but a weaker LCO staining than that displayed for wild type single and double inserted Ab 1-42 expressing flies ( Figure 3K-O). The with no curcumin added represented in black bars, 1, 10, and 100 mg curcumin per g yeast paste represented in yellow, orange, and red bars respectively. Significance was compared using a paired student's t-test between untreated and treated with curcumin (ns: non-significant; *: P,0.05, **: P,0.01; ***: P,0.001). doi:10.1371/journal.pone.0031424.g001  Figure S2A-E) displayed weak and restricted LCO and antibody amyloid staining surrounding the nuclei in twenty-day-old flies. Amyloid staining in the Ab 1-40 expressing flies has never earlier been reported. At earlier time points, LCO amyloid staining was only shown to a minute extent, but antibody staining was detected in the areas with high amounts of cell nuclei (Table 1). There was no difference between the untreated and curcumin treated flies. Single insert Ab 1-42 expressing flies ( Figure S2F) showed antibody staining, as well as some LCO staining, in both newly eclosed flies and in untreated ten day old flies. Interestingly, ten-day-old flies treated with curcumin displayed a different staining pattern than the untreated flies, where LCO positive aggregates were found in most regions of the brain that contained cell bodies. The intensity of the LCO staining appeared enhanced in the curcumin treated flies ( Figure S2G, H). In twenty-day-old flies the difference appeared to be retained between untreated and curcumin treated flies. Both untreated and curcumin treated flies had a strong amyloid staining, predominantly surrounding the nuclei, but regions of large aggregates were more commonly observed in curcumin treated flies than in untreated flies ( Figure S2I, J, Table 1). The Tau expressing flies showed few LCO and antibody-binding aggregates in young flies ( Figure S2K-M). At the age of 15 days, LCO binding aggregates were detected in approximately one out of five flies. There was no difference in staining upon curcumin treatment ( Figure S2N, O, Table 1). The antibody staining increased between day 0 and day 10, but decreased in day 15 brains. The aggregates were mostly found in regions were no or few cell nuclei were visible. The visual inspection of brain histology rendering assessment of LCO and antibody reactivity respectively curcumin decreased with increasing age. (M-P) Double insert Ab 1-42 expressing flies showed an activity enhancement upon curcumin treatment, but the effect of curcumin decreased with increasing age. (Q-T) Ab 1-42 E22G expressing flies showed an increased number of beam breaks during the initial hours of monitoring during curcumin treatment, but the number of active hours was not prolonged for 5 days old flies. A small curcumin induced increase in activity was observed for 10 day old flies. (U-X) Tau expressing flies, at all ages, showed a great enhancement of locomotor activity during the first hours of experiments upon curcumin treatment. Significance was compared using a paired student's t-test between untreated and treated with curcumin (ns: non-significant; *: P,0.05, **: P,0.01; ***: P,0.001). doi:10.1371/journal.pone.0031424.g002  Figure 3G and H). The amount of detectable aggregates increased with age. DAPI staining from regions with widespread p-FTAA-positive aggregates were decreased. (K-O) Ab 1-42 E22G expressing flies at day 0, day 5, and day 10, showed a spot-like staining from both p-FTAA and the Ab specific antibody, but a weaker p-FTAA staining was apparent (c.f. is summarized in Table 1, before the T 1/2 of lifespan for each genotype.

Spectral analysis of amyloid formation in Drosophila over time
Sections stained with the LCO, p-FTAA [29], were analyzed by hyperspectral imaging. As expected from previous imaging experiments in Drosophila models, the p-FTAA emission spectrum of compact plaques differed depending on genotype [32]. We herein assessed if curcumin treatment rendered a conformational difference of the amyloid deposits within the Drosophila transgenics. To this end we employed spectral mixing and quantification of the fluorescence intensities at 508 nm (excitation at 405 nm) and 612 nm (excitation at 560 nm) ( Figure S3) as described previously in detail for transgenic APP mice [33]. We used this parameter as an amyloid fibrillation index. A high ratio of 508/612 nm was interpreted as a well structured amyloid fibril whereas a low ratio was interpreted as alternative conformational states of the aggregates. A spectral shift over time, in presence or absence of curcumin treatment (0.001%) was clearly observable for the double inserted Ab 1-42 expressing flies ( Figure 4A and 4B). Newly eclosed flies had few aggregates detectable with p-FTAA. The aggregates had variable emission spectra, resulting in a wide variation in the amyloid fibrillation index (508/612 ratio values). Some aggregates with the distinct amyloid feature were also detected already at day 0 ( Figure 4A, triangles). The Ab aggregates within the double insert Ab 1-42 expressing flies displayed a higher amyloid fibrillation index, after ten days of curcumin treatment, indicating that curcumin promoted a more rapid conversion into more well organized fibrils. Untreated flies did not display the increased amyloid fibrillation index at day 10. At day 20, on the other hand, there were no differences in the amyloid fibrillation index between untreated and treated flies. Both groups of flies displayed a high amyloid fibrillation index indicating well organized Ab-fibrils. An extended amyloid fibrillation index in two dimensions (employing also the 543 nm peak) was also employed for the double inserted Ab 1-42 expressing flies. This analysis further supported the notion that curcumin treatment promoted amyloid fibrillation into well organized structures ( Figure S4). Control experiments, performing spectral imaging of curcumin treated flies without p-FTAA staining, did not result in any spectral images that were above noise level ( Figure S5). Control experiments, where curcumin was used as a histological labeling agent, did not result in any significant fluorescence in the spectral imaging ( Figure S5), ruling out that the spectral imaging results were influenced by curcumin fluorescence.
Protein aggregates within aged Ab 1-42 E22G expressing flies was previously shown to display a slightly more red shifted emission spectrum compared to wild type double insert Ab 1-42 expressing flies [32], as well as in aged mice [33]. This was evident also using the double excitation hyper spectral imaging method employed above (c.f. 10 day Ab 1-42 E22G expressing flies and 20 day wild type flies, Fig. 4A and 4C). The emission spectra of the Ab 1-42 E22G expressing flies differed from the double insert Ab 1-42 expressing flies. The wild type double insert Ab 1-42 expressing flies showed a clear double peak of 508 nm and 543 nm, whereas the Ab 1-42 E22G expressing flies only had a shoulder peak at 508 nm for late stage aggregates [32]. Because the peak at 508 nm increased for double insert Ab 1-42 expressing flies with increased age and increased fibril formation, it indicates that the Ab 1-42 E22G expressing flies do not produce as well ordered aggregates as the wild type double insert Ab 1-42 expressing flies. Usually the spectra of the Ab 1-42 E22G expressing flies was red shifted compared to the double insert Ab 1-42 expressing flies, and displayed the shoulder peak at 510 nm and the distinct second peak at 545 nm. Interestingly, the Ab 1-42 E22G expressing flies showed an elevated amyloid fibrillation index in 5 day old flies ( Figure 4C and D) compared to the amyloid fibrillation index at either 0 or 10 day old flies. Even so, no extended fibrous deposits were visible in the histological staining of Ab 1-42 E22G expressing flies ( Figure 3K-O). Both curcumin treated and untreated flies displayed indifferently the elevated Ab-fibrillation index at day 5. At day 10 the amyloid fibrillation index was lower for untreated flies as in the case for wild type expressing flies at day 10 ( Fig. 4C, D), albeit with smaller absolute differences than for the double inserted Ab 1-42 expressed flies at this age.
The Tau expressing flies showed significantly more red shifted emission spectra than either Ab genotypes as reported before [32]  ( Figure S6). The red shifted spectra resulted in consistently lower amyloid fibrillation index (508/612 nm ratios) at both analyzed time points ( Figure 4E and F). Although the same trend was found as for the single inserted Ab 1-42 expressing flies no significant spectral difference was observed for the Tau expressing flies upon curcumin treatment. Due to low detection rate and small aggregates, resulting in low spectral intensities, no spectra could be analyzed in Tau expressing flies at day 0.

Quantification of Ab-peptide in aged and curcumin treated Drosophila
The concentration of Ab is the primary factor determining aggregation rates and likely also influences Ab conformation  Table S3, Table S4). For the Ab expressing flies the amount of total Ab showed no significant difference between the untreated spectra, resulting in a wide variation in the 508/ 612 ratios. Some aggregates with the distinct amyloid feature were also detected (marked with triangles in the graph) in young flies. The double insert Ab 1-42 expressing flies displays a higher amyloid fibrillation index after ten days of curcumin treatment, indicating a more organized fibril formation due to curcumin ingestion. Untreated flies did not display an increased Ab-fibrillation index at day 10. At day 20, the overall amyloid fibrillation index had increased, but there were no differences in the amyloid-fibrillation index between untreated and treated flies. (C and D) The Ab 1-42 E22G expressing flies showed a rather high amyloid fibrillation index already in newly eclosed flies. Both treated and untreated flies have higher Ab-fibrillation index at day 5, but at day 10 the Ab-fibrillation index was lower, especially for untreated flies. (E and F) The Tau expressing flies showed more red shifted emission spectra than the Ab genotypes, with a higher contribution from the 612 nm emission (560 nm excitation) ( Figure S3), resulting in a lower amyloid fibrillation index. Note the different scale of the y-axis compared to A-D. A few aggregates with elevated amyloid fibrillation index were detected as marked by triangles in the graph. Overall the trend was the same as for Ab expressing flies but the statistical analysis revealed that no significant differences were observed for the Tau expressing flies upon curcumin treatment. No spectra could be analyzed in Tau expressing flies at day 0, due to low detection rate and minute aggregates, resulting in low spectral intensities. (ns: non-significant; *: P,0.05, **: P,0.01; ***: P,0.001). doi:10.1371/journal.pone.0031424.g004 flies and the curcumin treated flies at either time point. Moreover, the fraction of soluble Ab was not altered by curcumin treatment. Notably, the fraction of soluble Ab was below 5% of the total Ab for all genotypes at all time points. We hence concluded that treatment with curcumin does not shift the total amount of Ab or the distribution of soluble versus insoluble Ab. This finding supports the observation that curcumin does not decrease Ab deposition. It is important to note that any increase in insoluble Ab due to treatment would be difficult to detect with this assay, because .95% of Ab was insoluble in untreated flies.

Direct evidence for Ab 1-42 binding to curcumin in vitro
It is well established that curcumin in solution at neutral pH is rapidly degraded [34], and this can also be observed by ocular inspection of an aqueous curcumin solution (not shown). In our formulation of curcumin in yeast paste for the fly food we did not note (by eye) a decrease in yellow coloration over time. We also noted that in the presence of Ab 1-42 , curcumin solutions remained yellow over time. This is in contrast to observations of curcumin dissolved alone in PBS [34]. An absorbance assay was performed at 29uC to quantify this effect. In the absence of Ab 1-42 we observed a rapid decline in absorbance at 440 nm over time, with a half time of 1.5-3 h, depending upon curcumin concentration ( Figure 6A-C). In the presence of Ab 1-42 this decline in absorbance was abolished at curcumin concentrations below that of Ab  . In addition, we observed a noticeable red shifted absorbance spectrum of curcumin in the presence of excess Ab 1-42 , supporting the notion of sequestered curcumin molecules in a hydrophobic environment ( Figure 6D, E). Importantly, these results substantiate two important findings: i) Ab 1-42 directly binds curcumin (evident by the spectrochromic absorbance shift) and ii) Ab 1-42 sequesters curcumin from degradation (evident by the preserved curcumin absorbance over time).
In vitro aggregation of Ab 1-42 in the presence of curcumin Next, to compare our data with previous reports (e.g. [9]), we performed in vitro studies to address the effect of curcumin on the molecular aggregation behavior of Ab 1-42 . Samples were assayed using native PAGE Western blotting, transmission electron microscopy (TEM), and p-FTAA fluorescence. Native PAGE separation of freshly dissolved Ab 1-42 showed the presence of monomeric Ab 1-42 and a diffuse band representing soluble oligomeric Ab 1-42 for all samples ( Figure 7A). Native PAGE separation performed of the samples from the in vitro fibrillation at 60 minutes displayed an enhancement of insoluble Ab 1-42 in presence of curcumin, when compared to the control incubated with vehicle (1% ethanol). The enhanced band of insoluble material was observed at early time points, 60 minutes, with larger aggregates not being able to penetrate into the separation gel. This insoluble material appeared to increase at the expense of soluble oligomers and monomers. At longer fibrillation times, 180 minutes, the presence or absence of curcumin did not result in any difference in the amount of large aggregates nor oligomers or monomers. Even though TEM is not a quantitative method, TEM analysis was performed to obtain a morphological assessment of how curcumin influenced Ab 1-42 aggregation. Micrographs obtained for Ab 1-42 incubated for 60 minutes in the absence of curcumin displayed a great amount of morphologically disordered aggregates, as well as few amyloid fibrils associated with the aggregates. In the presence of curcumin, even at the lowest concentration, it appeared that the amount of fibrils was enhanced. At 180 min, large networks of entangled fibrils were present in all samples. Taken together, the PAGE and TEM analyses showed that curcumin does not inhibit fibrillation of Ab 1-42 , but on the contrary appeared to enhance fibrillation. A fluorescence based assay, using p-FTAA as a probe for formation of prefibrillar aggregates (including amyloid oligomers) and amyloid fibrils [29,35], was also performed. Conducting this assay in the presence of curcumin is difficult, because curcumin re-absorption of light will decrease the fluorescence signal and also competition between binding sites can occur. Hence, at least to accommodate the first reservation the fluorescence intensity was normalized for each curcumin concentration. As shown previously, p-FTAA fluoresces strongly in the presence of prefibrillar aggregates, and in the presence of only vehichle (2% EtOH) Ab 1-42 rapidly forms p-FTAA positive aggregates with a plateau at 60-240 min, followed by a decreased signal ( Figure 7D and Figure S7). Interestingly, the presence of curcumin appeared to suppress the formation of the initial prefibrillar phase, in a concentration dependent manner. These results support the notion that curcumin shifts the Ab 1-42 fibrillation pathway away from prefibrillar aggregates/oligomers and towards amyloid fibrils.

Discussion
In the present work we aimed to formulate a pharmacological treatment study of Drosophila AD models. The compound curcumin was selected from its well documented ability to be non-toxic, to mitigate neuropathology and to prevent cytotoxic Ab aggregate accumulation. A recent study by Alavez and colleagues in Caenorhabditis elegans shows the promise of curcumin (and Thioflavine T) as a drug candidate towards protein misfolding during aging and Ab 3-42 amyloidosis [36]. Our findings in Drosophila are rather different than that reported in the Alavez study. In C. elegans, no toxic effects were found on control worms, but rather a life enhancing effect. Furthermore, curcumin treated worms exhibited a decreased amount of Ab accumulation. The mechanistic outcome from our study was hence quite different even though the message is the same: curcumin prolongs lifespan and reduces neurodegeneration in an invertebrate model system for Ab amyloidosis.
As anticipated from previous reports on the potency of curcumin, the lifespan and activity behavior of AD fly models treated with curcumin displayed a positive pharmacological effect. Unexpectedly, control flies exhibited a substantially reduced survival as a function of curcumin concentration. Importantly, the pharmacological effect was directly dependent on genotype, rendering the strongest positive curcumin effect on Ab expressing flies exhibiting the worst phenotype. The curcumin induced pharmacological effect was in the following order: Ab 1-40 ,single insert Ab 1-42 ,double insert Ab 1-42 ,Ab 1-42 E22G . Intriguingly the reverse order of curcumin toxicity was evident: control.Ab 1-40 .single insert Ab 1-42 .double insert Ab 1-42 .Ab 1-42 E22G . This became most evident when comparing the median life span of the different genotypes as a function of curcumin concentration ( Figure S8). Hence the marked toxicity of curcumin for Drosophila controls was decreased or totally depleted for the amyloid expressing trangenes, indicating that Ab (and possibly also Tau) can function as a chemical detoxifier. If this is a normal function of Ab amyloid is not known, but, in the context of other Ab xenobiotic systems, a copper detoxification role within the context of transgenic Caenorhabditis elegans was recently reported [37]. We cannot rule out effects on other molecular pathways compared to direct binding to Ab, however we have shown that Ab 1-42 does bind curcumin in vitro. It is therefore particularly interesting that Ab appeared to suppress curcumin induced toxicity in the flies. It is hence possible that curcumin metabolites are responsible for the induced toxicity noted in the fly, and that this toxic effect was hindered by its sequestration within Ab. The increased lifespan upon curcumin treatment was especially striking for flies expressing the Ab peptide with the so called Arctic mutation; Ab 1-42 E22G . The Ab 1-42 E22G expressing flies increased their median survival with a striking 75% for the intermediate curcumin concentration. In the survival assay no positive effect on curcumin treatments was observed for Tau expressing flies, but rather a toxic effect at the highest curcumin concentration. Nevertheless the flies expressing Tau showed sustained locomotor activity and the largest increase in activity of all genotypes upon curcumin treatment at the intermediate concentration. This enhanced activity was sustained over time, in contrast to treatment of Ab transgenes, suggesting a different mechanism, possibly related to decreased oxidative stress [38,39] arising from antioxidant activity of curcumin.
Histological staining of fly brains followed by visual inspection showed that the treatment with curcumin at the intermediate concentration appeared to accelerate the formation of Ab amyloid fibrils (p-FTAA positive), as opposed to those reactive with Ab specific antibody. This was evident by analysis of the protein aggregation progression by microscopy at different ages (Table 1). At later time points (close to the day of death) the difference was less pronounced, but most importantly no decrease in total amount of amyloid deposition was observed as a function of curcumin treatment. In this fly model .95% of accumulated Ab 1-42 is insoluble, and a redistribution of soluble/insoluble Ab 1-42 can only be quantified if the soluble fraction was to increase during curcumin treatment. Our data showed that curcumin treatment did not affect the amount or the soluble/insoluble distribution of Ab deposition as quantified by the MSD immunoassay. Furthermore, aged flies expressing the Ab 1-42 peptides in the eyes by a strong driver (GMR-Gal4 crossings), analyzed by histological staining in the fluorescence microscope after staining with antibody, the LCO, and the nuclear marker DAPI, also displayed indifferent amyloid deposition patterns upon curcumin treatment ( Figure S9). Taken together, in neither of our experiments did we observe any decrease in the amount of Ab or Tau deposition following curcumin treatment.
One striking finding in this study was the hyper spectral analysis of p-FTAA fluorescence over time performed on double insert Ab 1-42 expressing flies: curcumin treatment accelerated the formation of well ordered amyloid fibrils in middle age. The notion that curcumin accelerated conformational conversion, rather than modulating the levels of Ab, was hence supported by the quantification of similar amounts of Ab in both treated and untreated brains. Our observations support recent findings in transgenic AD mouse models [20] and in vitro experiments [40], where it has been suggested that curcumin modulates the Ab amyloid cascade by accelerating fibril formation and decreasing oligomer formation (Figure 8). Fibrillation assays of recombinant Ab 1-42 peptide in absence and presence of curcumin was performed in order to verify such activity of the curcumin used in our Drosophila assay. The in vitro fibrillation assays of recombinant Ab 1-42 peptide in our study hence confirmed the few previous reports that curcumin decrease the population of soluble oligomers and appeared to accelerate formation of large amyloid fibrils in contrast to the majority of reports stating the fibrillation inhibitory mechanism for curcumin.
The Arctic mutant peptide Ab 1-42 E22G is well established to form soluble oligomeric protofibrils with high cytotoxicity [41], and is likely the reason for its strong toxicity in Drosophila [3,5]. The finding that sub-stoichiometric amounts of curcumin stabilize nucleation of fibril formation in vitro correlates very well with the notion that reduced oligomerization and increased fibril formation in the presence of curcumin markedly protects the Ab 1-42 E22G expressing flies. Our findings in the fly correlates well with those previously reported using mutagenesis where one additional mutation Ab 1-42 I31E in the context of Ab 1-42 E22G , mitigated neurotoxicity in Drosophila by enhancement of amyloid fibril conversion at the expense of oligomers [42]. The same mechanism could explain the neuroprotective effect of curcumin seen for the wild type Ab expressing flies.

Concluding remarks and outlook
Expressing Ab in the Drosophila CNS results in a significant neurotoxic effect, in contrast to that observed in mouse models of AD, which indicates the potency of these models for pharmacological treatment studies.We have demonstrated that curcumin exerts a general neuroprotective effect for Ab and tau Drosophila transgenes. We have demonstrated that this curcumin activity in the context of Ab is due to accelerated fibril conversion at the expense of putatively neurotoxic oligomers. It is plausible that the apparent toxicity of curcumin within Drosophila, which appears to be absent for mammalian cells, does suggest that the neuroprotective effect of curcumin can be even stronger than that reported here. The main drawback for curcumin as a drug for treatment of AD appears to be the poor bioavailability and stability in solution. With that in mind it is encouraging that curcumin analogues are synthesized as candidate drugs towards AD [43].

Fly stocks
Drosophila stocks were allowed to develop under a 12:12 hours light:dark cycle until eclosion at 26uC and posteclosion at 29uC, at 70% humidity. The flies were kept in 50 ml plastic vials containing standard Drosophila food (corn meal, molasses, yeast, and agar) and maintained post eclosion in 50 ml plastic vials containing 7 ml agar (20 g agar, 20 g sugar dissolved in 1 l of milliQ water) and yeast paste (dry bakery yeast dissolved in milliQ water) at 29uC under 12:12 hours light-dark cycles, in a setup of twenty flies per vial. The vials were changed every second day. Curcumin (28260, Fluka) was dissolved in 95% ethanol to a concentration of 10 mg/ ml and then diluted into the yeast paste into a final concentration of 1, 10 and 100 mg curcumin per g yeast paste, corresponding to (w/w) 0.0001%, 0.001%, and 0.01% curcumin added. Even distribution of curcumin within the yeast paste was easily checked by visual inspection due to the strong yellow color of the compound. All three curcumin concentrations were used in the lifespan and climbing assays, but only the intermediate concentration of 10 mg curcumin per g yeast paste was used in the activity assay, for flies corresponding to the immunohistochemistry combined with LCO staining, and for flies corresponding to quantification of Ab levels by MSD. The C155-Gal4 driver [44] was used for all experiments except for the eye degeneration assays were the GMR-Gal4 driver [45] was used instead. The UAS transgenes used in the experiments were

Survival Assay
The survival assay is a standard assay for monitoring the effect of genotype or environmental conditions on Drosophila lifespan, and is especially useful for C155-Gal4 lines expressing toxic proteins in the CNS. The number of surviving flies corresponding to the C155-Gal4 crossings was counted every second day. The survival proportions were calculated using the Kaplan-Meier estimation [46] by using the log-rank method in the GraphPad Prism 5.0 d software (GrapPad Software Inc., San Diego, CA, USA). The survival times described in the study are given as median 6 standard error of the median. The number of flies in the assay was between 100 and 60 flies of each genotype, as specified in Table S2. Significance was compared using a paired student's ttest between untreated (0% curcumin) and treated with curcumin (*: P,0.05, **: P,0.01; ***: P,0.001).

Drosophila Locomotor Activity Assay
The activity of individual flies (n = 16) corresponding to the C155-Gal4 crossings was recorded using Drosophila Activity Monitors (DAM2, TriKinetics, Waltham, MA, USA), and the number of beam breaks per hour during 24 hours was registered. The DAM2 unit was placed with a light source at the front of the monitor. The assay started with 10 hours of light followed by 12 hours in the dark, and 2 hours of light. The activity was measured at 5, 10, 15, and 20 days after eclosion for all genotypes and curcumin treatments, except for Ab 1-42 E22G expressing flies and the Tau expressing flies, which was only measured for 5-10 and 5-15 days, respectively, due to their shortened lifespan. The total number of beam breaks were calculated for all genotypes and curcumin treatments and compared using two-tailed student's ttest of unpaired samples (*: P,0.05, **: P,0.01; ***: P,0.001) with GraphPad Prism 5.0 d software (GrapPad Software Inc., San Diego, CA, USA).
At least 20 fly heads of each genotype that had been fed normal food or food containing 0.001% w/w curcumin (10 mg per gram of yeast) were assayed. An estimation of the aggregate load in each genotype was done by visual inspection in the fluorescence microscope.

Spectral analysis of amyloid produced in Drosophila
Consecutive sections to the histological analyses was used for protein spectral analyzes by the use of the LCO; p-FTAA [29,32]. The p-FTAA was diluted 1:500 (from a 1 mg/ml stock solution) in PBS and was used to stain sections as described before [32]. A fluorescence microscope (Leica Microsystems Inc., Bannockburn, IL, USA), with 405/40 nm and 560/40 nm longpass filter, attached with a spectral camera (Applied Spectral Imaging Inc.) was used to collect spectral images in an interval of 450 to 700 nm. LCO hyper spectral microscopy analysis of Drosophila brain aggregates was used to evaluate the conformation of deposited Ab and Tau. The ratio of the emission peak at 508 nm and the emission peak at 612 nm for the LCO, p-FTAA, taken at 405 nm excitation and 560 nm excitation ( Figure S3) was plotted as an amyloid fibrillation index. Spectral analysis of the two excitation spectra was performed by the SpectraViewH software (Applied Spectral Imaging Inc.) and were further analyzed by the GraphPad Prism 5.0 d software (GrapPad Software Inc., San Diego, CA, USA). The spectral shift of the LCO was analyzed by the fraction of the blue shifted peak (at 508 nm from the 405 nm excitation) and the red shifted peak (at 612 nm from the 560 nm excitation). The spectral shifts were estimated as the medium fraction with the standard deviation and compared using twotailed student's t-test of unpaired samples (*: P,0.05, **: P,0.01; ***: P,0.001).
To distinguish between different forms of aggregates formed in the double inserted Ab 1-42 expressing flies corresponding to the C155-Gal4 crossing in absence or presence of curcumin at different ages, the fraction of the peaks at 508 nm and 543 nm verses the peaks at 508 nm and 612 nm were analyzed by the GraphPad Prism 5.0 d software (GrapPad Software Inc., San Diego, CA, USA). The spectral shifts were estimated as the medium fraction with the standard deviation and compared using two-tailed student's t-test of unpaired samples (*: P,0.05, **: P,0.01; ***: P,0.001).
At least five different brains of the different genotypes at different time points were analyzed. The number of spectra taken was dependent on amount of amyloids detected. At least 15 spectra were collected of every genotype at the different time points.

Quantification of the Ab 1-42 -peptide in aged Drosophila
Five fly heads of newly eclosed, ten and twenty day old double inserted Ab 1-42 expressing flies and newly eclosed, five and ten day old Ab 1-42 E22G expressing flies, both corresponding to the C155-Gal4 crossing, were homogenized in 50 ml of extraction buffer (50 mM Hepes pH 7.3, 5 mM EDTA, Protease inhibitor (Complete TM , Roche Diagnostics)). The homogenate was incubated at room temperature for 10 min, sonicated for 4 minutes in a water bath and then centrifuged at 12 000 g for 5 minutes into a ''soluble'' and ''insoluble'' fraction. 20

In vitro Fibrillation Assay
Recombinant Ab 1-42 HFIP (A-1163-2, rPeptide, Borgart, GA, USA) was dissolved in 2 mM NaOH into a concentration of 1 mg/ml (222 mM). The peptide was stored at 220uC. 10 mM Ab 1-42 HFIP was allowed to fibrillate in a 96-well assay plate (3915, Corning Inc., NY, USA) in 10 mM phosphate buffer pH 7.5 in the absence or presence of curcumin in concentrations of 0.27, 2.7, and 27 mM dissolved in EtOH (corresponding to (w/ w) 0.0001%, 0.001%, and 0.01% added curcumin). The fibrillation was performed at 37uC, at 500 rpm. Aliquots were withdrawn at time points of 0, 60, and 180 minutes and were assayed by Western blotting (see below) and transmission electron microcopy (see below). The p-FTAA fluorescence assay was performed as described in [29].
Native PAGE Western blotting 15 ml aliquots from the fibrillation assay at time points; 0, 60, and 180 minutes, were mixed with 15 ml Native sample buffer (161-0738, BioRad, CA, USA) and run on a 12% acrylamide gel during native condition. Pre-stained protein standards (161-0374, BioRad, CA, USA) and synthetic Ab 1-42 peptide were used to indicate the apparent molecular weights of peptides and aggregates. After electrophoresis, proteins were blotted onto Immobilon-P transfer membrane (IPVH00010, Millipore, Billerica, MA, USA) set at 100 V for 1 hours at room temperature. After transfer, membranes were rinsed in distilled water followed by blocking using 4% BSA/TBST (TBS with 0.1% Tween) for 2 hours at room temperature. Blocked membranes were incubated with Monoclonal mouse anti-Amyloid b 1-16 (6E10) antibody (9320-02, Signet Laboratories Inc., Dedham, MA, USA) diluted 1:10 000 in 4% BSA/TBST for 1 hours at room temperature. After incubation with primary antibody, membranes were washed three times for 10 minutes in TBST, incubated with alkaline phosphatase-conjugate anti-mouse IgG (ab6729-1, AbCam, Cambridge, MA, USA), diluted 1:1 000 in 4% BSA/TBST for 30 minutes at room temperature, and washed again three times for 10 minutes in TBST. The membrane was developed using Immun-Star TM Chemiluminescent (170-5012, BioRad, CA, USA) and visualized using a LAS-400mini attached with a CCDcamera (Fujifilm corporation, Greenwood, SC, USA). The in vitro fibril formation assay was repeated at three different occasions with fresh solutions.
Transmission Electron Microscopy 5 ml aliquots from the fibrillation assay at time points; 0, 60, and 180 minutes, were applied on formvar and carbon coated copper grids for transmission electron microscopy (Carbon-B copper grids, Ted Pella Inc., Pedding, CA, USA). The sample was allowed to adhere to the grid for two minutes at room temperature. The grid was washed with dH 2 O and incubated for 20 seconds with 5 ml 1% uranyl acetate. Transmission electron micrographs were taken using a Jeol-1230 electron microscope operating at 100 kV.  Figure S6, of double insert Ab 1-42 flies at different ages. (A) Newly eclosed double insert Ab 1-42 flies have few and small aggregates detectable with p-FTAA with variable added emission spectra. The contribution from the 508 nm peak (405/40 excitation) was commonly smaller than from the 612 nm peak (560/40 excitation). (B) At day 10, the double insert Ab 1-42 flies shows extensive Ab aggregation, detectable with p-FTAA. The spectra of the aggregates showed a high contribution of the 508 nm peak (405/40 excitation) and a low contribution of the 612 nm peak (560/40 excitation). The shoulder peak at 508 nm was clearly visible, but was consistently much lower than the peak at 543 nm. This is likely an indicator of immature fibrils. (C) At day 20, the double insert Ab 1-42 flies have large aggregates formed that are clearly recognized by the LCO with added a more distinct double peak at 508 and 543 nm, indicating well organized amyloid fibrils. (TIF) Figure S4 2D spectral analysis of amyloid in double insert Ab 1-42 expressing Drosophila showing acceleration of fibrillation by curcumin ingestion. The fraction of the emission peak at 508 nm and the emission peak at 543 nm and the fraction of the emission peak at 508 nm and the emission peak at 612 nm of the LCO, p-FTAA, taken at 405/40 nm excitation and 560/40 nm excitation was plotted as a 2D amyloid fibrillation index. (A) The variable emission spectra from newly eclosed flies, resulted in a wide variation in the 2D amyloid fibrillation index, interpreted as a wide variation in the morphology of the aggregates. (B) After ten days there was still a variation in aggregate spectra detected by the LCO in untreated flies but with a shift towards low 508/612 nm ratios. (C) After ten days of curcumin treatment, the spectra were less wide spread and shifted towards higher 508/612 nm ratios indicating more well ordered aggregate morphology. At day 20, both untreated (D) and curcumin treated (E) flies showed the grouped 2D amyloid fibrillation indexes indicating amyloid fibrils with the same morphological structure as those observed in curcumin treated day 10 flies. Some amyloid was detected in Ab expressing flies (arrow), but the fluorescence emission was too low from the aggregates to confirm if the emitted light represent the bound curcumin to Ab aggregates and were only detected as background in the 470/40 excitation filter. (C) Curcumin staining of unfed control flies followed by p-FTAAstaining combined with DAPI staining, shows no amyloid staining in controls but in (F) Ab expressing flies. Since the curcumin histological staining increased the background staining of the samples, the large aggregates were still detected by the LCO probe (arrow head), but smaller aggregates surrounding the nuclei's were more diffuse (arrow). The emission spectra from the 470/40 excitation showed typical p-FTAA spectra in (F). No excitation of the 405/30 filter was possible due to the DAPI staining, and the characteristic double peak of the p-FTAA was hence lacking. (TIF) Figure S6 Spectral contribution for Ab and Tau aggregates found in Drosophila. A fluorescence microscope, with 405/40 nm and 560/40 nm longpass filter, attached with a spectral camera was used to collect hyper spectral images in an interval of 450 to 700 nm of aggregates found in brain tissue stained with the LCO, p-FTAA. Spectral additions of the two excitations was performed by the SpectraViewH software. The spectral shift of the LCO was analyzed by the fraction of the peak at 508 nm from the 405/40 nm excitation setup, and the peak at 612 nm from the 560/40 nm excitation. For 2D analysis of the Ab deposits, the peak at 543 nm, from the 405/40 nm excitation was used. (TIF) Figure S7 Fluorescence based assay using p-FTAA as a probe for recombinant Ab aggregation. Raw data graph corresponding to Figure 7D in main article, with error bars represented SEM. No curcumin added (vehicle control 2% EtOH) represented with black lines, 0.0001, 0.001, and 0.01% (w/v) curcumin is represented in yellow, orange, and red lines respectively. (TIF) Figure S8 Genotype dependent curcumin toxicity. Normalized median survival time as a function of curcumin concentration. The data on the y-axis was calculated from the median survival time for each genotype without treatment (T 1/2 (0%) ) versus treatment with the respective curcumin concentration (T 1/2 (x%) ). Control flies (black), Ab 1-40 expressing flies (cyan), single insert Ab