HDAC6 mutations rescue human tau-induced microtubule defects in Drosophila

Abstract

Neurons from the brains of Alzheimer’s disease (AD) and related tauopathy patients contain neurofibrillary tangles composed of hyperphosphorylated tau protein. Tau normally stabilizes microtubules (MTs); however, tau hyperphosphorylation leads to loss of this function with consequent MT destabilization and neuronal dysfunction. Accordingly, MT-stabilizing drugs such as paclitaxel and epothilone D have been shown as possible therapies for AD and related tauopathies. However, MT-stabilizing drugs have common side effects such as neuropathy and neutropenia. To find previously undescribed suppressors of tau-induced MT defects, we established a Drosophila model ectopically expressing human tau in muscle cells, which allow for clear visualization of the MT network. Overexpressed tau was hyperphosphorylated and resulted in decreased MT density and greater fragmentation, consistent with previous reports in AD patients and mouse models. From a genetic screen, we found that a histone deacetylase 6 (HDAC6) null mutation rescued tau-induced MT defects in both muscles and neurons. Genetic and pharmacological inhibition of the tubulin-specific deacetylase activity of HDAC6 indicates that the rescue effect may be mediated by increased MT acetylation. These findings reveal HDAC6 as a unique potential drug target for AD and related tauopathies.

Tau is a neuronal microtubule-associated protein (MAP) that binds to and stabilizes microtubules (MTs). However, in Alzheimer’s disease (AD), frontotemporal dementia with parkinsonism linked to chromosome 17 (FTDP-17), and other tauopathies, tau is hyperphosphorylated and aggregated into straight or paired helical filaments (PHFs) in the cell bodies and neurites of central neurons (1). In vitro biochemical studies demonstrated that hyperphosphorylated tau does not bind to MTs and thus does not promote MT stability (2, 3). Furthermore, mutations in tau lead to FTDP-17, and some of these mutations have been reported to reduce MT–tau binding affinity (4, 5), underscoring the importance of loss of normal MT-stabilizing function of tau in the pathogenesis of neurodegenerative tauopathies.

MTs are disrupted in the brains of patients and animal models of tauopathies. For example, MT density was reduced in both hippocampal neurons of transgenic mice expressing V337M mutant human tau and the spinal ventral root axons of transgenic mice expressing the smallest isoform of human tau (6, 7). Furthermore, administration of the MT-stabilizing agents paclitaxel and epothilone D to human tau-expressing mice results in improved MT density and axonal integrity (8), as well as enhanced cognitive performance (9, 10). However, paclitaxel has poor blood–brain barrier (BBB) permeability and thus is unsuitable for clinical treatment of brain diseases. Epothilone D is BBB-permeable; however, as a general MT stabilizer and genotoxic agent, it may have side effects such as neuropathy and neutropenia.

In this study, we aimed to find new strategies for mitigating tau toxicity by identifying interacting genes that can suppress tau-induced MT defects. MTs are so densely packed in axons that they cannot be resolved by conventional light microscopy (11). In contrast, Drosophila muscle cells are large—up to 40 × 150 μm of muscle 2 in the abdominal segment A3—and thus allow clear visualization of MT networks with high resolution (12, 13). Furthermore, Drosophila is a convenient model in which to conduct genetic screens for modifiers of a defined phenotype. Ectopic expression of wild-type and V337M mutant human tau in Drosophila muscle cells resulted in prominent MT defects. From a preselected genetic screen, we found that a null mutation in histone deacetylase 6 (HDAC6) 6 rescued the tau-induced MT defects in both muscles and neurons. Rescue of the tau-induced MT defects was further verified by genetic and pharmacological inhibition of the tubulin-specific deacetylase activity of HDAC6. This study thus identifies HDAC6 as a unique potential drug target for the treatment of AD and related tauopathies.

Results

Ectopic Expression of Wild-Type and Mutant Human Tau Disrupts MTs in Drosophila Muscle Cells.

Accumulation of hyperphosphorylated tau and MT defects are found in the brain neurons of both AD and FTDP-17 patients. Multiple tau mutations, including tau^V337M, have been found in FTDP-17 patients, whereas no tau mutations have been reported in AD patients. Wild-type and mutant tau have different biochemical properties such as MT-binding ability and degradation processes (4, 14). In addition, wild-type and mutant tau transgenic mice exhibit different neuronal and behavioral phenotypes (14, 15). We therefore analyzed changes in MT density and integrity in Drosophila muscles ectopically expressing either wild-type or V337M mutant human tau. Overexpression of tau^V337M or wild-type tau (tau^WT), driven by the muscle-specific C57–Gal4 (a transgenic line expressing the yeast transcription activator Gal4 specifically in muscles), led to MT fragmentation and reduced perinuclear MT density compared with the genetic control (C57–Gal4/+) (Fig. 1A). Specifically, the percentage of the tubulin-positive area in the defined perinuclear region was 24.47 ± 1.12% in cells overexpressing tau^V337M and 12.49 ± 1.02% in cells overexpressing tau^WT, compared with 45.77 ± 0.14% in the genetic control. Overexpression of tau^WT caused a more severe MT phenotype than mutant tau^V337M (P < 0.001), as indicated by a greater reduction in MT density and a greater increase in MT breakpoints (Fig. 1D). The greater disruptive effect of tau^WT on MTs did not result from higher ectopic tau^WT expression because Western analysis revealed similar levels of tau^V337M and tau^WT proteins in muscles (Fig. S1B). Overexpression of tau^V337M or tau^WT by a weaker muscle-specific driver Myosin heavy chain (Mhc)–Gal4 resulted in similar MT defects (Fig. S2), confirming that the deleterious effects on MTs were caused specifically by ectopically overexpressed human tau.

Fig. 1.

Ectopic expression of human tau^V337M or tau^WT in Drosophila muscles results in MT defects, which are rescued by a HDAC6 null mutation. (A) MTs were visualized by using an antibody against α-tubulin, and the nucleus was visualized by the DNA dye TO-PRO(R) 3 iodide. In genetic controls (C57–Gal4/+), a dense meshwork of MTs surrounded the nucleus. In muscle cells ectopically expressing mutant (tau^V337M) or wild-type (tau^WT) human tau driven by the muscle-specific C57–Gal4, MTs were broken into shorter fragments, and the density of perinuclear MTs within the 20-µm perinuclear area (circle) was significantly lower than the control. (Bars, 10 µm.) (B) Low-magnification images of larval muscles with muscle 2 indicated. (Scale bars, 300 μm.) (C) HDAC6^KO null mutations rescue the MT defects caused by tau expression. Like genetic controls (C57–Gal4/+), overexpression of tau^V337M or tau^WT in the HDAC6^KO null background (genotypes were tau^V337M HDAC6^KO/HDAC6^KO; C57–Gal4/+ or HDAC6^KO; C57–Gal4/UAS–tau^WT) exhibited dense perinuclear MTs. (Bars, 10 µm.) (D) Quantification of MT density in the 20-µm radial perinuclear area in different genotypes. A hemizygous HDAC6 mutation (HDAC6^KO/Df(1)ED7294) also rescued tau-mediated MT defects. Error bars indicate SEM. n ≥ 13 larvae analyzed. ***P < 0.001 (one-way ANOVA).

Ectopic tau may suppress tubulin expression, thus reducing available monomers for MT synthesis. We examined the amount of polymerized MTs by fractionation of cell lysates using ultracentrifugation (Fig. S3). Western analysis showed that polymerized MTs (in the pellets) were reduced to 62% of the genetic control by tau^V337M overexpression and to 69% of the control by tau^WT overexpression. However, the total α-tubulin content was similar in cell lysates from all three genotypes (Fig. S3). Together, our immunostaining and Western analysis demonstrate that ectopic expression of wild-type and mutant human tau in Drosophila muscle cells disrupts MT density and integrity.

HDAC6 Null Mutation Suppresses MT Defects Caused by Ectopic Expression of Human Tau.

To identify interacting genes that modulate the MT defects caused by ectopically expressed tau, we carried out a genetic screen by crossing flies overexpressing tau^V337M to selected mutants that affect MT dynamics and the protein level and phosphorylation status of tau. We then screened for modifications of the tau-induced MT defects in muscle cells. As shown in Table S1, several previously identified modifiers of tau toxicity (par-1, shaggy, and p150^Glued), as well as previously undescribed putative modifiers, were identified from the screen. We characterized the suppression of tau-induced MT defects by HDAC6 mutations, because an HDAC6 null mutant (HDAC6^KO) with translation start codon disrupted (16)—which had no measureable effect on the MT network—exerted the strongest rescue effect on MT defects in tau-expressing muscles among all candidates examined (Fig. 1 C and D and Table S1). As a control, the HDAC6 null mutation did not modify the MT defects caused by overexpression of other MT regulators such as the MT-severing protein Spastin (Fig. S4).

As shown in Fig. 1C, the HDAC6 null mutation fully rescued the defects in perinuclear MT density and integrity caused by tau^V337M or tau^WT overexpression. A hemizygous HDAC6 mutation [HDAC6^KO/Df(1)ED7294] also suppressed the tau-induced MT defects (Fig. 1D). These results demonstrate that the tau-induced MT defects can be rescued specifically and significantly by mutations in HDAC6.

HDAC6 Null Mutation Rescues Tau-Induced MT Defects in the Nervous System.

Although large Drosophila muscle cells allow for easy visualization of MT networks, endogenous tau is predominantly expressed in neurons. To investigate the effect of tau overexpression on MTs in the nervous system, we examined neuromuscular junction (NMJ) synapses stained with an antibody against the neuronal specific Futsch [the Drosophila ortholog of mammalian microtubule-associated protein 1B (MAP1B)] together with a Cy3-labeled HRP antibody to label the neuronal membrane (Fig. 2A). When tau^V337M and tau^WT were expressed in neurons driven by elav–Gal4, Futsch intensities in the distal 20-μm region of muscle 4 NMJ terminals were reduced to 69% and 64% of the genetic control, respectively, and they were rescued to the control level by the HDAC6 null mutation (Fig. 2 A and B). Similar rescue effects were observed in the whole NMJ terminals (Fig. 2B).

Fig. 2.

The HDAC6 null mutation rescues tau-induced MT defects at NMJ terminals and ddaE sensory neurons. (A) Confocal images of muscle-4 NMJ branches doubly labeled with anti-Futsch (green) and Cy3-conjugated anti-HRP (red). Compared with genetic controls, neuronal overexpression of either tau^V337M or tau^WT by elav–Gal4 resulted in decreased Futsch intensity at the NMJ terminals. The HDAC6^KO null mutation restored the decreased Futsch intensity caused by tau^WT or tau^V337M overexpression. (Bar, 10 μm.) (B) Statistical results of Futsch staining intensities in the whole NMJ terminals normalized to HRP intensity and Futsch intensities in the 20-µm distal parts of NMJ branches. n ≥ 23 NMJ terminals. (C) Confocal images of ddaE sensory neurons stained with anti-Futsch. Axons are indicated by arrowheads, and dendrites are denoted by arrows. Overexpression of either tau^V337M or tau^WT reduced Futsch intensities in the cell bodies and dendrites, which were suppressed by the HDAC6 null mutation. (D) Quantification results of Futsch staining intensities in the cell bodies and 20-µm proximal axons and dendrites. Futsch intensities in axons were not altered by tau expression. Error bars indicate SEM. n ≥ 20 sensory neurons. *P < 0.05; **P < 0.01; ***P < 0.001 (one-way ANOVA). ns, no significant difference.

To further demonstrate the rescue of tau-induced MT defects by the HDAC6 mutation in the nervous system, we performed anti-Futsch staining of larval dorsal da neuron E sensory neurons, which have distinct morphologies. Endogenous tau enriches in axons. However, tau overexpressed in mature hippocampal neurons is improperly distributed into somata and dendrites where disrupted MTs are observed (17). Consistent with this report, we found that Futsch intensities were decreased in the cell bodies and dendrites of sensory neurons expressing wild-type or mutant tau (Fig. 2 C and D). Specifically, Futsch intensities in the cell bodies of sensory neurons overexpressing tau^V337M and tau^WT were reduced to 88% and 73% of the genetic control, respectively, and they were significantly rescued to the control level by HDAC6 mutations (Fig. 2D). These results demonstrated that HDAC6 mutations rescue tau-induced MT defects in neurons.

HDAC6 Mutations Rescue the NMJ Morphological Abnormalities Caused by Tau Expression.

MTs play a critical role in the development of Drosophila larval NMJ synapses (12, 13, 18, 19). Previous studies demonstrated that overexpression of human tau in Drosophila motor neurons results in abnormal NMJ morphology (20, 21). Consistent with these observations, we found that there were significantly more satellite boutons—defined as small boutons that bud from major synaptic terminals or primary boutons—at the NMJ terminals of larvae overexpressing tau^V337M (11.5 ± 1.4) or tau^WT (9.6 ± 1.1) driven by elav–Gal4 compared with the genetic controls (4.1 ± 0.5) (Fig. 3B), and this anomaly was reversed by the HDAC6 null mutation, which produced normal NMJ development (Fig. 3). In addition, the average NMJ terminal length was 83% and 77% of the genetic controls in neurons overexpressing tau^V337M and tau^WT, respectively. Similarly, the HDAC6^KO null mutation rescued the decreased synaptic length caused by overexpression of tau^V337M or tau^WT to wild-type level (Fig. 3B). Reversal of these tau-induced morphological NMJ defects by an HDAC6 null mutation is consistent with the genetic interactions on MTs in neurons (Fig. 2).

Fig. 3.

The HDAC6 null mutation rescues the growth defects in Drosophila NMJ synapses induced by tau overexpression. (A) Confocal images of muscle-4 NMJ synapses of abdominal segment A3 doubly labeled with anti-cysteine string protein (CSP; red) and FITC-conjugated anti-HRP (green) to reveal synaptic vesicles and the neuronal membrane, respectively. Compared with the genetic controls, neuronal overexpression of tau^V337M or tau^WT resulted in more satellite boutons and shorter NMJ terminal length. The HDAC6^KO null mutation rescued these growth defects. (Bar, 10 μm.) (B) Quantification of the average number of satellite boutons and the average total NMJ terminal length for different genotypes. Error bars indicate SEM. n ≥ 18 NMJ terminals. *P < 0.05; ***P < 0.001 (one-way ANOVA).

Loss of HDAC6 Enzymatic Activity Is Critical for the Rescue of Tau-Induced MT Defects.

HDAC6 is a cytoplasmic enzyme that interacts with and deacetylates MTs in vitro and in vivo (22, 23). Furthermore, HDAC6 protein level is significantly higher, and the level of acetylated tubulins is reduced in AD brains (24, 25). To understand the underlying mechanism of the rescue, we first verified that the rescue of tau-induced MT defects by HDAC6 mutations was associated with increased MT acetylation in both muscles and neurons (Figs. S5 and S6).

To investigate whether loss of the deacetylase activity of HDAC6 was responsible for the rescue of tau-mediated MT defects, we generated transgenic lines expressing different HDAC6 mutants with mutations in one or both deacetylase domains (Fig. 4C). The second deacetylase domain (DD2) has been shown to deacetylate tubulin specifically (26, 27). Before we compared the effects of different HDAC6 mutations, we selected transgenic lines expressing different mutants at similar levels as detected by semiquantitative PCR. As expected, overexpression of HDAC6 carrying the H237A mutation in DD1 decreased the acetylation level of tubulin as the wild-type HDAC6, whereas that of H664A in DD2 or H237A H664A double mutants had no effect on tubulin acetylation (Fig. 4C). Reexpression of wild-type HDAC6 fully reversed the rescue of tau-induced MT defects by an HDAC6 null mutation, as did that of H237A mutant, whereas that of H237A H664A double mutants had no effect on the rescue (Fig. 4 A and B). Reexpression of H664A partially disrupted the rescue by the HDAC6 null; i.e., the HDAC6 mutation still rescued the tau-induced MT defects significantly when H664A was expressed. These results indicate that loss of the deacetylase activity conferred by the second deacetylase domain is critical for the rescue of tau-mediated MT defects.

Fig. 4.

Loss of the deacetylase activity of HDAC6 is responsible for the rescue of tau-induced MT defects. (A) Images of perinuclear MTs in different genotypes. Reexpression of wild-type HDAC6 fully reversed the rescue of tau-induced MT defects by an HDAC6 null mutation, as did the H237A mutant, whereas that of H237A H664A double mutants had no effect on the rescue. Reexpression of H664A mutant weakly disrupted the rescue by the HDAC6 null; i.e., the HDAC6 mutation still rescued the tau-induced MT defects significantly when H664A was reexpressed. (Bar, 10 µm.) (B) Quantification results of perinuclear MT densities of different genotypes. Each genotype was compared with the tau^V337M overexpression control unless otherwise indicated. Error bars indicate SEM. n ≥ 13 larvae analyzed. *P < 0.05; **P < 0.01; ***P < 0.001 (one-way ANOVA). (C Upper) HDAC6 domain organization. DD1 and DD2 indicate deacetylase domains 1 and 2, respectively; ZnF–UBP denotes zinc-finger ubiquitin binding domain. (Lower) Western analysis demonstrated that overexpression of H237A mutant decreased the acetylation level of tubulin as the wild-type HDAC6, whereas that of H664A or H237A H664A double mutants did not.

Tubacin, an Inhibitor of the Tubulin-Specific Deacetylase Activity of HDAC6, Rescues Tau-Induced MT Defects.

To further confirm that loss of the deacetylase activity of HDAC6 rescued tau-mediated MT defects, we treated tau-overexpressing flies with tubacin, a selective inhibitor of the tubulin-specific deacetylase activity of HDAC6 (26). Immunostaining revealed that tubacin partially reversed the reduced perinuclear MT density in muscle cells overexpressing tau^V337M or tau^WT (Fig. 5). Specifically, MT density was increased from 18.15 ± 1.36% (vehicle) to 28.34 ± 1.99% (tubacin) in tau^V337M-expressing muscle cells and from 16.07 ± 1.02% (vehicle) to 29.21 ± 1.32% (tubacin) in tau^WT-expressing muscle cells, whereas tubacin had no effect on MT density in C57–Gal4/+ controls (43.67 ± 1.36% in vehicle-treated and 39.93 ± 1.21% in tubacin-treated samples) (Fig. 5B). In agreement with previous reports in mammalian cell culture (28), Western analysis showed increased levels of acetylated tubulin in the muscle cells of animals treated with 50 μM tubacin compared with vehicle-treated controls (Fig. 5C). As a control, the total tubulin protein levels were unaffected by tubacin (Fig. 5C). The NMJ morphology abnormalities caused by tau expression were also rescued by tubacin treatment (Fig. S7). Thus, pharmacological inhibition of the tubulin-specific deacetylase activity of HDAC6 effectively rescues tau-induced MT defects.

Fig. 5.

Inhibition of HDAC6 by tubacin increases acetylation of tubulin and suppresses MT defects caused by tau overexpression. (A) MTs were visualized by α-tubulin antibody staining. Flies of different genotypes were fed cornmeal medium containing either vehicle (DMSO) or tubacin (50 μM), a selective inhibitor of the tubulin-specific deacetylase activity of HDAC6. MT defects caused by tau^V337M or tau^WT overexpression were partially reversed by tubacin. (B) Quantification of MT densities in the perinuclear area for the different genotypes. Comparisons were made between each genotype and the C57–Gal4/+ control (DMSO) unless otherwise indicated. Error bars indicate SEM. n ≥ 9 larvae. ***P < 0.001 (one-way ANOVA). ns, no significant difference. (C) Western analysis demonstrated that tubacin treatment did not alter the level of total α-tubulin but increased the level of acetylated α-tubulin in all three genotypes tested.

Discussion

Several screens based on tau’s neurotoxicity have been performed in Drosophila by using the “rough eye” phenotype as a readout. From these studies, multiple kinases, phosphatases, and cytoskeletal proteins have been shown to modulate tau neurotoxicity (20, 29, 30). MT defects, a hallmark of tauopathies, contribute directly to neurodegeneration (31–33). However, the precise role of tau in these MT deficits remains poorly understood, possibly due to the difficulty in observing MT fibers in neurons by light microscopy.

In this study, we ectopically expressed human wild-type tau and the FTDP-17–associated V337M mutant tau in Drosophila muscle cells. This ectopic expression recapitulated two central pathological features of tauopathies. First, both mutant and wild-type tau were phosphorylated on sites specific to paired helical filament (PHF) tau when overexpressed in muscles (Fig. S1). Second, overexpression of human tau in Drosophila muscles caused MT defects, such as reduced MT density and more MT fragments (Fig. 1), similar to those observed in AD brains and mouse models of AD (7, 34, 35). Therefore, we established a simple tauopathy model in Drosophila to facilitate studies on the deleterious effects of tau overexpression/hyperphosphorylation on MTs.

HDAC6, a unique member of the class II HDACs, is a cytoplasmic enzyme that contains two homologous deacetylase catalytic domains and one ubiquitin-binding zinc-finger domain, both of which are highly conserved between human and Drosophila (36). HDAC6 can deacetylate multiple substrates, including tubulin and the critical chaperone heat shock protein 90 (Hsp90) (37). In addition to its deacetylase activity, HDAC6 is critical for the transport of misfolded proteins by binding to both the MT-associated dynein motor and polyubiquitylated misfolded proteins, and it facilitates their destruction by promoting autophagosome–lysosome fusion (38, 39). Most previous studies on the role of HDAC6 in neurodegenerative diseases have focused on autophagy-mediated protein degradation because aberrant accumulation of misfolded proteins is considered to be a common pathogenic mechanism in a number of these disorders. For example, overexpression of HDAC6 has been shown to be neuroprotective in Parkinson disease and Huntington disease models by promoting autophagy-mediated degradation of aggregated proteins (16, 40–42). Conversely, HDAC6 loss-of-function mutations increase Hsp90 acetylation, which promotes proteasome-mediated degradation of target proteins, including tau (43, 44). Thus, it is possible that HDAC6 mutations rescue tau-mediated MT defects by promoting degradation of phosphorylated tau. However, the HDAC6 null mutation did not reduce the levels of phosphorylated tau in flies expressing tau^WT or tau^V337M (Fig. S8), inconsistent with the results from mammalian cell culture studies (44), probably due to the fact that the system we used was different from theirs. Our results indicate that the HDAC6 null mutation does not suppress MT defects by promoting tau degradation.

Introducing mutation of H664A in HDAC6 and treatment with tubacin both target the tubulin-specific deacetylase activity of HDAC6 (26, 27), rescued tau-mediated MT abnormalities (Figs. 4 and 5), suggesting that amelioration of MT defects is dependent on increased MT acetylation. It would be important to directly examine the role of tubulin acetylation against tau’s toxicity by mutating the K40 site of tubulin. However, acetylated MTs are reported to be more sensitive to the severing activity of katanin (45). How does increased acetylation level rescue tau-mediated MT defects? Previous studies in both cell cultures and isolated neurons reveal that overexpression of tau reduces MT binding to motor proteins and inhibits transport of cellular components, which leads to MT disruption and synaptic decay (17, 46). Conversely, enhancing MT acetylation leads to the recruitment of kinesin-1 and dynein motors to MTs, thereby stimulating anterograde and retrograde transport in vitro and in vivo (28, 47). Moreover, kinesin-1 stimulates MT elongation and maintains MT integrity by allowing the activation of c-Jun N-terminal kinase (48). It is therefore possible that increased MT acetylation rescues tau-induced MT defects by recruiting motor proteins to MTs, but this possibility needs to be confirmed.

In summary, our findings suggest that HDCA6 is a unique potential therapeutic target for tauopathies. Exploitation of HDCA6 suppression for eventual tauopathy treatment has several advantages. First, both HDAC6 mutant flies and knockout mice are viable and display no obvious developmental abnormalities (16, 49), greatly facilitating preclinical studies on HDCA6 function in tauopathy models. Second, there are a number of highly specific HDAC6 inhibitors available, such as tubacin and tubastatin A (26, 50). Moreover, these compounds have been studied extensively for their anticancer effects. Our current results suggest that these compounds may be potential neuroprotectants for the treatment of tauopathies. Third, because mutations of HDAC6 rescued both mutant and wild-type tau-induced MT defects, this potential treatment strategy could be applied to AD and tauopathies such as FTDP-17 that are associated with tau mutations. Finally, in AD brain neurons bearing neurofibrillary tangle, the staining intensity of acetylated α-tubulin is reduced (24). Furthermore, the HDAC6 protein level detected by Western analysis is significantly increased in the cortex and hippocampus of AD brains (25). Pharmacologically increasing MT acetylation by inhibition of HDAC6 or by increasing acetylase activity may antagonize the disruptive effects of tau on MTs. Such a drug could be a therapy alone or used as a complement to other emerging therapies for AD and related tauopathies that target other mechanisms of neurodegeneration.

Materials and Methods

All flies were cultured in standard cornmeal medium at 25 °C. Stocks used in this study were the muscle-specific C57–Gal4, the pan-neuronal elav–Gal4, and Df(1)ED7294 (Bloomington Stock Center). Upstream activation sequence (UAS)–tau^WT and UAS-tau^V337M were from M. Feany (Harvard Medical School, Boston). UAS–HDAC6 and an HDAC6 null (HDAC6^KO) were from R. Jiao (Institute of Biophysics, CAS, Beijing) (16).

Details of generation of transgenic flies, immunohistochemistry, HDAC6 inhibitor treatment, and Western blotting are described in SI Materials and Methods.