SIRT1 regulates O-GlcNAcylation of tau through OGT

Tau is modified with O-GlcNAcylation extensively in human brain. The O-GlcNAcylation levels of tau are decreased in Alzheimer’s disease (AD) brain. Sirtuin type 1 (SIRT1) is an enzyme that deacetylates proteins including transcriptional factors and associates with neurodegenerative diseases, such as AD. Aberrant SIRT1 expression levels in AD brain is in parallel with the accumulation of tau. cAMP response element binding protein (CREB), a cellular transcription factor, plays a critical role in learning and memory. In this present study, we found SIRT1 deacetylates CREB and inhibits phosphorylation of CREB at Ser133. The inactivated CREB suppresses OGT expression and therefore decreases the O-GlcNAcylation of tau and thus increases the phosphorylation of tau at specific sites. These findings suggest that SIRT1 may be a potential therapeutic target for treating tauopathies.


INTRODUCTION
Tau is encoded by the microtubule-associated protein tau (MAPT) gene located on the long arm of human chromosome 17 [1]. Tau contributes to tubulin assembly and stabilization of microtubules, thereby regulating normal functions of neurons [2].
O-GlcNAcylation, a dynamic posttranslational modification (PTM), is a single N-acetylglucosamine (GlcNAc) attached to serine and threonine residues of nuclear and cytoplasmic proteins [6]. Human brain tau is modified with O-linked N-acetylglucosamine (O-GlcNAc) [7]. A marked decrease of O-GlcNAcylation of tau protein has been showed in human AD brain [8]. The down-regulation of tau O-GlcNAcylation in AD can cause tau pathology and neurodegeneration [9]. The "Yin-Yang" reciprocal relationships between O-GlcNAcylation and phosphorylation have been suggested since phosphorylation and O-GlcNAcylation occur on the same Ser/Thr site of protein [10]. It has been shown that O-GlcNAcylation negatively regulates tau phosphorylation in a site-specific manner in cultured cells, in vivo and in metabolically active brain slices [11,12]. Tau-mediated neurodegeneration may be promoted by decreased O-GlcNAcylation through abnormal hyperphosphorylation and oligomerization of tau [9]. The only processing enzymes of O-GlcNAcylation are O-GlcNAc transferase (OGT) and O-GlcNAcase (OGA) [7]. OGT catalyzes the addition of N-acetylglucosamine to serine or threonine hydroxyl groups of target proteins [13]. OGA removes O-GlcNAc from proteins [14][15][16].
CREB is a cAMP-responsive transcription factor regulating the somatostatin gene [17]. CREB binds to the cAMP response element (CRE) in the promoters of its target genes, upon phosphorylation at Ser133 by different protein kinases, such as protein kinase A (PKA), calmodulin-dependent protein kinase (CaMK), mitogenactivated protein kinases (MAPK), and so on [18]. Transcription of memory-associated genes is associated with CREB phosphorylation, disruption of which in AD results in memory impairment [19]. Deacetylation prevents cAMP-dependent phosphorylation of CREB to affect its activity [20].
Whether and how SIRT1 regulates O-GlcNAcylation and phosphorylation of tau remain elusive. In the present study, we investigated the role of SIRT1 in OGT expression. We found that SIRT1 deacetylated CREB to suppress OGT expression and therefore to decrease the O-GlcNAcylation of tau and increase the phosphorylation of tau at specific sites.

SIRT1 reduces O-GlcNAcylation modification of tau
SIRT1 influences glucose metabolism [24]. To investigate whether the O-GlcNAcylation level of tau is regulated by SIRT1, GFP-Tau441 was co-overexpressed together with Myc-SIRT1 or Myc-H363Y, the deacetylase-dead mutation of SIRT1, in HEK-293A or COS7 cells. After 48 h, we purified GFP-tau protein from cell lysates by immunoprecipitation using anti-GFP antibody. We found that SIRT1 but not H363Y could decrease the O-GlcNAcylation level of tau protein significantly in both cell lines ( Figure 1A). In the mean time, we transfected shSIRT1 plasmids into HEK-293T cells to knockdown endogenous SIRT1 level and detected the O-GlcNAcylation level of immunoprecipitated GFP-tau. We found the O-GlcNAcylation level of GFP-tau was significantly increased due to decline of SIRT1 expression ( Figure  1B). These data suggested that SIRT1 could reduce O-GlcNAcylation modification of tau in cells.
To determine whether the altered O-GlcNAcylation may mediate the phosphorylation of tau, purified tau from HEK-293A cells was immunostained by using antibodies that recognize phosphorylated tau at their specific epitopes. We observed increased tau phosphorylation at Ser199 and Ser214 but not Thr212 due to SIRT1 overexpression (Figure 2A, 2B). However, these changes were not obviously observed in H363Y transfected cells. To further confirm the observations, we determined the phosphorylation levels of endogenous tau in the E18 rat cerebral cortical neurons. We detected the decreased phosphorylation levels of tau at Ser199 and Ser214 by infecting neurons with lentiviral-shSIRT1 ( Figure 2C, 2D). These data strongly suggest that the decrease in O-GlcNAcylation of tau protein may be accompanied by hyperphosphorylation of tau at some phosphorylation sites.

SIRT1 inhibits the expression of OGT
The O-GlcNAc transferase (OGT) regulates the O-GlcNAc modification on tau proteins [29]. To determine whether SIRT1 controls the mRNA and protein levels of OGT, we transfected HEK-293A cells with pcDNA3.1/Myc-SIRT1 or pcDNA3.1/Myc-H363Y. As expected, SIRT1 reduced the mRNA level of OGT ( Figure 3A, 3B). Additionally, we detected the changes of OGT protein levels in HEK-293A cells transfected with SIRT1 or H363Y plasmids. We found the OGT protein level was decreased significantly due to SIRT1 overexpression, whereas the H363Y transfection has little impact ( Figure 3C, 3D). These results suggest that SIRT1 may regulate the expression of OGT both at the mRNA and protein levels.

SIRT1 negatively regulates the expression of luciferase driven by OGT promoter
To understand the molecular mechanisms underlying OGT expression regulation, the promoter of the human OGT gene was analyzed by MatInspector software analysis [30,31], a Genomatix internationally renowned program for the identification of transcription factor AGING binding sites. The bioinformatic analysis revealed an array of putative nuclear factor binding sites, especially several potential CRE-like elements ( Figure 4A), suggesting CREB may be involved in regulating OGT expression. To investigate the transcriptional regulation of OGT, we inserted the promoter region of human OGT, -1463 ~ +37, into pGL3-basic vector to generate reporter plasmid of pGL3/OGT1500. We transfected pGL3/OGT1500 together with pRL-TK into HEK-293A cells and then measured luciferase activity using the dual luciferase assay. We found that pGL3/OGT1500 increased luciferase activity by almost 30-fold ( Figure 4A) compared with pGL3-basic, indicating that the promoter of human OGT drives the luciferase expression. To detect the effect of SIRT1 on OGT transcription, pcDNA3.1, pcDNA3.1/Myc-SIRT1 or pcDNA3.1/Myc-H363Y was co-transfected with pGL3/OGT1500 into HEK-293A cells. Cell lysates were analyzed using western blots and luciferase activity assay. We observed a half decrease in luciferase activity in pcDNA3.1/Myc-SIRT1 transfected HEK-293A cells compared with the control transfected cells ( Figure 4B), suggesting that overexpression of SIRT1 but not H363Y inhibited the expression of luciferase driven by OGT promoter. These results indicate that SIRT1 may suppress OGT transcription.

CREB takes part in the expression regulation of OGT
Several CRE-like elements appeared densely in the OGT promoter according to MatInspector software analysis ( Figure 4A). Therefore, we want to explore whether CREB participates in the regulation of OGT expression. We overexpressed CREB or knocked down CREB by siRNA of CREB in HEK-293A cells. The protein levels of OGT were examined by western blots using anti-OGT antibody and mRNA levels of OGT were determined by RT-PCR from total RNA. We observed that both mRNA and protein levels of OGT were declined when CREB were overexpressed ( Figure 5A, 5B). In the contrary, CREB knockdown elevated OGT expression at mRNA and protein level ( Figure 5A, 5B). These data suggested that CREB may play a pivotal role in the expression regulation of OGT.
To investigate the molecular mechanisms by which CREB involved in the transcription regulation of OGT, we transfected pGL3/OGT1500 alone or together with pCI/HA-CREB into HEK-293A cells and then immunoprecipitated CREB with anti-HA antibody from  There are six CRE-like elements within human OGT promoter region -1463 ~ +37 bp. pGL3/ OGT1500 or pGL3-basic vector was co-transfected together with pRL-Tk into HEK-293A cells. After 48 h transfection, the luciferase activity was measured and normalized with Renilla luciferase (pRL-TK). The relative activity of luciferase was presented as mean ± S.D., **, p < 0.01. (B) pGL3/ OGT1500 or pGL3-basic vector was co-transfected together with pRL-Tk into HEK-293A cells overexpressing Myc-SIRT1 or Myc-H363Y. After 48 h transfection, the luciferase activity was measured and normalized with Renilla luciferase (pRL-TK). The relative activity of luciferase was presented as mean ± S.D., **, p < 0.01. AGING Figure 5. CREB is involved in the regulation of OGT expression. HEK-293A cells were transfected with pCI-neo, pCI/HA-CREB or siRNA of CREB respectively. (A) mRNA levels of OGT and GAPDH were measured by RT-PCR. The quantification of relative mRNA level of OGT after normalization with the mRNA level of GAPDH was represented as mean ± S.D. (n = 3); *, p < 0.05, **, p < 0.01. (B) Protein levels of OGT, CREB and GAPDH were examined by western blot using anti-OGT, anti-CREB and anti-GAPDH antibody. Relative OGT or CREB level was quantified after normalization with the protein level of GAPDH and presented as mean ± S.D. (n=3), *, p < 0.05; **, p < 0.01. (C) pCI/HA-CREB was cotransfected with pGL3/OGT1500 into HEK-293T cells. HA-CREB was immunoprecipitated with anti-HA antibody. Co-immunoprecipitated DNA of OGT promoter with CREB was determined by PCR with three sets of primers specific to CRE elements as indicated for amplifying the DNA. The PCR product was separated by agarose electrophoresis. (D) In HEK-293A cells transfected with pGL3/ OGT1500 and pRL-TK, pCI-neo, pCI/HA-CREB or siRNA of CREB was transfected with or without overexpression of SIRT1. After 48 h transfection, the luciferase activity was measured and normalized with Renilla luciferase. The relative activity of luciferase was presented as mean ± S.D. (n=3), *, p < 0.05, **, p < 0.01, ***, p< 0.001 versus control group; ###, p< 0.001. (E) pCI-neo, pCI/HA-CREB or siRNA of CREB was transfected into HEK-293A cells overexpressing SIRT1 or not. Protein levels of SIRT1, OGT, CREB or GAPDH were detected by western blot using anti-Myc, anti-OGT, anti-CREB or anti-GAPDH antibody. Relative OGT or CREB level was quantified after normalization with the protein level of GAPDH and presented as mean ± S.D. (n=3), *, p < 0.05; **, p < 0.01; ***, p < 0.001 versus control group; ##, p< 0.01.
AGING the cell lysates. The co-immunoprecipitated DNA of OGT promoter with CREB by anti-HA was amplified with PCR by using three sets of primers against different regions (-1446 to -734, -839 to -487, -476 to +32) of OGT promoter respectively ( Figure 5C, upper panel). We observed that only -1446 to -734 of the OGT promoter was co-immunoprecipitated with CREB ( Figure 5C, lower panel), suggesting that CREB could act on the promoter of OGT.
Domenico Accili et al. demonstrated that CREB is deacetylated by SIRT1 in SirBACO mice [20]. It is possible that SIRT1 modulates the expression of OGT dependent on CREB. To verify the speculation, we knocked down CREB by siRNA merely or overexpressed SIRT1 simultaneously in HEK-293A cells which were transfected with pGL3/OGT1500. Then the dual luciferase assay was employed to measure luciferase activity. The results showed that knockdown of CREB reversed the inhibition effect of SIRT1 on the luciferase activity driven by OGT promoter ( Figure 5D).
To further convince the function of CREB in OGT transcription, we transfected siRNA of CREB alone or together with pcDNA3.1/Myc-SIRT1 into HEK-293A cells. Endogenous OGT was detected by western blot using anti-OGT antibody after 48 h transfection. SIRT1 couldn't decrease the protein level of OGT when CREB was knocked down ( Figure 5E). These results suggested that SIRT1 may act on CREB to negatively regulate the OGT expression.

SIRT1 physically interacts with CREB
To address whether there is interaction between SIRT1 and CREB. GST-pull down was used to detect the protein-protein interaction in vitro. We found that SIRT1 overexpressed in HEK-293T cells was pulled down with GST-CREB, but not with GST itself ( Figure 6A). To further validate the interaction between SIRT1 and CREB, SIRT1 with Myc-tag and CREB with HA-tag were overexpressed in HEK293A cells and CREB was immunoprecipitated with anti-HA antibody. Then we analyzed whether SIRT1 was co-immunoprecipitated with CREB by western blot using anti-Myc antibody. We detected that Myc-SIRT1 was co-immunoprecipitated with HA-CREB in the immunocomplex ( Figure 6B). These data indicate that SIRT1 physically interacted with CREB which is consistent with the results presented by Domenico Accili et al. [20].

SIRT1 deacetylates CREB and inactivites CREB
SIRT1 influences the expression of OGT via CREB. To investigate the role of SIRT1 in CREB acetylation, we co-transfected pCI/HA-CREB with pcDNA3.1/Myc-SIRT1 or pcDNA3.1/Myc-H363Y and immunoprecipitated CREB with anti-HA antibody. The acetylation level of immunoprecipitated CREB was analyzed by western blot developed with anti-acetylatedlysine antibody. As shown in Figure 7A and 7B, the level of acetylated CREB was significantly decreased in the cells with overexpression of SIRT1, whereas overexpression of H363Y, the deacetylase-dead mutant (B) SIRT1 could be co-immunoprecipitated by HA-CREB using anti-HA antibody. CREB tagged with HA and SIRT1 tagged with Myc were co-expressed in HEK-293A cells for 48 h. The cell extract was incubated with anti-HA antibody coupled onto protein G beads. The bound proteins were subjected to western blots using anti-HA antibody to CREB and anti-Myc antibody to SIRT1. AGING of SIRT1, had no effect on the acetylation level of CREB. These results indicate that SIRT1 deacetylates CREB.
A causal link has been established between phosphorylation of CREB at Ser133 and trans-activation of cAMP-responsive genes [32]. Since SIRT1 interacted with and deacetylated CREB, we ask whether SIRT1 affects the activity of CREB. The Immunopurified CREB was also used to test the phosphorylation level of CREB at Ser133 by western blot using anti-pS133-CREB antibody. We found that overexpression of SIRT1 but not H363Y decreased the phosphorylation level of CREB at Ser133 ( Figure 7A, 7C), these data are consistent with the findings of Domenico Accili et al. [20].

DISCUSSION
Elevation of tau O-GlcNAcylation was suggested as pharmacological means to attenuate the formation of tau paired helical filament via decreasing tau phosphorylation [7]. In AD patient's brain, glucose uptake/metabolism is impaired which contributes to neurodegeneration via decline of O-GlcNAcylation and abnormal hyperphosphorylation of tau [33]. GLUT1 and GLUT3 are the two major brain glucose transporters (GLUT) which are responsible for the transport of glucose from blood into the neuron. Studies also show that the reduction of GLUT1 and GLUT3 is associated with the decrease in O-GlcNAcylation and hyperphosphorylation of tau in AD brain [34]. In the present study, we show for the first time that SIRT1 downregulates the O-GlcNAcylation of tau by controlling the expression of OGT.
Sirtuins are important in regulating and maintaining glucose and lipid homeostasis [35]. SIRT1, the most well studied member of the mammalian sirtuin family, plays a pivotal role in metabolism, including mitochondrial biogenesis, glycolysis, hypoxia and angiogenesis [36]. SIRT1 is a key regulator in glucose and lipid metabolism by deacetylating certain proteins through its deacetylase activity [37]. We provide the evidence that SIRT1 negatively influence the O-GlcNAcylation level of tau protein via its deacetylase activity. The H363Y form does not alter the mRNA AGING levels of OGT, yet affects the protein levels of OGT ( Figure 3). We can't detect the interaction between SIRT1 and OGT (data not shown). The real reasons need to be investigated further.
Several transcriptional factors involved in transcriptional regulation of key genes in a wide range of metabolic activities are deacetylated by SIRT1, such as nuclear factor-kappa beta (NFκβ) [38], extracellular signal regulated kinase (ERK) [39], the forkhead box subgroup O (FoxO) family [40], peroxisome proliferator-activated receptors γ (PPARγ), and its transcriptional coactivator PPARγ coactivator 1-α (PGC-1α) [41,42]. CREB, a cellular transcription factor, participates in hepatic gluconeogenesis [43] and lipid synthesis [44]. There are mammalian studies report that the CREB family both activates and represses transcription by a similar mechanism [45]. It has been reported that SIRT1 directly deacetylates CREB at Lys136 to modulate its protein kinase A (PKA)-dependent phosphorylation in transgenic mice overexpressing SIRT1 (SirBACO) [20]. Present study further convinces the interaction between SIRT1 and CREB. Our data also support that SIRT1 deacetylates CREB and decreases phosphorylation level of CREB at Ser133 in cells.
O-GlcNAcylation negatively regulated site-specific phosphorylation of tau [11]. Aberrance of O-GlcNAcylation would result in hyperphosphorylation of tau [35,46]. In this study, we show that SIRT1 inhibits the O-GlcNAcylation of tau and gives rise to the increased phosphorylation level of tau at Ser199, Ser214 but not Thr212 in which Ser199 is mainly modulated by GSK3β and Ser 214 is reported to be modulated by CDK5 [47]. However, the exact O-GlcNAcylation sites of tau are still not well mapped.
In summary, SIRT1 interacts with and deacetylates CREB, leading to CREB inhibition by reducing the phosphorylation level of CREB at Ser133. Repressed CREB inhibits the expression of OGT, resulting in decreasing O-GlcNAcylation of tau and increasing phosphorylation of tau at specific site. Our results provide a novel insight into molecular mechanisms of the regulation of tau phosphorylation. These findings suggest that SIRT1 may be potential therapeutics target for treating tauopathies.

Primary neuronal cell culture and transduction of viral vectors
Sprague Dawley (SD) rat brains were obtained at embryonic day 18 (E18). The E18 SD rat forebrains were dissected and dissociated in trypsin (0.05% with EDTA) and DNAse (10 μg/mL) and then resuspended in neurobasal medium with 2% B27 and penicillin (20 U/ml) and streptomycin (5 μg/ml) (Gibco, Rockville, MD, USA). For western blots experiments, 6-well plates were seeded with ~1×10 6 neurons per well. Half of the medium was replaced by fresh medium after 3 days. Then the cerebral cortical neurons were infected with lentiviral-shSIRT1 or its empty vectors (Genepharma, shanghai, China). The virus-containing medium was half-refreshed 24 h later and then subjected for western blot analysis after 72 h.

Plasmids and antibodies
pGL3-basic, pRL-Tk (thymidine kinase promoter driven Renilla luciferase), and dual luciferase assay kit were bought from Promega (Madison, WI, USA). pCI/GFP-tau441 was a gift from Dr. Fei Liu (Department of Neurochemistry, New York State Institute for Basic Research in Developmental Disabilities). shSIRT1 plasmid was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Mouse monoclonal anti-SIRT1, anti-acetylated-lysine antibody and polyclonal anti-pS133-CREB were from Cell Signaling Technology (Danves, MA, USA). Polyclonal anti-CREB, rabbit polyclonal anti-HA, mouse monoclonal anti-HA and anti-GAPDH antibody were from Sigma (St. Louis, MO, USA). Anti-GFP was a product from Proteintech (Rosemont, IL, USA). RL2 was bought from Affinity BioReagents (Affinity Bioreagents, Golden, CO, USA) and was used for O-GlcNAc detection. Levels of tau phosphorylation at each specific site were determined by using phosphorylationdependent and site-specific tau antibodies from BioSource International (Camarillo, CA, USA), and total tau level was determined by using R134d, a phosphorylation-independent pan-tau polyclonal antibody produced in laboratory at the New York State Institute for Basic Research in Developmental Disabilities by using the longest human recombinant tau isoform to immune rabbit [48]. Mouse monoclonal anti-Myc, rabbit anti-GAPDH and human siRNA of CREB were purchased from Santa Cruz (Santa Cruz, CA, USA). Peroxidase-conjugated anti-mouse and antirabbit IgG were obtained from Jackson Immuno Research Laboratories (West Grove, PA, USA). ECL Kit was from Thermo Scientific (Rockford, IL, USA).

Plasmid construction
pcDNA3.1/SIRT1 was constructed by subcloning SIRT1 coding region which was PCR amplificated from Flag-SIRT1 plasmid purchased from Addgene (Cambridge, MA, USA) into mammalian expression vector pcDNA3.1 tagged with Myc at C-terminus by BamHI and NotI. Mammalian expression vector pCI/CREB AGING tagged with HA (hemagglutinin) at N-terminus was constructed as described previously [49]. pGEX-4T-1/CREB was constructed by subcloning CREB coding region from pCI/CREB. Luciferase driven by promoter of human OGT in pGL3-basic was constructed and confirmed by sequencing.

Reverse-transcription PCR
Total RNA was isolated from cultured cells by using the RNeasy Mini kit (Qiagen, Valencia, CA, USA) and used for first-strand cDNA synthesis with Oligo-(dT)15-18 using the RT2 First-Strand Kit (Invitrogen, CA, USA) according to the manufacturer's instructions. The cDNA of OGT or GAPDH was amplified by PCR using PrimeSTAR HS DNA Polymerase (Takara, Shiga, Japan) at 98 °C for 5 min, 98 °C for 10 s, at 55 °C for 15 s, and at 72 °C for 30 s for 30 cycles and then at 72 °C for 10 min for extension. The PCR products were resolved on 1 % agarose gels and viewed using the Molecular Imager system (Bio-Rad, Hercules, CA, USA). The primers used for this study are as follows: OGT forward 5'-AGG CAG TTC GCT TGT ATC GT-3' and reverse 5'-TAG AGT AGG CAT CAG CAA AGG T -3'; GAPDH forward 5'-GGT GG T CTC CTC TGA CTT CAA CA -3' and reverse 5'-GTT GCT GTA GCC AAA TTC GT T GT -3'.
Plasmid construction of luciferase reporter driven by human tau promoter and luciferase assay A 1.5-kb fragment of the human OGT genomic DNA from −1463 -+37 bp was amplified by PCR and cloned into pGL3-basic (Promega) by Kpn I and Sma I (New England Biolabs, Ipswich, MA, USA) to generate pGL3/OGT1500. The sequence and orientation of the clone was confirmed by DNA sequence analysis. HEK-293A cells were co-transfected with pcDNA3.1, pcDNA3.1/Myc-SIRT1, or pcDNA3.1/Myc-H363Y and pGL3/OGT1500, or their control vectors and pRL-Tk for 48 h. The cells were lysed using the passive lysis buffer (Promega). The luciferase activity was measured by the dual luciferase assay kit (Promega) according to manufacturer's manuals. The firefly luciferase activity and Renilla luciferase activity were measured subsequently and the firefly luciferase activity was normalized with Renilla luciferase activity.

Co-immunoprecipitation
HEK-293A cells were co-transfected with pCI/HA-CREB and pcDNA3.1/Myc-SIRT1 or pcDNA3.1/Myc-H363Y for 48 h. The cells were washed twice with phosphate-buffered saline (PBS) and lysed by sonication in lysate buffer (50 mM Tris-HCl, pH7.4, 150 mM NaCl, 50 mM NaF, 1 mM Na3VO4, 2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 2 μg/ml aprotinin, 2 μg/ml leupeptin, and 2 μg/ml pepstatin). Insoluble materials were removed by centrifugation at 18,000 g for 10 min. Protein G beads were incubated with mouse anti-HA overnight at 4 °C, and then the antibody bound beads were incubated with the cell lysate at 4 °C for 4 h. The beads were washed with lysate buffer twice and with Tris-buffered saline (TBS) twice, and then bound proteins were eluted by boiling in 2× Laemmli sample buffer for 5 min. The eluted samples were subjected to western blot analysis with the indicated primary antibodies.

Immuno-affinity purification of tau and CREB
HEK-293A cells were transfected with pCI/GFP-tau441 or pCI/HA-CREB for 48 h. HEK-293T cells were transfected with shSIRT1 or control shRNA plasmid (Santa Cruz, CA, USA) together with pCI/GFP-tau441 for 48 h and followed by selection with 2 μg/ml puromycin for 72 h. The recombinant GFP-tau or HA-CREB was immunoaffinity-purified from cell extracts in RIPA buffer (50mM Tris-HCl, pH 7.4, 150mM NaCl, 1% NP-40, 2mM EDTA) with monoclonal anti-GFP or anti-HA antibody that was pre-linked covalently onto protein Gagarose beads (Pierce, Rockford, IL). The beads were washed twice each with lysate buffer and with TBS. The bound proteins were eluted by boiling in 2× Laemmli sample buffer. The eluted samples were subjected to western blots with the indicated primary antibodies.

Statistical analysis
Where appropriate, the data are presented as the means ± S.D. Data points were compared by the unpaired two-tailed Student's t-test for two groups' comparison, oneway ANOVA and two-way ANOVA. The calculated pvalues are indicated in the figures.