Abeta42-Induced Neurodegeneration via an Age-Dependent Autophagic-Lysosomal Injury in Drosophila

The mechanism of widespread neuronal death occurring in Alzheimer's disease (AD) remains enigmatic even after extensive investigation during the last two decades. Amyloid beta 42 peptide (Aβ1–42) is believed to play a causative role in the development of AD. Here we expressed human Aβ1–42 and amyloid beta 40 (Aβ1–40) in Drosophila neurons. Aβ1–42 but not Aβ1–40 causes an extensive accumulation of autophagic vesicles that become increasingly dysfunctional with age. Aβ1–42-induced impairment of the degradative function, as well as the structural integrity, of post-lysosomal autophagic vesicles triggers a neurodegenerative cascade that can be enhanced by autophagy activation or partially rescued by autophagy inhibition. Compromise and leakage from post-lysosomal vesicles result in cytosolic acidification, additional damage to membranes and organelles, and erosive destruction of cytoplasm leading to eventual neuron death. Neuronal autophagy initially appears to play a pro-survival role that changes in an age-dependent way to a pro-death role in the context of Aβ1–42 expression. Our in vivo observations provide a mechanistic understanding for the differential neurotoxicity of Aβ1–42 and Aβ1–40, and reveal an Aβ1–42-induced death execution pathway mediated by an age-dependent autophagic-lysosomal injury.


Introduction
The pathological hallmarks of Alzheimer's disease (AD) are amyloid plaques, neurofibrillary tangles and widespread neuronal loss. A century-old puzzle about the causal relationship between amyloid formation and neurodegeneration remains unresolved due to the lack of a definitive pathogenic pathway linking aggregate-prone proteins with neuronal death [1]. Amyloid beta (Ab) with 40 and 42 amino acids in length (Ab 1-40 and Ab 1-42 , respectively), the main components of amyloid plaques, are aggregate-prone peptides generated from proteolytic processing of amyloid precursor protein (APP) [2]. Ab  has been shown to be more neurotoxic than Ab 1-40 and thus more directly linked to development of AD [2,3]. The underlying mechanism of Ab species-specific neurotoxicity, however, is still absent.
Classically, extracellular deposition of Ab was thought to be important in AD pathogenesis. More recently, evidence has demonstrated that intraneuronal Ab may play a crucial role in the early progression of the disease [4,5]. Intraneuronal protein aggregates are primarily degraded by macroautophagy (usually referred as to ''autophagy''), a lysosome-mediated catabolic pathway responsible for turnover of long-lived proteins and organelles [6][7][8]. Although basal autophagy is undetectable in healthy neurons [9], the pathway is important to maintain neuronal homeostasis [10,11]. Autophagy has been shown to be extensively involved in Alzheimer's [12][13][14], Parkinson's, lysosomal storage diseases, myopathies, cancers, etc. [6,15]. However, it is largely unknown if autophagy has a protective or deleterious effect on these diseases [15][16][17]. Activation of autophagy in APP transgenic mice by genetic induction of Beclin1 results in reduced Ab deposition [18], suggesting that autophagy functions in Ab clearance. Mouse models expressing mutant presenilin 1 and APP demonstrated that Ab peptides are preferentially produced or deposited in autophagic compartments [13], raising the possibility that a physical interaction may occur between Ab 1-42 and autophagic vesicles. Ab 1-42 expression in nematode muscle cells results in an accumulation of autophagic vesicles that associate with animal paralysis [19], directly linking Ab 1-42 proteotoxicity with autophagy malfunction. It is thus important to further establish if autophagy malfunction is a cause or an effect of AD pathogenesis. Additionally, the phenotype and fate of neurons with dysfunctional autophagy is poorly characterized.
To identify an Ab 1-42 -induced pathogenic pathway, we use the Drosophila Gal4-UAS system to express human Ab  or Ab  in two different subtypes of neurons in flies. Both transgenes incorporate a rat preproenkephalin secretory signal peptide to direct secretion after expression and this has been confirmed both in vitro [20] and in vivo [20,21]. Ab 1-42 expression induces an age-dependent impairment of neuronal autophagy at a post-lysosomal stage leading to extensive neuronal damage and death. Our data provide the first experimental evidence for an autophagy-mediated neurodegeneration that may be responsible for Ab 1-42 -specific neurotoxicity.

Results
Differential neurotoxicity of Ab  and Ab  Human Ab  or Ab  transgene is expressed in subtypes of Drosophila neurons where soluble GFP is also expressed as a cytosolic reporter that is independent of Ab expression. GFP labels somas and neuropil of targeted neurons; while Ab 1-42 immunostaining is primarily limited to neuronal somas (Fig. S1). When expression is limited to cholinergic neurons, Ab 1-42 results in a 38.1% of decrease in mean lifespan relative to control (log-rank P,0.0001, Fig. 1A) suggesting a significant Ab 1-42 neurotoxicity. In contrast, Ab 1-40 expression does not shorten fly lifespan. Locomotor activity of Ab 1-42 flies shows an accelerated decrease compared with Ab 1-40 or control flies (Fig. 1B). Similar results were obtained for Ab  or Ab 1-42 expression limited to GABAergic (and glutamate motor) neurons (not shown). Relative expression levels of Ab transgenes measured by reverse transcription quantitative PCR (RT-qPCR) show that Ab 1-40 expression is significantly higher than Ab 1-42 (Fig. 1C), thus ruling out the possibility that Ab 1-42 -specific neurotoxicity is associated with a higher level of the transgene expression. Cytosolic GFP fluores-cence in control (not shown) or Ab 1-40 flies (Fig. 1D) shows relatively homogeneous distribution in neurons. In contrast, region and age-matched Ab 1-42 samples exhibit numerous punctate structures with high GFP fluorescence (fluorescent puncta) relative to the surrounding cytosol with lower GFP fluorescence (Fig. 1E). These puncta show a significantly age-dependent increase (Fig. 1F) that has a negative correlation with animal climbing ability.
Ab 1-42 -induced fluorescent puncta are large autophagic vesicles Electron microscopy was used to identify subcellular structures that could account for the puncta in Ab 1-42 flies. Normal neuronal somas in control flies exhibit well-defined nuclei surrounded by limited cytoplasm (Fig. 2A). However, many neuronal somas in Ab 1-42 flies exhibit an increased volume of cytoplasm and numerous large autophagic vesicles (Fig. 2B). The increased cytoplasmic volume is consistent with the large size of many Ab 1-42 -targeted neurons as shown in Fig. 1E. To test if autophagic vesicles represent Ab 1-42 -induced puncta, we expressed a transgenic fusion protein between autophagy-specific gene 8a (Atg8a) and GFP. Atg8a-GFP climbing assay, N = 160 for all three cohorts). Note that survival rates correlate well with climbing ability in control and Ab 1-40 flies. However, 88% of Ab 1-42 flies at 16 days survive with only 5% maintaining active climbing ability. Ab 1-42 flies thus have accelerated neurological deficits that precede animal death. (C) Levels of Ab transcripts in fly heads are significantly higher for Ab 1-40 relative to Ab 1-42 (data are the mean+SEM, N = 3 for each group, two-tailed P value by student's t test). (D-E) Cytosolic GFP fluorescence exhibits an even distribution in Ab 1-40 flies (16-day-old adult, D) in contrast to an extensive accumulation of punctate structures in an age-and region-matched Ab 1-42 sample (E). GFP fluorescence in the Ab 1-42 sample is decreased in cytosol (arrowheads) but especially bright in puncta (arrows). Some neuronal somas appear abnormally large (stars). Cellular boundaries also appear to be indistinct (arrowheads). Note that cytosolic GFP expression is independent of the expression of Ab 1-40 or Ab 1-42 thus the fluorescent puncta are not likely to be the structure of Ab 1-42 aggregation. (F) An age-dependent increase of fluorescent puncta in Ab 1-42targeted neurons (data are mean+SEM, two-tailed P values by student's t test, n = 9 for each group). Scale bars = 5 mm. doi:10.1371/journal.pone.0004201.g001 moves from an even distribution in cytosol to a punctate distribution in autophagic vesicles following autophagy induction [22,23]. Using a cytosolic RFP reporter to distinguish Ab 1-42 -induced puncta from the autophagy reporter, we observe formation of numerous RFP puncta in targeted neurons that extensively colocalize with punctate Atg8a-GFP (Fig. 2C), suggesting that Ab 1-42 -induced puncta are autophagic vesicles. Ab 1-42 expression induces an age-dependent decrease in autophagic degradative function Healthy neurons are thought to have a high efficiency of autophagy degradation; thus autophagic vesicles are usually undetectable due to their rapid turnover [9,24]. To test if induction of normal autophagy could result in accumulation of fluorescent puncta in the absence of Ab 1-42 expression, we fed 1 mM rapamycin, an autophagy inducer [25], to control flies. Continuous rapamycin feeding does not result in puncta formation in neurons examined in up to 16-day-old adults ( Fig. 3A), suggesting that induction of normal autophagy per se does not result in puncta accumulation. Therefore, the numerous puncta in Ab 1-42 -targeted neurons (Fig. 3D) are abnormal autophagic vesicles with undigested cargo as indexed by GFP. Electron micrographs reveal numerous autophagic vesicles in neurons from 1-day-old Ab 1-42 flies (Fig. 3B); however, no fluorescent puncta are detectible in neurons at this age ( Fig. 3C), suggesting that autophagic vesicles in young Ab 1-42 flies (1-5 days) are functionally normal. The agedependent increase of fluorescent puncta in Ab 1-42 -targeted neurons (Fig. 1F) indicates that the degradative function of autophagy is likely to become progressively impaired with aging. Anti-Ab immunostaining shows colocalization between fluorescent puncta and Ab 1-42 (Fig. 3E), indicating an accumulation of Ab 1-42 along with other undigested cargo. Ab 1-40 expression under the same experimental conditions does not result in an age-dependent accumulation of fluorescent puncta.
Autophagic dysfunction is not due to defective vesicle fusion Newly formed autophagosomes are known to fuse with lysosome-related vesicles such as autophagic vacuoles, endosomes or lysosomes to acquire catabolic enzymes necessary for their degradative function [6]. To test if Ab 1-42 -induced puncta are autophagosomes that have failed in vesicle fusion, we stained fly brains with LysoTracker Red, an acidophilic chemical that marks lysosomes or other lysosome-related vesicles [26,27]. Many of the puncta are positively stained by LysoTracker Red (Fig. 4A), suggesting that many, but not all, of the accumulated puncta are  (Fig. 5D). Consistently, confocal micrographs demonstrate that some Ab 1-42 -targeted neurons develop large areas completely devoid of GFP fluorescence (Fig. 5E). These areas are large, irregular, absent of well-defined edges (Fig. 5E) and not bounded by any limiting membrane (Fig. 5D), suggesting that they may represent a type of unlimited digestion or erosive destruction of normal cytoplasmic compo-nents. Neurons with erosive areas often lack DAPI staining or the DAPI staining appears smeared (Fig. 5F), consistent with a partially or completely destroyed nucleus (Figs. 5D and S4). Neuropil areas also exhibit damage (Fig. S5).

Autophagic leakage contributes to the erosive destruction
Local electron lucent areas in cytosol exhibit a radial dispersion surrounding autophagic vesicles (Fig. 6A), raising the possibility that cytoplasmic erosion may be initiated by a leakage of the catabolic contents of post-lysosomal autophagic vesicles, possibly due to a compromise in their membrane integrity. The multilamellar material in the cytosol of neurons from Ab 1-42 flies provides strong evidence supporting this possibility. Multilamellae, derived from lipid accumulation within autophagic vesicles [28,29], are usually well-packed and contained within the vesicles (Figs. 6B and S6). However, the unexpected appearance of multilamellae in cytosol in the vicinity of autophagic vesicles (Figs. 6C and S7) suggests a leakage of autophagic contents from the vesicles. The erosive areas in degenerative neurons are frequently observed with the appearance of either recognizable autophagic vesicles, multilamellae or both (Fig. S4), suggesting an association between autophagic leakage and erosive destruction of cytoplasm. This subcellular morphology has never been observed in neurons from age-matched control or Ab 1-40 flies. Additionally, the diffuse LysoTracker staining occurs at and beyond the erosive areas with decreased or absent GFP fluorescence in Ab 1-42 -targeted neurons (Fig. 6D), indicating that cytoplasmic acidification, likely due to leakage of post-lysosomal vesicles, may precede erosive destruc-   flies older than middle age begin to exhibit a small number of puncta in a few neurons but there is no obvious age-dependent increase (Fig. 7B). These morphological features suggest that neurons expressing Ab 1-40 can maintain relatively normal neuronal integrity, consistent with the normal lifespan and climbing ability of Ab 1-40 flies (Fig. 1A-B). Neurons expressing Ab 1-42 in 1-day-old adults also shows homogenous GFP distribution and clear cellular boundaries of neuronal somas (Fig. 7C, 1d). However, Ab 1-42 -targeted neurons in flies over 6 days old exhibit a progressive accumulation of dysfunctional autophagic vesicles (GFP puncta, quantitated in Fig. 1F) and a decrease in cytosolic GFP fluorescence followed by indistinct somal boundaries (Fig. 7C, 6d-16d). Taken together, these morphological changes and their relative time scale indicate that Ab 1-42 expression induces an age-dependent deterioration in neuronal integrity resulting from an autophagy-derived injury. The age-dependent loss of neuronal integrity in Ab 1-42 flies correlates well with the reduced lifespan and climbing ability ( Fig. 1A-B).

Autophagy activity modulate Ab 1-42 neurotoxicity
To test if autophagy activity affects Ab 1-42 neurotoxicity, we downregulate autophagy in Ab 1-42 flies by using a loss-of-function allele of autophagy-specific gene1 (Atg1 D3D ). Flies heterozygous for Atg1 D3D (Atg1 +/2 ) exhibit an expected 50% decrease in Atg1 transcript levels (Fig. 8A). Ab 1-42 flies with the Atg1 +/2 genotype have a 10.9% increase (log-rank P,0.0001) in mean lifespan compared with Atg1 +/+ genotype (Fig. 8B). To rule out the possibility that the lifespan change may result from potential variation in genetic background among fly cohorts, lifespan assay for control flies with and without the Atg1 D3D allele was also performed in parallel with Ab 1-42 flies. In contrast to Ab 1-42 flies, control flies with the Atg1 +/2 genotype have a 13.6% decrease (logrank P,0.0001) in mean lifespan compared with Atg1 +/+ genotype (Fig. 8B). Lifespan decrease in normal flies due to autophagy inhibition is consistent with previous observations in mice [10,11], confirming the importance of normal autophagy for animal survival. However, the reverse effects of autophagy inhibition on lifespan between control and Ab 1-42 flies suggest that the significant interaction between Ab 1-42 expression and autophagy activity is not due to any potential influence of genetic background. To additionally rule out the possibility that the Atg1 +/2 genotype may influence Ab 1-42 expression, the relative expression levels of Ab 1-42 transgene were measured by RT-qPCR. Normalized Ab 1-42 transcript levels in Atg1 +/2 flies are not significantly different from Atg1 +/+ flies (Fig. 8C), suggesting that the lifespan extension in Atg1 +/2 genotype is not due to a change in Ab 1-42 expression. These data suggest that downregulating autophagy has a protective effect on Ab 1-42 neurotoxicity. In addition, Ab 1-42 flies with the Atg1 +/2 genotype also show a significantly lower accumulation of fluorescent puncta compared to the Atg1 +/+ genotype (Fig. 8D) confirming the important contribution of dysfunctional autophagic vesicles to Ab 1-42induced neurodegeneration.
Autophagy activity can also be downregulated specifically in targeted neurons using an autophagy-specific gene 5 (Atg5) RNAi transgene under control of Gal4-UAS system [23]. To confirm the interaction between Ab 1-42 expression and autophagy activity, we expressed Atg5 RNAi transgene (UAS-Atg5 RNAi ) specifically in Ab 1-42 -targeted neurons. Neuron-specific Atg5 RNAi expression also results in a lifespan decrease for control flies and again a significant lifespan extension for Ab 1-42 flies (Fig. S8A). Atg5 RNAi expression also has no significant effect on Ab 1-42 expression as measured by RT-qPCR (Fig. S8B). To test if autophagy activation has differential effects on flies expressing different Ab transgenes, Ab 1-40 or Ab 1-42 flies were fed with 1 mM rapamycin to increase autophagy activity. Ab 1-40 flies show no obvious rapamycindependent changes in lifespan. However, rapamycin treatment of Ab 1-42 flies results in a significantly shortened lifespan (Fig. 8E), suggesting that enhancement of autophagy may also enhance Ab 1-42 neurotoxicity.
Aging is an independent factor affecting the degradative function of neuronal autophagy Constitutive autophagic vesicles in healthy neurons are rarely detectable [9]. Consistently, we do not observe autophagy vesicles by electron microscopy or fluorescent puncta in neurons from young control flies in the absence of Ab expression. However, neurons in control flies at middle age or older begin to exhibit puncta (Fig. S9A-B) consistent with autophagic vesicles observed in electron micrographs of brains from old control flies (not shown). Most of the puncta colocalize with LysoTracker Red staining (Fig. S9C) indicating that they also represent inefficient autophagy vesicles.

Discussion
Autophagy maintains neuronal homeostasis. It has been shown to protect neurons from degeneration in the absence of any additional aggregated protein [10,11] and improves the survival of animals expressing expanded polyglutamine proteins associated with Huntington disease [30]. The protection may depend on autophagy's ability to efficiently degrade protein aggregates. Consistently, abnormal autophagy has not been detected in patients' brains with Huntington disease [18,31]. Unfortunately, not all aggregate-prone proteins are amenable to autophagic degradation [32], raising the possibility that different types of aggregate-prone proteins associated with different neurodegenerative diseases may differentially affect autophagic clearance. Here we show that two AD-associated peptides, Ab 1-40 and Ab 1-42 , have differential effects on neuronal autophagy when expressed in Drosophila neurons. Ab 1-42 induces numerous autophagic vesicles in cytosol with an age-dependent defect in their degradative function and a compromise in their structural integrity that associates with accelerated neurological deficits and a shortened lifespan of the animals. The massive accumulation of autophagic vesicles and their large size may result from cargo storage due to impaired degradative function. Ab 1-40 expression, in contrast, does not show any detectible autophagic changes in neurons or neurological defects in animals, suggesting that Ab 1-40 is likely to be processed efficiently by neuronal autophagy. The differential effects of Ab 1-42 and Ab 1-40 on neuronal autophagy could be the underlying cause of their differential neurotoxicity. This finding may explain the paradoxical observations that APP proteolysis primarily generates Ab 1-40 in neurons [33], while it is predominantly Ab 1-42 that exhibits intraneuronal accumulation [4,5].
Abnormal autophagy is a prominent neuropathological phenotype of AD [12,14]. The abnormality has been proposed to result from a failure of fusion between autophagosomes and lysosomes making them unable to complete their degradative function [12,13,15,34]. Here we provide compelling evidence that Ab 1-42 -induced dysfunction of autophagic vesicles may occur at a post-lysosomal fusion stage. Moreover, the multilamellar structures outside of autophagic vesicles along with cytosolic acidification indicate a compromise of and then a leakage from the postlysosomal vesicles. These events cause further membrane and organelle damage as well as erosive destruction. The detailed pathogenic processes for the erosive destruction are currently unknown. However, this abnormal phenotype in our Drosophila model is consistent with previous histopathological observations of AD brains where affected neurons with an intracellular accumulation of Ab 1-42 experience cell lysis that is associated with the appearance of lysosomal enzymes in cytoplasm [35]. Thus the pathological features we observed are not species-specific, but may reflect a common consequence of Ab 1-42 pathology in neurons. We also observe that normal aging decreases the efficiency of autophagic degradation in agreement with previous reports [36,37]. Due to this common cellular consequence, aging could thus facilitates Ab 1-42 neurotoxicity, or vice versa, in agreement with similar neuropathological features shared by normal aging and AD [38]. Autophagy inhibition via a haploinsufficiency of Atg1 or targeted neuron-specific expression of Atg5 RNAi extends lifespan of Ab 1-42 flies in contrast to the deleterious effects on flies without Ab 1-42 expression, suggesting that Ab 1-42 expression may shift neuronal autophagy to a pathogenic condition. Taken together, we propose an autophagy-mediated pathogenic process where functional and intact autophagy has an early pro-survival effect that shifts to a later prodeath effect due to chronic deterioration in both degradative function and structural integrity. Our data suggest a mechanism for a dual role of autophagy that has been observed in different cellular contexts [15,16].
In summary, we express human Ab 1-42 and Ab 1-40 separately in Drosophila neurons revealing an Ab 1-42 -dependent pathogenic pathway linking autophagy malfunction with progressive neurodegeneration. Ab 1-42 impairs the degradative function and structural integrity of neuronal autophagy but not the maturation process of autophagic vesicle fusion. A death execution pathway may be triggered by leakage of post-lysosomal autophagic vesicles leading to cytosolic acidification, subcellular damage to membranes and organelles, and erosive destruction of cytoplasm. Even though the direct expression of Ab constructs containing a preproenkephalin secretory signal in Drosophila neurons may not mirror the normal conditions of Ab generation from full-length of APP, our observations suggest a mechanism for differential neurotoxicity of Ab 1-42 and Ab 1-40 as well as a cellular pathway that is responsible for Ab 1-42 -induced neuronal death. Future studies will be required to uncover detailed molecular mechanisms underlying these cellular changes.

Lifespan assay
Groups of 20 individual flies were collected within 24 hours of eclosion and placed in fresh food vials (2.3 cm diameter68.4 cm height). Vials were incubated at 28uC and live flies were regularly transferred to new food vials while the number of dead flies was counted. Parallel cohorts were assayed at the same time under the same conditions and the experimenter was blinded to the genotypes or experimental conditions. The survival rates were calculated using the LIFETEST procedure and log rank test in SAS software. The

Climbing assay
Reactive climbing assay is as described [21] with slight modifications. Ten female flies were placed in a plastic vial and gently tapped to the bottom. The number of flies that reached a mark at the top of the vial within 10 seconds was recorded. Ten trials were performed to get an average number for each time point. The data represent combined results from a cohort of flies tested every 5 days for each genotype.

Fluorescence microscopy
Adult fly brains were dissected in phosphate buffered saline (PBS), and fixed in PBS with 4% formaldehyde for 30 minutes for microscopic observation of endogenous GFP (or RFP/YFP). For LysoTracker staining, freshly dissected fly brains were incubated in PBS with 100 nM LysoTracker Red (Molecular Probes) for 15 minutes, washed 2 times with PBS and immediately observed. For Ab immunostaining, dissected brains were fixed in PBS with 4% paraformaldehyde at 4uC overnight. Fixed brains were permeabilized in 1% Triton X-100 in PBS for 5 hours and treated with 70% formic acid in PBS for 20 minutes, and immunostained with anti-Ab antibody 4G8 (Signet Laboratories) followed by detection with a Texas-Red conjugated secondary antibody. Samples were observed by confocal microscopy (Zeiss LSM 510). For quantitative morphological analyses, objects (fluorescent puncta) were manually counted in representative confocal images using ImageProPlus (Media Cybernetics). At least 3 nonoverlapping optical sections were sampled from confocal Zsections taken from 3 individual specimens. The counter was blinded to sample identities (such as genotypes, ages, or experimental conditions). Data were analyzed by ANOVA followed by pairwise Student's t-tests corrected for multiple comparisons.

Electron microscopy
Dissected brains were fixed in 1.6% paraformaldehyde with 2% glutaraldehyde and 0.06 M cacodylate buffer at 4uC for 24 hours.
Brains were post-fixed in osmium and embedded in eponate. Ultrathin sections were stained with uranyl acetate and Sato's lead. Specimens were observed with an FEI Tecnai transmission electron microscopy. Independent observations from 3-5 animals were performed for each experimental condition.

Rapamycin feeding
Rapamycin feeding was as described [25]. Flies were allowed to mate on normal fly food for 2-3 days and then transferred to fresh food containing 1 mM rapamycin (Sigma) or an equal volume of DMSO (vehicle for dissolving rapamycin) for egg collection, embryogenesis and larval growth. Female flies were collected within 24 hours after eclosion and incubated at 28uC in vials containing food supplemented with rapamycin or DMSO for lifespan assays.

Reverse Transcription and quantitative real-time PCR
Fifty fly heads were collected on dry ice and total RNA was isolated using RNA STAT-60 (Tel-Test). RNA samples were treated with Drosophila incorporating a heterozygous loss-of-function allele Atg1 D3D (Atg1 +/2 ) exhibit a significant decrease in expression levels of Atg1 mRNA in fly brains (data are normalized mean+SEM relative to GAPDH, two-tailed P values by Student's t-test, n = 3 for each group). (B) Control flies with Atg1 +/2 genotype have a shortened mean lifespan compared to Atg1 +/+ genotype (213.6%, log-rank P,0.0001). In contrast, Ab 1-42 flies with Atg1 +/2 genotype have extended lifespan relative to Atg1 +/+ genotype (+10.9%, log-rank P,0.0001) (data are the mean6SEM). (C) Normalized expression levels of Ab 1-42 transcripts exhibit no significant difference in Ab 1-42 fly heads between Atg1 +/+ and Atg1 +/2 genotypes (data are the mean+SEM, N = 3 for each group, two-tailed P value by student's t test). (D) Ab 1-42 flies with Atg1 +/2 genotype have significantly fewer fluorescent puncta in targeted neurons relative to Atg1 +/+ genotype (fly age is 11 days, data are the mean+SEM, two-tailed P value by Student's t-test, n = 9 for each group). (E) Autophagy activation by rapamycin feeding (1 mM) results in a shorter lifespan for Ab 1-42 flies (226.0%, log-rank P,0.0001), but has no obvious effect on the lifespan of Ab 1-40 flies (21.5%, log-rank P = 0.076) relative to flies fed with the same amount of DMSO (data are the mean6SEM). N is the sample size of fly cohorts for each experimental condition. doi:10.1371/journal.pone.0004201.g008 DNase I to remove genomic DNA and reverse transcribed to cDNA using the iScript cDNA Synthesis Kit (Bio-Rad). Three biological replicates of the same experimental condition were performed. Genespecific transcription levels were determined in triplicate by real-time PCR using SYBR Green Supermix (Bio-Rad) and an IQ5 real-time PCR machine (Bio-Rad). Primers were 59-CTTCCAGGCGTCG-CATCC-39 and 59-GTCTTCAGTTGTCCCTTCTTCG-39 for Drosophila Atg1, 59-CTACGCTATGACAACACCGC-39 and 59-AGACTTTGCATCTGGCTGCT-39 for Ab 1-42 / Ab 1-40 transgenes, or 59-CCACTGCCGAGGAGGTCAACTAC-39 and 59-ATGCTCAGGGTGATTGCGTATGC-39 for Drosophila glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a reference. Ct values of real-time PCR were analyzed by a custom SAS program relying on a published algorithm [42] to calculate mean normalized expression relative to GAPDH. Representative data from 3 separate experiments are presented as mean6SEM. Two-tailed P values were calculated by Student's t-test between parallel groups. Figure S1 Coexpression of cytosolic GFP reporter and Ab 1-42 (or Ab 1-40 ) in Drosophila brains using UAS-Gal4 technique. An optic lobe of an adult fly brain is shown here. Soluble GFP fluorescence (green) distributes in both neuronal somas and neuropil (outlined in cyan). Only neuronal somas are additionally labeled by Ab 1-42 immunostaining using anti-Ab antibody 4G8 (red). DAPI staining cellular nuclei (blue) is confined to cell somas. Scale bar = 20 mm. Multilamellae can spontaneously form from lipids accumulating within autophagic-lysosomal vesicles [1,2] especially at acid pH [3]. There are several different sized multilamellar stacks formed independently in a large autophagy vesicle. Disruption or incomplete digestion of membranes from many small vesicles sequestered within autophagy vesicles is the source of multilamellae (arrows). Scale bar = 0.5 mm. Supporting References: 1.