|Home | About | Journals | Submit | Contact Us | Français|
Intracytoplasmic proteinaceous inclusions, primarily composed of tau or α-synuclein (α-syn), are predominant pathological features of Alzheimer’s disease (AD) and Parkinson’s disease (PD), respectively. However, the co-existence of these pathological aggregates is identified in many neurodegenerative disorders, including spectrum disorders of AD and PD. While α-syn can spontaneously polymerize into amyloidogenic fibrils, in vitro, tau polymerization requires an inducing agent. The current study presents a human-derived cellular model, in which recombinant, pre-formed α-syn fibrils cross-seed intracellular tau to promote the formation of neurofibrillary tangle-like aggregates. These aggregates were hyperphosphorylated, Triton-insoluble, and thioflavin S-positive, either co-mingling with endogenously expressed α-syn aggregates, or induced by only exogenously applied recombinant α-syn fibrils. Further, filamentous, amyloidogenic tau took over the cellular soma, displacing the nucleus and isolating or displacing organelles, likely preventing cellular function. While a significant proportion of wild-type tau formed these cellular inclusions, the P301L mutation in tau increased aggregation propensity resulting from α-syn seeds to over 50% of total tau protein. The role of phosphorylation on the development of these tau aggregates was investigated by co-expressing glycogen synthase kinase 3 beta or MAP/microtubule affinity-regulating kinase 2. Expression of either kinase inhibited the formation of α-syn-induced tau aggregates. Analyses of phosphorylation sites suggest that multiple complex factors may be associated with this effect, and that Triton-soluble versus Triton-insoluble tau may be independently targeted by kinases. The current work not only provides an exceptional cellular model of tau pathology, but also examines α-syn-induced tau inclusion formation and provides novel insights into hyperphosphorylation observed in disease.
Accumulations of brain proteinaceous, hyperphosphorylated inclusions, composed of amyloidogenic, misfolded protein are major pathological features of Alzheimer’s disease (AD) and Parkinson’s disease (PD). Intracellular localization of these pathological aggregates can be identified in these disorders, primarily composed of tau forming neurofibrilliary tangles (NFTs) in AD or of α-synuclein (α-syn) forming Lewy pathology in PD. The aberrant polymerization and accumulation of tau or α-syn additionally classifies a host of other neurodegenerative disorders, termed as tauopathies and synucleinopathies, respectively (Buee et al., 2000; Duda et al., 2000; Goedert et al., 1998; Lee et al., 2001).
Familial mutations in MAPT and SCNA (the genes for tau and α-syn, respectively) have also been identified that are causative of disease. Genetic mutations in tau are causative of the tauopathy frontotemporal dementia with parkinsonism linked to chromosome 17 (FTDP-17) (Hutton et al., 1998; Lee et al., 2001; Goedert and Spillantini, 2006), and mutations in α-syn result in familial forms of PD (Polymeropoulos et al., 1997). Mutations in these proteins increase the propensity of filamentous aggregate formation in vitro and in transgenic animal models (Arrasate et al., 1999; Conway et al., 1998; Giasson et al., 1999; Giasson et al., 2002; Lewis et al., 2000), further supporting the association between these proteins and disease.
Tau and α-syn amyloidogenic fibrils are both nucleation-dependent (Congdon et al., 2008; Wood et al., 1999) and hyperphosphorylated in their pathological forms (Fujiwara et al., 2002; Kahle et al., 2002; Anderson et al., 2006; Nishie et al., 2004; Neumann et al., 2002; Waxman and Giasson, 2008; Billingsley and Kincaid, 1997; Buee et al., 2000). Antibodies recognizing phospho-epitopes S396/S404 (PHF1) and S202/T205 (AT8) on tau are among the most characterized identifiers of NFTs (Otvos, Jr. et al., 1994; Goedert et al., 1995), and phosphorylation of these sites by glycogen synthase kinase 3 beta (GSK3β) may promote tau aggregate formation (Sato et al., 2002; Rankin et al., 2007; Liu et al., 2007). However, MAP/microtubule affinity-regulating kinase 2 (MARK2) targets the core, repeat region required for tau filament formation, potentially inhibiting aggregation (Drewes et al., 1997; Drewes et al., 1995; Crowther et al., 1989; Crowther et al., 1992; Goode and Feinstein, 1994; Schneider et al., 1999).
While α-syn can spontaneously self-polymerize in vitro (Conway et al., 1998; Giasson et al., 1999), tau requires the presence of an inducing agent (Goedert et al., 1996). α-Syn induces tau polymerization in vitro (Giasson et al., 2003), and tau and α-syn pathology co-exist in patients with familial forms of PD and in spectrum diagnoses between PD and AD, termed dementia with Lewy bodies (DLB) and Lewy body variant of Alzheimer’s disease (LBVAD) (Duda et al., 2002; Hansen et al., 1990; McKeith et al., 1996; Giasson et al., 2003). However, the manner in which these proteins interact to promote aggregation has yet to be modeled in vivo.
The current work provides the first cellular modeled system, mimicking co-existing tau and α-syn pathology. Using previously described methods for cellular seeding (Waxman and Giasson, 2010), recombinant, pre-formed α-syn fibrils robustly promoted intracellular tau aggregation with filamentous morphology with similar characteristics to that observed post-mortem. We further characterized the hyperphosphorylation of tau and the effects of kinase overexpression within this model system.
The human α-syn cDNA was cloned into the bacterial expression vector pRK172, using Nde I and Hind III restriction sites. The pRK172 DNA construct expressing N-terminal truncated 21–140 α-syn (with a Met codon added before amino acid 21) was generously provided by Dr. Virginia Lee (University of Pennsylvania, Philadelphia, PA). α-Syn proteins were expressed in E. coli BL21 (DE3) and purified as previously described (Giasson et al., 2001;Greenbaum et al., 2005). Human full-length tau cDNA (2N/4R isoform) was cloned into the bacterial expression vector pRK172, using NdeI and EcoRI restriction sites (kindly provide by Dr. Michel Goedert). pRK172 plasmid expressing human tau with the P301L mutation was created with oligonucleotides corresponding to the amino acid substitutions by QuickChange site-directed mutagenesis (Stratagene, La Jolla, CA). Recombinant full-length wild-type (WT) or P301L tau was expressed in E. coli BL21 and purified, as previously described (Giasson et al., 2003; Hong et al., 1998).
For cellular experiments, α-syn proteins were assembled into filaments by incubation at 37ºC at concentrations greater than 5 mg/ml in sterile phosphate buffered saline (PBS, Invitrogen) with continuous shaking at 1050 rpm (Thermomixer R, Eppendorf, Westbury, NY). Experimentation was planned so that α-syn would be visibly assembled (by filamentous clusters observed in the solution) by the day of cellular experimentation. α-Syn fibrils were diluted to a concentration of 1–3 mg/ml in sterile PBS and treated by water bath sonication for a minimum of 2 hours. In experiments where large fibrils were separated from small oligomers, diluted fibrillized α-syn (prior to sonication) was centrifuged at 16,000 x g for 5 min, and the supernatant was removed and the pellet was re-suspended in sterile PBS. Protein concentrations of the original diluted fibrils and the supernatant were measured by BCA protein assay (Pierce). Concentration of the large fibrils was determined as the concentrations of [total - supernatant]. All samples were sonicated in a water bath prior to addition to cell culture media. Cells expressing both α-syn and tau were treated with 1 μM of recombinant 21–140 α-syn fibril mix, and cells expressing only tau were treated with 1 μM of wild-type α-syn fibril mix, unless otherwise specified.
QBI293 cells, derived from human embryonic kidney cells, were maintained using Dulbecco’s Modified Eagle’s Medium (DMEM; Invitrogen) supplemented with 10% fetal bovine serum (FBS) and 100 U/ml penicillin/100 μg/ml streptomycin. The mammalian-expression vector pcDNA3.1 cloned with WT human α-syn cDNA was previously described (Paxinou et al., 2001). The pcDNA3.1 expression plasmid containing cDNA for human WT full-length tau (2N/4R) or tau containing the P301L mutation was generously provided by Dr. Virginia Lee (University of Pennsylvania, Philadelphia, PA). The S262A and S356A mutations in WT or P301L tau were created with oligonucleotides corresponding to the amino acid substitutions by QuickChange site-directed mutagenesis (Stratagene). The following amino acid substitutions were completed by serial QuickChange reactions: S262A/S356A, S202A/T205A, S202E/T205E. All mutations and the absence of copy errors were confirmed by sequencing the entire length of the tau cDNA. The mammalian-expression vector pCMV, cloned with myc-tagged MARK2 cDNA was generously provided by Dr. Marina Picciotto (Yale University, New Haven, CT). Human HA-tagged GSK3β cDNA in pcDNA3 was generated by Dr. Jim Woodgett (He et al., 1995) and acquired from Addgene Inc. (Cambridge, MA).
Cells were plated onto poly-D-lysine coated 6-well plates and transfected at approximately 30% confluency, using calcium phosphate precipitation, as previously described (Waxman and Giasson, 2010). Cells were transfected with an equal proportion of tau-containing expression plasmid and α-syn-containing expression plasmid or pcDNA3.1 (empty vector), except in experiments where MARK2 over-expression was evaluated. In MARK2 experiments, cells were transfected with a 3:1 ratio of tau to MARK2 expression plasmids. Four hours after transfection, 1 μM (final concentration) of sonicated α-syn fibrils was added drop-wise to media, after which cells were incubated over night at 37ºC, 5% CO2.
Approximately 16 hours after transfection, calcium phosphate was washed two-times with PBS, and media was replaced with warm DMEM containing 3% FBS and 100 U/ml penicillin/100 μg/ml streptomycin (reduced serum). Cells were harvested for biochemical fractionation or fixed 72 hours after transfection, unless otherwise specified.
Cells were washed one time in ice-cold PBS, and samples were harvested in 25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 20 mM NaF, and a cocktail of protease inhibitors containing 1 mM phenylmethylsulfonyl and 1 mg/ml each of pepstatin, leupeptin, N-tosyl-L-phenylalanyl chloromethyl ketone, N-tosyl-lysine chloromethyl ketone and soybean trypsin inhibitor. Samples were sedimented at 100,000 x g for 30 min at 4ºC. Supernatants were removed and pellets were sonicated in 1.5X Laemmli sample buffer (75 mM Tris-HCl, pH 6.8, 3% SDS, 15% glycerol, 3.75 mM EDTA, pH 7.4). Sonication in Laemmli buffer was sufficient to reduce Triton-insoluble tau to monomeric form. SDS sample buffer (10 mM Tris, pH 6.8, 1 mM EDTA, 40 mM DTT, 1% SDS, 10% sucrose) was added and samples were heated to 100ºC for 5 min prior to Western blot analysis. Equal proportions of supernatant (Triton-soluble protein) and pellet (Triton-insoluble protein) were loaded and analyzed as percent pelleted.
Microtubule binding assays were performed, as previously described (Vogelsberg-Ragaglia et al., 2000). Briefly, cells were harvested with homogenization into pre-warmed 100 mM MES, pH 6.8, 1 mM EGTA, 0.5 mM MgSO4, 2 mM DTT, 0.75 M NaCl, 20 mM NaF, 0.1% Triton X-100, 20 μM taxol, 2 mM GTP, and a cocktail of protease inhibitors at 37ºC. After centrifugation for 20 min at 50,000 x g, the supernatant containing non-microtubule bound protein was removed from the microtubule bound fraction (pellet). SDS sample buffer was added and samples were heated to 100ºC for 5 min prior to Western blot analysis. Equal proportions of supernatant and pellet were loaded and analyzed as percent pelleted.
SNL4 is a polyclonal rabbit antibody raised against a synthetic peptide corresponding to amino acids 2–12 of α-syn (Giasson et al., 2000). pSer129 is a mouse monoclonal antibody that specifically recognizes phosphorylated of α-syn at S129 (Waxman and Giasson, 2008). Rabbit polyclonal anti-tau antibody 17025 (generously provided by Dr. Virginia Lee) or goat polyclonal anti-tau antibody (sc-1995, Santa Cruz Biotechnology, Santa Cruz, CA) was used to examine general tau immunoreactivity. 17025 was used for Western blot analyses and co-labeling with thioflavin S reactivity; sc-1995 was used for all other anti-tau double-immunofluorescence immunolabeling. AT8 (Thermo-fisher) is specific towards phosphorylation sites S202 and T205 in tau (Goedert et al., 1995), PHF1 (generously provided by Dr. Peter Davies, Albert Einstein University, NY, NY) is specific towards phosphorylation sites S396 and S404 in tau (Otvos, Jr. et al., 1994), and 12E8 (generously provided by Dr. Peter Seubert, Elan Pharmaceuticals, San Francisco, CA) specifically recognizes phosphorylation at S262 and S356 in tau (Seubert et al., 1995). Anti-myc antibody clone 9E10 (Sigma-Aldrich) was used to detect MARK2 overexpression. Additional antibodies include anti-vimentin (Sigma-Aldrich), anti-lamp1 (BD Biosciences, San Jose, CA), anti-γ-tubulin (Sigma-Aldrich), anti-β-tubulin (Sigma-Aldrich), and anti-mannosidase II (Chemicon).
Protein samples were resolved by SDS-PAGE on 8% gels for tau or MARK2 and 15% polyacrylamide gels for α-syn, followed by electrophoretic transfer onto nitrocellulose membranes. Membranes were blocked in Tris buffered saline (TBS) with 5% dry milk, and incubated overnight with antibodies SNL4, 9E10, or 17025 diluted in TBS/5% dry milk or phospho-specific antibodies diluted in TBS/5% bovine serum albumin (BSA). Each incubation was followed by goat anti-mouse conjugated horseradish peroxidase (HRP) (Amersham Biosciences, Piscataway, NJ) or goat anti-rabbit HRP (Cell Signaling Technology, Danvers, MA), and immunoreactivity was detected using chemiluminescent reagent (NEN, Boston, MA) followed by exposure onto X-ray film.
Double-immunofluorescence of transfected cells was completed as previously described (Mazzulli et al., 2006; Waxman et al., 2009). Cells were fixed at −20ºC with 100% MeOH for 20 min, followed by 50% MeOH and 50% acetone for 5 min. For Thioflavin S immunostaining, coverslips were fixed in 4% paraformaldehyde for 8 minutes, followed by PBS/0.1% Triton X-100 for 10 min. Following washes with PBS, coverslips were blocked with PBS containing 3% bovine serum albumin (BSA), 1% dry milk, and 1% fish gelatin, and primary antibodies were diluted into blocking solution for 1–2 h at room temperature. Following PBS washes, coverslips were incubated in secondary antibodies conjugated to Alexa488 or Alexa594 for 1 h. For double-immunofluorescence between pSer129 and PHF1 or AT8, secondary antibodies goat anti-mouse IgG1 conjugated to Alexa488 (for PHF1 or AT8) and IgG2A conjugated to Alexa594 (for pSer129) were used. Thioflavin S (Sigma-Aldrich) immunostaining was performed after secondary antibody incubation at a concentration of 0.05%, after which coverslips were washed three times in 70% ethanol followed by washes in water. Immunofluorescence of post-mortem brain samples was completed with paraffin-embedded brain sections as previously described (Giasson et al., 2006). Nuclei were counterstained with Hoechst trihydrochloride trihydrate 33342 (Invitrogen), and coverslips were mounted using Fluoromount-G (Southern Biotech, Birmingham, AL).
Low magnification double-immunofluorescence pictures were captured on an Olympus BX51 fluorescence microscope mounted with a DP71 digital camera (Olympus, Center Valley, PA). Confocal microscopy images were captured on a Zeiss Axiovert 200M inverted confocal microscope mounted with a Zeiss LSM510 META NLO digital camera utilizing Zeiss LSM510 META V3.2 confocal microscope software (Zeiss, Thornwood, NY). Confocal images were captured with 63x oil optics. Representative images were of one Z-plane of <0.7 μm, or 10–12 stacked Z-planes, where indicated, for 3-Dimensional (3D) projection images. 3D projection images were rotated on a central axis for optimal representation.
Cells that were transfected and, in some experiments, treated with recombinant α-syn fibrils as described above were fixed with 2% paraformaldehyde/2.5% glutaraldehyde 0.1M sodium cacodylate buffer, pH 7.4, at 4ºC overnight. Cells were harvested, pelleted and washed in the same buffer prior to secondary fixation in 2% osmium tetroxide. After rinses to remove osmium tetroxide, cells were stained en bloc in 2% aqueous uranyl acetate and then dehydrated in a graded ethanol series prior to infiltration and embedding in EMbed-812 (Electron Microscopy Sciences, Fort Washington, PA). Epone blocks were cut and sections were stained with uranyl acetate and lead citrate. Sections were examined with a JEOL 1010 transmission electron microscope (Peabody, MA) fitted with a Hamamatsu digital camera (Bridgewater, MA) and AMT Advantage image capture software (Danvers, MA). For immuno-EM, cells were fixed for 60 min in 4% paraformaldehyde/0.25% glutaraldehyde in PBS, permeabilized with ethanol, blocked with 5% horse serum/1% BSA/0.2% cold water fish skin gelatin in PBS, incubated with anti-tau antibody17025 or anti-α-syn antibody SNL4 diluted in blocking solution, followed by incubation with a goat anti–rabbit HRP-conjugated antibody (Santa Cruz Biotechnology, Inc.), and developed with 3,3′-diaminobenzidine. The chemical product was enhanced using a modification of the Rodriguez silver/gold enhancement method (Teclemariam-Mesbah et al., 1997). Cells were then further fixed in 2% paraformaldehyde/2.5% glutaraldehyde/0.1M sodium cacodylate buffer, pH 7.4 and prepared for ultrastructural analysis as described above, with the omission of lead citrate staining.
Recombinant-generated α-syn proteins were pre-assembled into filaments by incubation at 37ºC in 100 mM sodium acetate, pH 7.4 with continuous shaking at 1050 rpm (Thermomixer R, Eppendorf). After assembly, α-syn protein was treated by waterbath sonication for a minimum of 2 hours. After which, pre-fibrillized, sonicated α-syn was diluted to 0.25 mg/ml in 100 mM sodium acetate, pH 7.4 and incubated with tau (WT or P301L) at 1 mg/ml or recombinant, non-fibrillized α-syn at 0.3 mg/ml for 3 days at 37ºC, 1050 rpm. A fraction of each sample was set aside for K114 fluorometry. The remainder of each sample was centrifuged at 100,000 x g for 30 min. SDS-sample buffer was added to pellets and supernatants, which were heated to 100ºC for 10 min. Equal volumes of supernatants and pellets were separated by SDS-PAGE and were quantified by densitometry of Coomassie Blue R-250 stained gels.
Assembled α-syn polymers were absorbed onto 300 mesh carbon coated copper grids, stained with 1% uranyl acetate and visualized with a Joel 1010 transmission electron microscope. Images were captured with a Hamamatsu digital camera using AMT software.
α-Syn and tau fibrils are amyloidogenic, and their formation can be quantified using the fluorescent amyloid binding dye K114 (Crystal et al., 2003; Conway et al., 2000). K114 is derived from the structure of Congo Red and demonstrates a tremendous increase in fluorescence in solution assays upon binding to amyloidogenic fibrils (Crystal et al., 2003). This assay was conducted, as previously described (Crystal et al., 2003), by incubating a fraction of each sample with the K114 (50 μM) in 100 mM glycine, pH 8.6 and measuring fluorescence (λex=380 nm, λem=550 nm, cutoff = 530 nm) with a SpectraMax Gemini fluorometer and SoftMax Pro 4.0 software.
Cells were counted by capture and counting of 1–4 random fields per experiment, independently assessing immunofluorescence of AT8, PHF1, or anti-tau antibodies. Western blot data were quantified by ImageJ software (NIH, Bethesda, MD). Changes in Triton-insoluble protein were calculated as [Triton-insoluble/(soluble+Triton-insoluble)]*100, and direct comparisons were performed by paired t-tests, using GraphPad InStat software (San Diego, CA). Additional data between two groups were analyzed by two-way, parametric t-tests or by non-parametric t-tests or U-tests (in cases of significantly different variances). Analyses of phospho-specific immunoreactivity were performed by standardizing densitometry to total anti-tau immunoreactivity (17025) and the control (normally WT tau in the absence of fibril treatment) condition. Comparisons to the control condition after standardization were performed by one-sample t-tests. Comparisons between multiple groups were performed by one-way ANOVA, followed by Bonferroni post-test analyses. Each experiment was performed a minimum of three independent times.
Recent evidence suggests that α-syn is capable of membrane translocation and recycling (Liu et al., 2009; Ahn et al., 2006) and fibrillized α-syn may be transferred between neurons (Lee et al., 2008; Desplats et al., 2009). Based on this paradigm, we have recently developed a model system of cellular α-syn aggregation, using preformed, recombinant α-syn fibrils to seed intracellular α-syn aggregate formation (Waxman and Giasson, 2010). We hypothesized that a similar mechanism may be involved in the formation of tau containing NFTs that are observed concurrently with Lewy pathology in vivo.
To examine the ability of α-syn aggregation to induce cellular tau aggregation, QBI293 cells were initially co-transfected with expression plasmids for α-syn and full-length tau (2N/R4). Under basal conditions using methanol fixation, α-syn staining was diffuse, and tau appeared bundled with some diffuse immunoreactivity (Figure 1A,B). Co-localization of the two proteins seemed incidental. To examine the effects of recombinant fibril seeds on tau localization, cells were treated with recombinant, pre-fibrillized 21–140 α-syn protein after transfection. Previous studies show that 21–140 α-syn is capable of seeding intracellular amyloid fibrils similarly to full-length α-syn in cellular seeding experiments (Waxman and Giasson, 2010), and 21–140 α-syn was used so that an anti-α-syn antibody directed to the N-terminus of the protein (SNL4) could be used to specifically identify changes to intracellularly expressed full-length α-syn. Using double-immunofluorescence between anti-α-syn and anti-tau antibodies, large intracellular aggregates, composed of both α-syn and tau proteins and in most cases encompassing the entire cell, were formed (Figure 1C-E). Co-localization between α-syn and tau was observed in these cells, and in some cases these aggregates appeared round with Lewy body-type morphology containing a tau immuno-positive “tail” (i.e. Figure 1E).
Since pathological aggregates can be more specifically recognized with antibodies specific to phospho-epitopes on α-syn or tau, we performed double-immunofluorescence between pSer129 [recognizing phospho-S129 on α-syn (Waxman and Giasson, 2008)], and PHF1 or AT8 [recognizing phospho-S396/S404 (Otvos, Jr. et al., 1994) or phospho-S202/T205 (Goedert et al., 1995), respectively, on tau]. In the absence of recombinant α-syn fibril treatment, a paucity of phospho-specific immunoreactivity was observed (Figure 2A for pSer129 and PHF1, data not shown for AT8). However, in the presence of recombinant, pre-fibrillized α-syn, robust immuno-positive pSer129 α-syn aggregates were observed on the edge with some overlay of PHF1 (Figure 2B,C) or AT8 (Figure 2D) immunoreactivity. In addition, some cellular morphology bared remarkable similarity to α-syn and tau containing aggregates observed in the brain of patients with the LBVAD diagnosis (see comparison in Figure 2D&E).
Although α-syn and tau pathology can localize and intermingle within the same aggregate in vivo, most Lewy pathology and NFTs co-exist in the same brain region, but in different neurons (Giasson et al., 2003;Lee et al., 2004). We therefore hypothesized that preformed α-syn fibrils may seed tau aggregation even in the absence of α-syn overexpression. QBI293 cells transfected with only the expression plasmid for full-length tau, displayed rare PHF1 (data not shown) or AT8 (Figure 3A) immunoreactivity and typically only in cells that were rounded in morphology. Treatment with recombinant, pre-fibrillized full-length α-syn significantly increased the number of tau expressing cells that were PHF1 and AT8 immuno-positive (Figure 3B,E-F). In fact, by confocal microscopy, these PHF1 and AT8 immunoreactive cells often appeared NFT-like (Figure 3C). We also found that the propensity of these cells to form these AT8-or PHF1-positive, NFT-like aggregates was further promoted when QBI293 cells were transfected with an expression plasmid containing full-length tau with the FTDP-17 mutation P301L (Figure 3D-F).
Biochemical cellular fractionation further supported the aggregation of intracellular tau, promoted by the addition of recombinant, pre-formed α-syn fibrils (Figure 4). Both in the absence (pcDNA) and presence (α-syn) of α-syn co-expression, recombinant 21–140 α-syn fibrils induced the formation of Triton-insoluble tau, with an increased propensity of P301L tau over that of WT tau. In cells expressing both α-syn and tau and treated with recombinant, pre-fibrillized 21–140 α-syn protein, both tau and full-length α-syn (intracellularly expressed α-syn protein) developed Triton-insoluble aggregates (Figure 4A). However, intracellular α-syn co-expression was not required for significant development of Triton-insoluble tau, indicating that small amounts of pre-formed, exogenously applied fibrils were sufficient to seed aggregation of tau. Phospho-specific antibodies were efficient at detecting α-syn-induced Triton-insoluble tau, with PHF1 and AT8 providing the most robust phospho-specific signals. The percentage of Triton-insoluble tau also increased over the course of time (Figure 4B). Interestingly, when both tau and α-syn were concurrently expressed and treated with pre-formed, recombinant fibrils to form intracellular aggregates, the P301L mutation of tau displayed a significantly greater propensity to aggregate than α-syn in the same cells (at 48 h, p = 0.003 by two-tailed, parametric t-test, n = 5), indicating that cellular seeding of P301L tau could be promoted more readily than α-syn. Further, α-syn aggregation propensity was not affected by tau co-expression (data not shown).
To further characterize the ability of α-syn to promote cellular tau aggregation, populations of polymers of recombinant α-syn were separated and compared in this cellular assay. After fibrillization of recombinant α-syn protein, large α-syn fibrils were isolated from small α-syn oligomers by centrifugation, and each population was subjected to sonication. Electron microscopy revealed various morphologies of α-syn polymers after sonication. Total recombinant α-syn fibrils (fibril mix) displayed a combination of larger fibrils and smaller oligomers (Figure 5A). Isolated small oligomers appeared similar to the fibril mix; however, fewer larger fibrils were observed (Figure 5B). Isolated large fibrils were composed of larger aggregated structures after sonication (Figure 5C). However, they often had an amorphous appearance and smaller circular structures were observed (arrowheads). The polymerized 21–140 (N-terminal truncated) α-syn protein appeared more resistant to sonication, as the majority of polymers observed were longer fibrils (Figure 5D). Although there were differences between experiments depending on α-syn recombinant preparation and experimental variability, each polymer type was capable of inducing intracellular tau aggregation with similar efficiency, while recombinant, soluble (monomeric, non-fibrillized) α-syn did not typically augment tau aggregation propensity (Figure 5E).
Confocal microscopy analyses were then used to evaluate the formation of amyloidogenic aggregates in QBI293 cells expressing WT or P301L tau and treated with α-syn fibrils. AT8- and anti-tau-positive (17025) aggregates co-localized with thioflavin S reactivity (Figure 6A,B), while recombinant, pre-fibrillized α-syn protein, as visualized with antibody SNL4, rarely co-localized with a robust thioflavin S signal (Figure 6C). Further, SNL4 immunoreactivity that was thioflavin-positive was only identified extracellularly, rather than co-localizing with the large thioflavin-positive intracellular aggregates (arrow). Surprisingly, thioflavin S reactivity was often observed throughout the entire cell, suggesting widespread, rather than localized, intracellular tau amyloidogenic inclusions. While the core of the tau aggregates typically exhibited reduced antibody immunoreactivity, likely due to reduced antibody penetrance, thioflavin S, a chemical dye, was capable of completely labeling the entire cellular aggregate. Thioflavin S reactivity was not observed with cells that were transfected with WT or P301L tau, but not treated with recombinant α-syn fibrils (data not shown).
Ultrastructural analyses demonstrated that the intracellular tau inclusions were comprised of 16.6 ± 3.8 [SD] nm wide, predominantly straight fibrils, similar in size and morphology to tau fibrils observed in human brain inclusions (Iqbal et al., 1975;Terry et al., 1964;Goedert, 1999) (Figure 7). In addition, these studies provided insights into the effects of this amyloidogenic, aggregated tau. Cellular morphology was significantly disrupted by the formation of these intracellular tau aggregates with the cellular soma packed throughout with fibrous protein (Figure 7A-E). This resulted in displacement or retraction of the nucleus (Figure 7A,B). In addition, organelles were displaced and/or isolated amidst the fibrous network with mitochondria appearing “trapped” and “sick” (arrowheads). Confocal microscopy analyses between tau and cellular markers lamp1 (for lysosomes) or mannosidase II (for golgi) indicated that these organelles were also frequently trapped within these large cytoplasmic aggregates (data not shown).
To identify the composition of these large cellular aggregates, we immunolabeled P301L tau transfected cells with anti-tau antibodies prior to ultrastructural analyses. In the absence of recombinant fibril treatment, tau immunolabeling appeared diffuse and cytosolic (Figure 7G), while in the presence of recombinant fibril treatment, cellular fibrillar aggregates were ubiquitously labeled (Figure 7H). Similar findings were observed in cells transfected with WT tau (data not shown). Immunolabeling with anti-α-syn antibodies identified only small clusters of intracellular fibrils among the large cellular aggregates (Figure 7I, arrows). In cells that were transfected with α-syn instead of tau, recombinant α-syn fibrils induced the formation of fibrillar α-syn inclusion (Waxman and Giasson, 2010), with robust, ubiquitous immunolabeling of intracellular fibrils with anti-α-syn antibodies (data not shown).
Under non-pathological conditions tau binds to and participates in microtubule assembly (Gustke et al., 1994; Hong et al., 1998; Lee et al., 2001; Goedert and Spillantini, 2006). Confocal microscopy analyses of tau versus cytoskeletal markers suggested the presence of β-tubulin and vimentin inside the fibrous tau aggregate (Figure 8A,C). Large, bulbous tau aggregates were observed, using 3D reconstruction of confocal microscopy images, and these aggregates often displaced γ-tubulin outside the center of cellular plane (Figure 8B).
We then performed biochemical analyses to identify the percent of tau protein bound to microtubules under basal conditions and to identify the propensity of β-tubulin to aggregate under conditions forming cellular tau aggregates. Approximately 40% of cellular tau bound to microtubules, as determined through biochemical microtubule binding assays (Figure 9). Surprisingly, no differences in microtubule-bound proteins were identified between WT and P301L tau proteins. Further, under biochemical cellular fractionation conditions to identify aggregated protein, no Triton-insoluble β-tubulin was identified.
Based on microscopy analyses and biochemical data, recombinant, pre-fibrillized α-syn caused a profound increase in P301L tau aggregation compared to WT tau. Since P301L tau did not differentially bind to microtubules in our system, we examined the direct effects of α-syn seeding on tau aggregation, in vitro. Recombinant, pre-fibrillized, sonicated α-syn was incubated with recombinant-generated WT or P301L tau under assembly conditions in vitro. After three days, α-syn significantly increased the sedimentation and K114 fluorometry of P301L tau over that of WT tau (Figure 10A-C), while in the absence of α-syn neither WT nor P301L tau aggregates after 7 days (Figure 10A – right panel). We then aimed to identify any potential changes in rate of aggregation between WT tau, P301L tau, and α-syn. In these experiments, soluble, recombinant α-syn was used at a concentration of 0.3 mg/ml such that α-syn and tau were approximately equimolar. At this concentration α-syn only minimally begins to aggregate after 7 days of agitation (Figure 10A – left panel). Each of these soluble recombinant proteins were incubated with recombinant, pre-fibrillized, sonicated α-syn, and the amount of sedimented protein was assessed after 1, 3, or 7 days. WT tau and α-syn reached maximum sedimentation after 1 day of incubation with pre-fibrillized α-syn (Figure 10D). However, P301L tau sedimentation continued to increase over the course of 7 days. Under these in vitro conditions, P301L tau aggregation propensity resulting from α-syn fibril seeds was reduced when compared to α-syn.
Tau contains many phosphorylation sites, many of which have been implicated in the formation of NFTs (Grundke-Iqbal et al., 1986; Goedert et al., 1992; Hanger et al., 2007; Alonso et al., 2001; Billingsley and Kincaid, 1997; Buee et al., 2000). Based on analyses of the hyperphosphorylation of tau cellular aggregates (representative immunoblots provided in Figure 4), AT8 immunoreactivity was over-represented in the Triton-insoluble fraction of tau (when compared to 17025; for P301L + fibrils, p = 0.003 by one-sample t-test) and 12E8 was under-represented (when compared to 17025; for P301L + fibrils, p = 0.05 by one-sample t-test; n = 5). Kinases such as GSK3β and MARK2 may therefore alter propensities of tau aggregate formation (Sato et al., 2002; Rankin et al., 2007; Liu et al., 2007; Schneider et al., 1999).
To examine the role of kinases and phosphorylation on tau aggregate formation, QBI293 cells were co-transfected with expression plasmids for WT or P301L tau and GSK3β or pcDNA. Overexpression of GSK3β robustly increased phospho-tau immunoreactivity with antibodies PHF1, AT8, AT180, and AT270, but not 12E8 (data not shown; consistent with (Godemann et al., 1999)). These changes in immunoreactivities were immeasurable due to the profound increases resulting from GSK3β overexpression. We then examined the effects of GSK3β on tau by biochemical cellular fractionation. In the absence of α-syn fibril treatment, GSK3β overexpression increased the amount of Triton-insoluble tau; however, it also attenuated the propensity of tau to aggregate in the presence of recombinant α-syn fibrils (Figure 11A-B).
To examine the effects of phosphorylation on the 12E8 epitopes S262/S356, QBI293 cells were co-transfected with expression plasmids for WT or P301L tau and MARK2 or pcDNA. In the presence of fibril treatment, MARK2 significantly inhibited the formation of Triton-insoluble WT and P301L tau (Figure 11C). Concurrent with this effect, overexpression of MARK2 resulted in approximately a 10-fold increase in 12E8 immunoreactivity (Figure 12A).
To discern the effects of specific phosphorylation sites affected by MARK2 expression on the formation of tau aggregates, we co-expressed MARK2 in cells expressing tau with mutations S262A and/or S356A. In the absence of MARK2 overexpression, S262A and/or S356A mutations did not alter the propensities of tau aggregation in the presence of α-syn fibril treatment (Figure 12A-B). Fibril treatment with S262A tau and MARK2 co-expression exhibited similar inhibition as that observed with WT tau and MARK2. However, surprisingly, S356A tau or S262A/S356A tau and MARK2 co-expression resulted in a significantly greater loss of tau aggregation when compared to WT tau and MARK2 co-expression. These results were only observed for tau without the P301L mutation. The P301L mutation in combination with the S262A and/or S356A mutations resulted in non-significant differences between fibril-treated samples when multiple-comparisons were taken into account.
Interestingly, in the absence of MARK2, the S262A mutation reduced 12E8 immunoreactivity of Triton-soluble tau to 29 ± 18% [SD], when compared to WT tau, and P301L/S262A tau reduced 12E8 immunoreactivity of Triton-soluble tau to 45 ± 30% [SD], but only reduced 12E8 immunoreactivity of Triton-insoluble tau to 69 ± 11% [SD], when compared to the same conditions of P301L tau. Consistent with this finding, the S356A mutations reduced 12E8 immunoreactivity of Triton-soluble tau to 69 ± 27% [SD] and 56 ± 16% [SD] (when compared to WT and P301L tau, respectively), while 12E8 immunoreactivity of Triton-insoluble tau was reduced to 37 ± 31% [SD] of the same condition with P301L tau (n = 3 for all conditions). In the absence of MARK2 co-expression, Triton-insoluble WT tau was not observed with 12E8; therefore, no changes were observed in this fraction with S262A or S356A mutations. Presence of both S262A and S356A mutations completely abrogated 12E8 immunoreactivity in both fractions.
MARK2 overexpression also significantly decreased AT8 immunoreactivity of Triton-soluble tau (by 84 ± 11% [SD] for WT tau when standardized to 17025 immunoreactivity; p = 0.008 by one-sample t-test; n = 4). Furthermore, the S356A and the S262A/S356A mutations in tau reduced AT8 immunoreactivity to 51 ± 21% [SD] and 21 ± 14% [SD], respectively (compared to WT tau when standardized to 17025 immunoreactivity; p = 0.06 and p = 0.01 by one-sample t-test; n=3). These effects appeared isolated, since relative AT8 immunoreactivity of Triton-insoluble tau was not altered when controlled for 17025 immunoreactivity, and PHF1 immunoreactivity was not affected by MARK2 or S356A mutations in the either fraction.
To discern potential effects caused by alterations of the residues associated with the AT8 epitope, we mutated residues S202 and T205 to either Ala or Glu. The Ala residue was used to remove any basal phosphorylation of these residues, while the Glu was used as a phospho-mimetic. Cells transfected with S202A/S205A or S202E/S205E tau and treated with recombinant fibrils each demonstrated significant reductions in the propensity of tau to aggregate, but to varying degrees (Figure 12C). With cells transfected to express P301L tau, only the additional mutations of S202E/S205E modestly reduced the propensity of tau to aggregate in the presence of recombinant fibrils.
The current work provides in-depth analyses of the formation of NFT-like inclusions that develop in response to cellular seeding by recombinant, pre-formed α-syn fibrils. This study establishes an important connection between the formation of α-syn-containing Lewy pathology and the formation of NFTs, as observed in pathological brain. Hyperphosphorylated, Triton-insoluble, thioflavin S-positive, filamentous tau could be induced to intermingle with endogenously expressed, aggregated α-syn or prompted to aggregate from the addition of only a small amount of α-syn fibril seeds. Furthermore, a significant number of tau-expressing cells (~11%) and a significant proportion of total WT tau protein (up to ~30%) formed aggregates in the presence of fibrillized α-syn seeds. These studies show that the entry of only a small amount of fibrillized α-syn is sufficient to induce the formation of tau fibrillar inclusions that encompass almost the entire cell, even in the absence of α-syn expression. This finding is consistent with the previously suggestions that only a minute amount of amyloidogenic α-syn may be necessary to induce tau aggregation with pathological consequences (Lee et al., 2004). Furthermore, this cellular model is uniquely capable of examining factors associated with tau inclusion formation, mimicking pathology observed post-mortem.
Recent cellular studies have shown intracellular tau aggregation in response to treatment with recombinant, pre-formed tau fibrils, producing small intracellular, hyperphosphorylated tau aggregates (Frost et al., 2009;Nonaka et al., 2010). However, unlike α-syn that can polymerize by itself, tau aggregation requires an inducing agent (Goedert et al., 1996). This work suggests that α-syn polymers may initially induce tau aggregation, but once this process is initiated, aggregated forms of tau may propagate the aggregation process, further promoting amyloid formation.
In the cellular model described here, the intracellular tau inclusions developed to encompass the entire cellular soma, similar to NFT in diseased human brains. This is unlike α-syn cellular aggregates (Waxman and Giasson, 2010), which typically expand around the γ-tubulin-positive centrosome and enlarge in a sphere-like manner, but remain relatively localized. These tau aggregates, instead, displaced the centrosome and isolated organelles, which likely ceased cellular activity. Based on cytoskeletal localization, these tau aggregates do not form aggresomes, but instead trap and mingle with cytoskeletal markers. While these microscopic data are difficult to interpret, the overwhelming toxic appearance, including dying mitochondria, was apparent, indicating that these tau inclusions result in cellular injury. Further, ultrastructural analyses of these pathological aggregates bare striking structural similarity to tau aggregates characterized in animal models of tauopathies (Lin et al., 2003) and to NFTs analyzed postmortem (Iqbal et al., 1975;Terry et al., 1964).
Interestingly, the P301L mutation in tau that is causal of FTDP-17, significantly promotes the cellular aggregation of tau. The P301L mutation enhances tau polymerization in vitro in studies performed with heparin (Nacharaju et al., 1999;Barghorn et al., 2000;Goedert et al., 1999). However, in vitro studies performed with pre-fibrillized α-syn show that WT tau aggregation reached maximum after 1 day, while P301L tau continued to aggregate through 7 days of incubation. This difference could be due to differential protein interactions or differences in structural folding between WT and P301L tau. Furthermore in cells expressing both P301L tau and α-syn, P301L tau augmented the propensity of tau aggregation in the presence of recombinant seeds more than that observed with α-syn in the same samples. This difference was not recapitulated in vitro, as α-syn seeds promoted α-syn aggregation more readily than P301L tau. This disconnect between in vitro and in situ may be due to the ability of recombinant α-syn to self-aggregate, while intracellular expressed α-syn requires seeding, or other cellular factors may promote P301L tau aggregation in vivo.
In vitro studies suggest that the P301L mutation reduces binding of tau to microtubules, thereby increasing the pool of tau available to aggregate (Hong et al., 1998; Sun and Gamblin, 2009). However, in this study tau binding to microtubules was not altered by this mutation, similar to other reports (Vogelsberg-Ragaglia et al., 2000), suggesting a lack of microtubule involvement in the increased propensity of P301L tau to develop aggregates. While the current study suggests that cytoskeletal markers may be present and potentially isolated within tau inclusions, biochemical cellular fractionation did not isolate β-tubulin in the Triton-insoluble fraction, indicating that microtubules are independent from tau inclusions.
The relationships between kinase activity, phosphorylation of tau, and NFT formation have been a long-standing debate. Tau is phosphorylated at an abundant number of sites, and most of these sites are hyperphosphorylated in the pathological form of tau. GSK3β and its phosphorylation of the PHF1 and AT8 epitopes have been widely implicated in the formation of pathological tau aggregates (Sato et al., 2002; Rankin et al., 2007; Liu et al., 2007; Engel et al., 2006; Jeganathan et al., 2008; Kwok et al., 2008). However, co-expression of GSK3β in this study resulted in two opposing processes: one, in which basal expression of tau increased in the Triton-insoluble fraction, and the second, which significantly decreased α-syn-induced tau aggregate formation. GSK3β phosphorylates tau at many sites, including but not limited to T181, S202/T205, T231/S235, and S396/S404 (Godemann et al., 1999; Hanger et al., 2007; Dayanandan et al., 1999; Ishiguro et al., 1993). Therefore, while phosphorylation at one site may increase the propensity of tau to aggregate, another site may induce inhibitory processes. However, structural forms of Triton-insoluble tau promoted by GSK3β expression may be independent from the tau fibrils created by recombinant α-syn seeds, which would suggest only inhibitory processes involved with the expression of GSK3β on tau inclusion formation. Nevertheless, these data are consistent with a previous report of GSK3β-mediated inhibition of tau fibrillization observed in vitro (Schneider et al., 1999).
MARK2 expression was also capable of significantly reducing tau aggregate formation. This is not surprising, since MARK2 primarily phosphorylates S262 and S356 within the tau repeat region (Drewes et al., 1997; Drewes et al., 1995), the region that forms the core of tau filaments and is required for tau aggregation (Crowther et al., 1989; Crowther et al., 1992; Goode and Feinstein, 1994), and MARK2 phosphorylation of tau can inhibit fibrillization in vitro (Schneider et al., 1999). However, upon investigation of the potential residues involved in MARK2-mediated inhibition, single-point mutations did not identify either S262 or S356 as responsible. An uncharacterized phosphorylation site on tau or other biological effects may therefore be responsible. Interestingly, single-point mutation analyses identified S356 phosphorylation as potentially promoting tau aggregation, consistent with Triton-insoluble tau being enriched in phospho-S356 over phospho-S262.
An additional unexpected effect of MARK2 was that co-expression of MARK2 reduced AT8 immunoreactivity. This effect was not generalized to PHF1 immunoreactivity, suggesting a change in kinase specificity in regards to the S202/T205 sites. Studies suggest that AT8 phosphorylation promotes the formation of tau aggregates (Rankin et al., 2007;Jeganathan et al., 2008;Biernat et al., 1992). We therefore examined the role of the AT8 epitope by mutating these residues to block phosphorylation or add phospho-mimetics. Mutations to either A or E both inhibited the propensity of tau to aggregate, with a stronger effect noted for E. These data suggest that any alterations to the S202/T205 residues may reduce the ability of tau to aggregate, rather than only abrogating or enhancing phosphorylation-like activity. Therefore, the role phosphorylation at the AT8 epitopes may not be adequately assessed by these means.
However, changes in AT8 immunoreactivity provided relevant information as to substrate targeting of tau by kinases. MARK2 co-expression or the P301L mutation in tau significantly decreased AT8 immunoreactivity of Triton-soluble tau (Boekhoorn et al., 2006; Han and Paudel, 2009) (also see Figure 4; AT8 immunoreactivity P301L, 66 ± 24% [SD] less than WT tau, p = 0.003; n=5), but AT8 immunoreactivity in the Triton-insoluble fraction was not affected. These data suggest that tau is likely phosphorylated at the AT8 epitope after protein aggregation and that kinases may independently target soluble or aggregated tau. These studies suggest that investigations into factors that modulate the phosphorylation of the soluble forms of tau may not be generalized to the aggregated, NFT-like form of tau.
Investigations into the formation of pathological tau aggregates have been limited, to date, by the availability of adequate cellular models. While in vitro studies are informative, the cellular milieu needs to be considered for potential therapeutic intervention. The current study has extended cellular modeling to include the ability of α-syn to induce the robust formation of filamentous tau aggregates with characteristics of NFTs. The ability of a small amount of preformed α-syn to induce large intracellular amyloid tau inclusions is consistent with the frequent co-existence of α-syn and tau inclusions in the diseased brain, and our study supports the notion that minute amounts of α-syn can be a physiological inducer of tau aggregation (Giasson et al., 2003). This ability of α-syn is also consistent with the mounting evidence that α-syn can be transported across cell membranes (Liu et al., 2009; Ahn et al., 2006; Lee et al., 2008; Desplats et al., 2009) and that amyloid diseases may spread by prion-like mechanisms (Lee et al., 2010; Goedert et al., 2010). Therefore, this research provides a likely mechanism to explain the frequent co-existence of hyperphosphorylated α-syn and tau pathology observed postmortem in disease. It is through the understanding of these factors that are associated with the formation of pathological aggregates and result in cross-seeding of pathological proteins that we may develop future therapeutics for disease.
This work was funded by grants from the National Institute on Aging (AG09215) and the National Institute of Neurological Disorders and Stroke (NS053488). We thank the Biochemical Imaging Core Facility supported by Abramson Cancer Institute at the University of Pennsylvania for assistance in the EM and confocal microscopy studies.
Conflict of Interest: None