Tau Ser262 phosphorylation is critical for Abeta42-induced tau toxicity in a transgenic Drosophila model of Alzheimer's disease.

The amyloid-beta 42 (Abeta42) peptide has been suggested to promote tau phosphorylation and toxicity in Alzheimer's disease (AD) pathogenesis; however, the underlying mechanisms are not fully understood. Using transgenic Drosophila expressing both human Abeta42 and tau, we show here that tau phosphorylation at Ser262 plays a critical role in Abeta42-induced tau toxicity. Co-expression of Abeta42 increased tau phosphorylation at AD-related sites including Ser262, and enhanced tau-induced neurodegeneration. In contrast, formation of either sarkosyl-insoluble tau or paired helical filaments was not induced by Abeta42. Co-expression of Abeta42 and tau carrying the non-phosphorylatable Ser262Ala mutation did not cause neurodegeneration, suggesting that the Ser262 phosphorylation site is required for the pathogenic interaction between Abeta42 and tau. We have recently reported that the DNA damage-activated Checkpoint kinase 2 (Chk2) phosphorylates tau at Ser262 and enhances tau toxicity in a transgenic Drosophila model. We detected that expression of Chk2, as well as a number of genes involved in DNA repair pathways, was increased in the Abeta42 fly brains. The induction of a DNA repair response is protective against Abeta42 toxicity, since blocking the function of the tumor suppressor p53, a key transcription factor for the induction of DNA repair genes, in neurons exacerbated Abeta42-induced neuronal dysfunction. Our results demonstrate that tau phosphorylation at Ser262 is crucial for Abeta42-induced tau toxicity in vivo, and suggest a new model of AD progression in which activation of DNA repair pathways is protective against Abeta42 toxicity but may trigger tau phosphorylation and toxicity in AD pathogenesis.


INTRODUCTION
Alzheimer's disease (AD) is a progressive neurodegenerative disease without effective therapies (1,2). Pathologically, AD is defined by an extensive loss of neurons and by formation of two characteristic protein deposits in the brain, extracellular amyloid plaques and intracellular neurofibrillary tangles [NFTs (1)]. The major components of amyloid plaques are the 40 or 42 amino acid amyloid-b peptides (Ab40 or Ab42) (3,4). Ab peptides are derived from a type 1 transmembrane protein, the amyloid precursor protein (APP), by sequential cleavage by band g-secretases (5). Molecular genetic studies of early-onset familial AD patients have identified causative mutations in APP, Presenilin 1 and Presenilin 2 (6), which increase Ab42 production and/or Ab aggregation (7,8). These results provide a strong causative link between Ab42 and AD (7).
NFTs are intracellular protein inclusions composed of the hyperphosphorylated microtubule-associated protein tau (9 -12). NFTs are detected in many neurodegenerative diseases (13), and multiple tau gene mutations and polymorphisms are associated with tauopathies, including hereditary frontotemporal dementia and parkinsonism linked to chromosome 17 (13). Tau mutations have not been associated with any known form of familial AD to date; however, tau haplotypes driving slightly higher tau expression increase the AD risk (14,15), suggesting that tau plays a role in the pathogenesis of AD as a modulator of disease progression. * To whom correspondence should be addressed at: 900 Walnut Street, JHN410, Philadelphia, PA 19107, USA. Email: koichi.iijima@jefferson.edu (K.I.); kanae.iijima-ando@jefferson.edu (K.I. -A.) An imbalance in phosphorylation and/or dephosphorylation of tau has been suggested to initiate the abnormal metabolism and toxicity of tau in AD (13,16,17). At least 30 putative Ser/ Thr phosphorylation sites in tau are phosphorylated in NFTs (16). In vitro and in vivo studies have demonstrated that tau phosphorylation at some of the disease-associated sites plays critical roles in tau binding to microtubules (18 -20) and tau fibril formation (21)(22)(23)(24). Approximately half of AD-related sites are targets for serine/proline (SP) or threonine/proline (TP) kinases (25). In transgenic animal models, overexpression of kinases that phosphorylate tau at SP/TP sites, including GSK-3b and Cdk5 modify tau phosphorylation, NFT formation and tau toxicity (26)(27)(28)(29)(30)(31)(32)(33). In addition, phosphorylation of tau at AD-related, non-SP/TP sites such as Ser262/356 increases tau phosphorylation at SP/TP sites and promotes tau toxicity (34)(35)(36).
Accumulating evidence suggests that Ab and tau synergistically contribute to the pathogenesis of AD (17). Studies in human AD cases following Ab immunization have shown decreases in amyloid burden and in phosphorylated tau in neurites surrounding the amyloid plaques (37,38). In transgenic mice overproducing human Ab and tau proteins, Ab facilitates the abnormal phosphorylation of tau at AD-related sites and enhances the formation of NFTs (39)(40)(41). Ab immunization removes amyloid pathology as well as early stage tau lesions (42), and ameliorates cognitive decline in transgenic mice that form plaques and tangles (43)(44)(45). Knockdown of tau expression suppresses Ab-induced neurotoxicity in cultured neurons (46,47), and lowering or eliminating endogenous tau expression in transgenic mice suppresses Ab-induced behavioral deficits (48). In a Drosophila model expressing human Ab42 and tau, Ab42 synergistically enhances tau-induced neurodegeneration, and tau phosphorylation at AD-related SP/TP sites is important for Ab42-induced tau toxicity (49). These reports suggest that Ab lies upstream of aberrant phosphorylation and toxicity of tau in the pathogenesis of AD. However, the molecular mechanisms by which Ab induces abnormal phosphorylation and toxicity of tau in vivo are not fully understood.
The transgenic Drosophila models of tauopathy, in which human tau is overexpressed, have been used as effective genetic model systems to reveal the mechanisms underlying human tau-induced neurodegeneration (33,34,(49)(50)(51)(52)(53)(54)(55)(56)(57). Accumulation of disease-associated conformational changes and phospho-epitopes in tau has been detected in the fly brain and eye (34,50,53). NFT formation is not observed in fly neurons (50), indicating that tau toxicity is not conferred by large insoluble aggregates of tau in the Drosophila models. These results suggest that Drosophila models of tauopathies may recapitulate early pre-tangle events in tau-associated neurodegeneration (58).
In order to study the pathogenic interactions between Ab42and tau-induced toxicity in vivo, we use a transgenic Drosophila expressing human tau (50) in combination with a transgenic Drosophila model of human Ab42 toxicity (59-61) (Please see the 'Materials and Methods' section for the details of the Ab42 fly model). Double transgenic flies expressing human Ab42 and tau enable the examination of the effect of Ab42 on tau pathology and toxicity, in the absence of the effects of APP and other fragments of APP including C-terminal and N-terminal fragments and other Ab species.
Here we show that tau phosphorylation at Ser262 plays a critical role in Ab42-induced tau toxicity. We also demonstrate that expression of DNA repair genes, including DNA damage-activated Checkpoint kinase 2 (Chk2), is increased in Ab42 fly neurons as a protective response against Ab42 toxicity. Since Chk2 phosphorylates tau at Ser262 and enhances tau toxicity (36), our results suggest that the increased activity of the DNA repair pathways in response to Ab42 may be one of the mechanisms that mediate Ab42-induced tau phosphorylation and toxicity in vivo.

Human Ab42 enhances human tau-induced toxicity in transgenic fly eyes and brains
Expression of wild-type human tau (0N4R, see the 'Materials and Methods' section) in Drosophila eyes using the panretinal gmr-GAL4 driver causes eye degeneration characterized by small eye size, rough surface and reduced retinal thickness (33,50) (Fig. 1B, F and I). Co-expression of Ab42 with tau significantly enhanced the reduction in the external size of eyes (Fig. 1C) as previously reported (49) and internal retina thickness ( Fig. 1G and I), whereas Ab42 expression alone did not significantly affect eye structures at this level of expression (Fig. 1D, H and I). Similar results were obtained from two independent Ab42 transgenic fly lines (Ab42#1 and Ab42#2, Fig. 1I). These results indicate that Ab42 expression exacerbates tau toxicity in vivo and that the fly eye can be used to investigate the molecular mechanisms of Ab42-induced tau toxicity.
Expression of tau in neurons by the pan-neuronal elav-GAL4 driver caused an abnormality in fly brain structures. The mushroom body structures are paired structures, formed from approximately 2500 cells in each hemisphere, which play a central role in olfactory learning and memory in flies (Fig. 1J) (62). The calyx is a dendritic region in the mushroom body, and the calyx neuropil in flies expressing tau were smaller than those in controls ( Fig. 1K) or were missing in some of the fly brains ( Fig. 1N) (63). Co-expression of Ab42 and tau caused a complete loss of calyx structures ( Fig. 1L and N). In contrast, a normal calyx structure was observed in flies expressing Ab42 alone ( Fig. 1M and N). These results indicate that Ab42 enhances tau-induced toxicity also in fly brain neurons.
Human Ab42 increases human tau phosphorylation at Ser202, Thr231 and Ser262 in transgenic fly eyes and brains We examined whether enhancement of tau-induced toxicity by co-expression of Ab42 in fly eyes and brains was accompanied by an increase in tau phosphorylation levels at AD-related sites. Tau phosphorylation at 16 AD-related sites (indicated in Fig. 2A), in the presence or absence of Ab42 expression, was examined by western blotting using phospho-tau specific antibodies. This systematic analysis revealed that tau phos-

2948
Human Molecular Genetics, 2010, Vol. 19,No. 15 phorylation levels were significantly increased at Ser202, Thr231 and Ser262 when tau was co-expressed with Ab42 in both eyes (using the gmr-GAL4 driver, Fig. 2B) and brains (using the pan-neuronal elav-GAL4 driver, Fig. 2C). These results were confirmed in three independent transgenic fly lines carrying Ab42, and using two different antibodies to detect Ser202, Thr231 and Ser262 phosphorylation.
To examine whether Ab42 affects tau solubility in fly brains, fly brains expressing human tau in the presence or absence of Ab42 were extracted with sarkosyl, and the sarkosyl-soluble and -insoluble fractions were subjected to western blotting with anti-human tau antibody. Expression of Ab42 did not affect the distribution of tau in the sarkosyl-soluble and -insoluble fractions (Fig. 2D). Moreover, paired helical filaments were not detected in the sarkosyl-insoluble fractions from fly brains co-expressing Ab42 and tau by transmission electron microscopy. The Ser262 phosphorylation site of tau is critical for the pathogenic interaction between Ab42 and tau in transgenic flies Tau phosphorylation at Ser262 has been shown to play a critical role in tau toxicity (34)(35)(36). To test whether the Ser262 phosphorylation site is required for the pathogenic interaction between Ab42 and tau, we have established transgenic fly lines carrying human tau with an alanine mutation at the Ser262 site (S262A tau), which express comparable levels of S262A tau to wild-type human tau (Fig. 3A) (36). Consistent with the previous report using S262A/S356A tau (34), phosphorylation at Ser202 (AT8 epitope) was found to be significantly lower in S262A tau than in wild-type tau, whereas phosphorylation at Thr231 (AT180 epitope) was not significantly altered (Fig. 3A). The S262A single mutant dramatically suppressed tau toxicity in the retina as reported previously (Fig. 3B) (36), and also in the brain ( Fig. 3D and F). Co-expression of Ab42 and S262A tau using the pan-retinal gmr-GAL4 driver did not cause any reduction in eye size (Fig. 3C). In addition, co-expression of Ab42 and S262A tau by the pan-neuronal elav-GAL4 driver did not cause structural defects in the mushroom body ( Fig. 3E and F). These results indicate that the Ser262 site is critical for the pathogenic interaction between Ab42 and tau.
Expression of Chk2, as well as a number of genes involved in DNA repair pathways, is increased by human Ab42 expression in transgenic fly brains We have recently shown that the human DNA damage-activated Chk2 phosphorylate tau at Ser262 in vitro (36). Overexpression of Drosophila Chk2 enhances tau toxicity, and the Ser262 phosphorylation site is critical for the Positions of the phosphorylation sites tested in this study are shown. Gray boxes, four repeats of the microtubule-binding domain. (B and C) Fly heads expressing human tau alone (tau) or with Ab42 (tau + Ab42) driven by the pan-retinal gmr-GAL4 at 1 dae (B) or pan-neuronal elav-GAL4 at 25 dae (C) were subjected to western blotting with anti-tau (total tau), anti-phospho-tau (pS202, pT231 and pS262) and anti-Ab42 antibodies. Flies carrying the driver only were used as the negative control (control). The phosphorylation levels in the eye and brain of flies co-expressing tau and Ab42 (tau + Ab42) are shown as a ratio relative to that in flies expressing tau alone (tau). Representative blots are shown. Asterisks indicate significant differences from tau alone (tau) [n ¼ 4 or 5, * P , 0.05 (Student's t-test)]. (D) Co-expression of Ab42 did not increase sarkosyl-insoluble tau in the fly brain. Western blotting of sarkosylsoluble and -insoluble fractions of head extracts from flies expressing tau alone (tau), or tau and Ab42 (tau + Ab42), driven by the pan-neuronal elav-GAL4 driver. Head extracts from flies expressing Ab42 alone (Ab42) was used as a negative control. Flies are at 35 dae.

2950
Human Molecular Genetics, 2010, Vol. 19,No. 15 toxic interaction between Chk2 and tau in a transgenic fly model (36). Using quantitative real-time PCR, we found that mRNA levels of Chk2 were increased in the fly brain by the expression of human Ab42 in neurons (Fig. 4). The increase in Chk2 mRNA levels was more prominent in the fly brains expressing Ab42 with the familial Alzheimer's disease Arctic mutation (Ab42Arc: E22G substitution), which shows more enhanced accumulation and toxicity than Ab42 (60) (Fig. 4). These results suggest that the Ab42-induced increase in Chk2 expression may underlie the increase in tau phosphorylation and the enhancement of tau toxicity. We examined whether a genetic reduction of Chk2 ameliorates the enhancement of tau toxicity caused by Ab42. Because the homozygous null mutants of Chk2 are lethal (64), we tested the effect of a heterozygous loss-of-function mutation of Chk2 on tau-induced retinal degeneration in the presence or absence of Ab42. A heterozygous loss-of-function mutation of Chk2 did not significantly suppress the enhancement of tau-induced retinal degeneration caused by Ab42 (data not shown), suggesting that loss of one copy of Chk2 is not sufficient to reduce the enhancement of tau toxicity caused by Ab42.
Why is Chk2 expression upregulated in the Ab42 fly brain? Chk2 is a DNA damage transducer, which is activated in response to double-strand breaks in DNA (65,66). The nonhomologous end-joining DNA repair pathway is the major repair pathway for DNA double-strand breaks in post-mitotic neurons (67). Genes involved in non-homologous end-joining DNA repair pathways (Ku70, rad50, rad54 and Ligase 4) and Nijmegen breakage syndrome (nbs), a component of the double-strand break sensor MRN complex, were upregulated in Ab42 and Ab42Arc fly brains (Fig. 4F). Increased expression of the replication factor C subunit 40 (RfC40) and proliferating cell nuclear antigen (PCNA), which is important for both DNA synthesis and DNA repair (68) and is abnormally re-expressed in human AD brains and animal models of AD (51,69,70), was also detected. In addition, expression of the Drosophila homologs of genes involved in direct repair (O-6-alkylguanine-DNA alkyltransferase), base excision repair (XRCC1) and mitochondrial single-strand

Activation of DNA repair pathways is a protective response against Ab42-induced toxicity
While genes involved in the DNA repair response are upregulated (Fig. 4), damaged DNA and apoptosis were not detected in the brains of flies expressing Ab42, as indicated by TUNEL staining (71) and EM analysis (60). These results suggest that expression of DNA repair genes are induced as a protective response against Ab42 toxicity. We tested this possibility by blocking the function of the tumor suppressor p53, which is a transcription factor that regulates DNA damage-induced transcription (72)(73)(74). Drosophila p53 regulates induction of pro-apoptotic genes and DNA repair genes, including components of the non-homologous end-joining repair pathway, after DNA damage (75). Two dominant negative forms of p53 have been used to disrupt p53 functions in Drosophila (76). DN-p53-259H carries a point mutation in the p53 DNAbinding domain, and DN-p53-Ct is a C-terminal p53 fragment. Both mutants form tetramers with endogenous p53, but fail to bind DNA and disrupt p53 functions (76).
To test whether a reduction of p53 function would enhance Ab42 toxicity, we examined the effect of neuronal expression of the dominant negative forms of p53 on Ab42-induced locomotor defects. Ab42 flies show age-dependent, progressive locomotor dysfunction starting around two weeks after eclosion, which can be detected by a climbing assay (59,60). In this assay, flies were placed in an empty plastic vial and tapped to the bottom. The number of flies at the top, middle or bottom of the vial was scored after 10 s. The neuronal expression of DN-p53-259H or DN-p53-Ct enhanced the locomotor defects induced by Ab42 (Fig. 5A, 19 day and 29 day). In contrast, neuronal expression of DN-p53-259H or DN-p53-Ct alone did not cause locomotor defects at up to the age of 36 days after eclosion (Fig. 5B). These results indicate that the induction of DNA repair responses is protective against Ab42 toxicity.

DISCUSSION
Elucidation of the mechanisms by which Ab42 induces abnormal phosphorylation and toxicity of tau is crucial to understanding the complex pathogenesis of AD. We have demonstrated here that, in transgenic Drosophila expressing both human Ab42 and tau, Ab42 increases tau phosphorylation at AD-related sites including Ser262 and enhances tau-induced neurodegeneration (Figs 1 and 2). Co-expression of Ab42 and tau carrying the non-phosphorylatable Ser262Ala mutation did not cause neurodegeneration (Fig. 3), suggesting that the Ser262 phosphorylation site is required for the pathogenic interaction between Ab42 and tau.
Tau phosphorylation at Ser262 is increased in pre-tangle neurons in AD (77,78). Increased tau phosphorylation at Ser262 is observed in cellular and animal models such as the cultured neurons treated with Ab42 (79), brains of double transgenic mice expressing human APP and tau (80), and monkey cortex after injection of Ab (81). Our results are consistent with these reports and suggest that the double transgenic fly model co-expressing human Ab42 and tau recapitulates a pathological phosphorylation of tau induced by Ab42 in mammalian neurons. In transgenic mice overproducing human Ab and tau, Ab enhances the formation of NFT (39 -41). In contrast, in the double transgenic fly model, neither sarkosyl-insoluble tau nor PHF tau was detected (Fig. 2). These results suggest that large tau aggregates are not involved in the enhancement of Ab42-induced tau toxicity in the transgenic fly model.
What are the mechanisms by which Ab42 enhances tau toxicity through Ser262 phosphorylation? Ser262 is located in the microtubule-binding domain of tau, and phosphorylation at

2952
Human Molecular Genetics, 2010, Vol. 19,No. 15 Ser262 reduces tau binding to microtubules (19), which may increase the chances of abnormal phosphorylation at other AD-related sites and, consequently, enhance tau toxicity (17,82,83). In the Drosophila model, tau phosphorylation at Ser262 triggers a temporally ordered series of phosphorylations at several proline-directed kinase target sites (SP/TP sites) and generates disease-associated phospho-epitopes (34). In the double transgenic fly model, we detected that phosphorylation of tau at two of the SP/TP sites, Ser202 and Thr231, was increased by Ab42 (Fig. 2). Moreover, studies of transgenic flies co-expressing Ab42 and tau have revealed that phosphorylation at AD-related SP/TP sites is involved in Ab42-induced tau toxicity (49,84). These results suggest that the increase in tau phosphorylation at SP/TP sites followed by Ser262 phosphorylation may be one of the mechanisms underlying Ab42-induced enhancement of tau toxicity. Interestingly, a recent study has shown that the introduction of the S262A/ S356A mutation to tau can suppress the toxicity of tau hyperphosphorylated at SP/TP sites (35). This raises a possibility that tau phosphorylation at Ser262 affects tau toxicity not only by increasing tau phosphorylation at SP/TP sites but also controlling toxicity of tau phosphorylated at SP/TP sites. Widespread single and double-strand DNA breaks have been detected in neurons in the brains of patients with AD and with mild cognitive impairment (85)(86)(87)(88)(89)(90)(91)(92)(93)(94)(95)(96)(97)(98). More DNA damage was found in the aging hippocampus, one of the vulnerable regions of the brain in AD, than in the aging cerebellum (99). In postmortem brains from patients, the neurons that show NFT formation in AD are the same as those that show age-related accumulation of DNA damage (100). The Ab42 peptide is known to cause oxidative stress (101,102), and damage to nucleic acids caused by reactive oxygen species includes base modifications such as 8-hydroxydeoxyguanosine, single-strand breaks and double-strand breaks if single-strand breaks are in close proximity (103). Expression of genes involved in DNA repair responses, including the DNA damage-activated Chk2, was increased in response to human Ab42 expression in the fly brain (Fig. 4). Since Chk2 phosphorylates tau at Ser262 and enhances tau toxicity in a transgenic Drosophila model (36), these results suggest that increased expression of the DNA repair transducer Chk2 may be one of the mechanisms underlying Ab42-induced phosphorylation and toxicity of tau in vivo.
While genes involved in the DNA repair response are upregulated (Fig. 4), damaged DNA and apoptosis were not detected in the brains of flies expressing Ab42 (60,71). Furthermore, blocking the function of p53, which mediates expression of DNA repair genes, enhanced Ab42-induced behavioral deficits (Fig. 5). These results suggest that the upregulation of genes involved in DNA repair pathways is protective against Ab42 toxicity, but the increased activity of DNA damage-activated kinases such as Chk2 may cause tau phosphorylation and toxicity in AD progression.
In summary, this study has demonstrated that tau phosphorylation at Ser262 is critical for Ab42-induced tau toxicity. Additionally, our results suggest that the activation of DNA damage-activated kinases by Ab42 may be involved in the pathogenic interaction between Ab42 and tau. Increases in DNA repair gene expression have been reported in aged brains (104) and in brains from Down's syndrome patients (105), and DNA repair efficiency is changed in AD brains (90,(106)(107)(108)(109)(110)(111)(112). The DNA damage-activated Chk1 and Chk2 are expressed in post-mitotic neurons in the brain (113,114), and it will be important to investigate whether Chk1 and Chk2 are activated in AD brains.

Transgenic fly models of Ab42 toxicity
We have established the transgenic fly line carrying human Ab42 and Ab42 with Arctic mutation, which has been described previously in detail (59,60,115). Briefly, to produce human Ab42 in the secretory pathway of fly neurons, the Ab42 peptide sequence is directly fused to a secretion signal peptide at the N-terminus. Mass spectrometry analysis has revealed that the Ab42 transgenic flies produce the intact human Ab42 peptide in the fly brain (59,60), and immuno-electron microscopy has shown that the expressed Ab42 is localized to the secretory pathways in neurons in the fly brain (60). These Ab42 flies show late-onset, progressive short-term memory defects, locomotor dysfunctions, neurodegeneration and premature death, accompanied by the formation of Ab42 deposits (59 -61).

Fly stocks
The transgenic fly line carrying the human 0N4R tau, which has four tubulin-binding domains (R) at the C-terminal region and no N-terminal insert (N), was a kind gift from Dr Mel Feany (Harvard Medical School) (50). We have previously established the transgenic fly lines carrying S262A mutant tau (36). Other fly stocks were obtained from: Drs Wei Du (the University of Chicago) (Chk2[E51]) and the Bloomington Drosophila Stock Center (Indiana University) (UAS-DN-p53-Ct, UAS-DN-p53-259H, gmr-GAL4 and elav-GAL4). Crosses were maintained on standard cornmealbased Drosophila medium at 258C.

Histological analysis
To analyze internal eye structure, heads of female flies at 1 day-after-eclosion (dae) were fixed in Bouin's fixative (EMS) for 48 h at room temperature, incubated 24 h in 50 mM Tris/150 mM NaCl and embedded in paraffin. Serial sections (6 mm thickness) through the entire heads were prepared, stained with hematoxylin and eosin (Vector), and examined by bright-field microscopy. Images of the sections that include the retinal were captured, and retina thickness was measured using Image J. To analyze fly brain structures, paraffin sections of heads of 1 dae males were prepared as described previously (115). Heads from five to ten flies were analyzed for each genotype.

Western blotting
Total tau was probed with anti-tau monoclonal antibody (Tau46, Zymed), and phosphorylated tau was probed with phospho-tau specific antibodies against phospho-Thr175/181 (AT270, Pierce), phospho-Ser199 (Biosource), phospho-Ser202 (CP13, Peter Davis), phospho-Ser409 (Biosource) and phospho-Ser422 (Biosource). Fifteen fly heads for each genotype were collected at 1-3 dae and homogenized in SDS-Tris-Glycine sample buffer, separated by 10% Tris-Glycine gel and transferred to nitrocellulose membrane. The membranes were blocked with 5% milk (Nestle), blotted with the antibodies described above, incubated with appropriate secondary antibody and developed using ECL plus Western Blotting Detection Reagents (GE Healthcare). The signal intensity was quantified using ImageJ (NIH). Western blots were repeated a minimum of three times with different animals and representative blots are shown.

Extraction of sarkosyl-soluble tau
Sarcosyl-insoluble tau was prepared as described in (50,116,117). Briefly, fly heads were homogenized in 10 volumes buffer and centrifuged for 20 min at 15 000g. The supernatant was brought to 1% N-lauroylsarcosinate, incubated for 1 h at room temperature with shaking and then further centrifuged for 1 h at 100 000g. The resultant highspeed pellet was re-suspended at 10 ml per 50 mg of starting material. Tau levels in the sarcosyl-soluble and -insoluble fraction were analyzed by western blot. This sarcosylinsoluble fraction was subjected to transmission electron microscopy. Flies were at 35 dae.

Climbing assay
The climbing assay was performed as previously described (60). Approximately 25 flies were placed in an empty plastic vial. The vial was gently tapped to knock the flies to the bottom, and the number of flies at the top, middle, or bottom of the vial was scored after 10 s. Experiments were repeated more than three times, and a representative result was shown.