Report

Initiation and Synergistic Fibrillization of Tau and Alpha-Synuclein

See allHide authors and affiliations

Science  25 Apr 2003:
Vol. 300, Issue 5619, pp. 636-640
DOI: 10.1126/science.1082324

Abstract

Alpha-synuclein (α-syn) and tau polymerize into amyloid fibrils and form intraneuronal filamentous inclusions characteristic of neurodegenerative diseases. We demonstrate that α-syn induces fibrillization of tau and that coincubation of tau and α-syn synergistically promotes fibrillization of both proteins. The in vivo relevance of these findings is grounded in the co-occurrence of α-syn and tau filamentous amyloid inclusions in humans, in single transgenic mice that express A53T human α-syn in neurons, and in oligodendrocytes of bigenic mice that express wild-type human α-syn plus P301L mutant tau. This suggests that interactions between α-syn and tau can promote their fibrillization and drive the formation of pathological inclusions in human neurodegenerative diseases.

Tau and α-synuclein (α-syn) are abundant brain proteins with distinct biological functions and intraneuronal distributions. Tau is a microtubule binding protein that stabilizes and promotes microtubule polymerization in neuronal perikarya and processes (1, 2), whereas α-syn is localized mainly in axon terminals where it may regulate synaptic functions (3). However, tau and α-syn polymerize into fibrils that accumulate to form pathological inclusions in distinct neurodegenerative disorders now classified as tauopathies and synucleinopathies, respectively (1, 2, 4, 5). For example, neurofibrillary tangles composed of tau are hallmarks of Alzheimer's disease, and Lewy bodies, the signature pathological inclusions of Parkinson's disease, are composed of α-syn.

The discoveries of mutations in α-syn and tau that result in synucleinopathies and tauopathies, respectively (6, 7), and that promote the fibrillization of the mutant proteins in vitro and in transgenic mice, causing neuronal dysfunction (813), support the notion that filamentous inclusions formed by α-syn and tau are toxic. However, whereas α-syn readily self-polymerizes in vitro (8, 9), tau requires cofactors (such as glycosaminoglycans or nucleic acids) (14), and it remains enigmatic how these proteins fibrillize in vivo.

Recent studies of an individual with Parkinson's disease (IX-47) of the Contursi kindred with the rare A53T α-syn mutation revealed widespread α-syn and tau inclusions (15). Post-mortem examination of another affected member (IX-51) of this kindred also demonstrated abundant α-syn and tau inclusions (figs. S1 and S2). Thus, a pathogenic mutation in α-syn that is known to increase the propensity of α-syn to fibrillize (8, 9) also promotes formation of tau inclusions in humans. Co-occurrence of tau and α-syn inclusions is also observed in the brains of individuals with sporadic neurodegenerative disorders (16, 17) (fig. S1).

These findings suggest that the formation of pathological inclusions composed of α-syn or tau share common mechanisms. We monitored the effect of incubating the proteins together on their in vitro fibrillization by K114 fluorometry (Fig. 1A), a sensitive and quantitative measure of amyloid fibril formation (18). Centrifugation at 100,000g followed by quantitative Western blotting (Figs. 1, B and C, and 2A) was used to monitor the polymerization of each protein individually. For comparison, tau fibrillization in the absence of α-syn was induced with heparin (Figs. 1 and 2A; fig. S3) (14). We monitored the effects of coincubating tau and α-syn after incubation of tau at 1 mg per mL of 100 mM Na acetate, pH 7.4, while varying the concentration of α-syn (0.25 to 5 mg/mL). Incubation of tau T40 (2N/4R) (19), the longest tau isoform expressed in the central nervous system (CNS) (20), alone did not result in detectable K114 fluorescence (Fig. 1A) or in a time-dependent increase in pelletable tau polymers (Fig. 1C). The incubation of α-syn alone resulted in the formation of amyloid-like, pelletable polymers, but polymerization was greater at higher concentrations of α-syn (at 2.5 and 5 mg/mL) (Figs. 1, A and B, and 2A). The coincubation of tau T40 and α-syn resulted in greater K114 fluorescence compared to samples that contained only α-syn (Fig. 1A). Sedimentation analysis demonstrated that coincubation of tau and α-syn induced polymerization of both proteins (Figs. 1, B and C, and 2A). The presence of α-syn, even at concentrations as low as 0.25 mg/mL, resulted in the accumulation of pelletable polymers of tau T40. Furthermore the polymerization of α-syn was promoted by the presence of tau T40, especially at lower concentrations of α-syn (at 0.25 mg/mL and 1 mg/mL).

Fig.1.

Coincubation of tau T40 and α-syn promotes the polymerization of both proteins. (A) The formation of amyloidogenicfibrils was monitored by K114 fluorometry. AFU, arbitrary fluorescence units. (B and C) Polymerization of (B) α-syn and (C) tau T40 was assayed by quantitative sedimentation analysis. Assembly was monitored after 24-, 48-, and 96-hour incubations with constant agitation at 37°C in 100 mM Na acetate, pH 7.4 (9). Tau T40 was incubated at a concentration of 1 mg/mL for all experiments. Heparin (hep, 50 μM) was used to induce the assembly of tau T40 (14). α-syn was incubated at concentrations of 0.25, 1, 2.5, and 5 mg/mL. β-syn and Δ71-82 α-syn were incubated at 5 mg/mL. All syn proteins were incubated alone or in the presence of tau T40 as indicated. For quantitative sedimentation analysis, samples were fractionated into supernatants and pellets by centrifugation at 100,000g, followed by Western blot analysis with 125I-conjugated secondary antibodies (10, 21). The membranes were dried and exposed to a PhosphorImager plate. The radioactive signal was quantified with ImageQuant software (33) (n = 4 samples).

Fig.2.

Centrifugal sedimentation and Western blot analysis of tau and α-syn polymerization. All six major CNS isoforms of tau (T40, T39, T43, T44, T37, and T34) at a concentration of 1 mg/mL were incubated under assembly conditions for 96 hours and processed as described in Fig. 1. The rabbit 17026 antibody to tau was used to detect tau, and the mouse Syn 102 antibody to α-syn, which reacts equally with α-and β-syn, was used to detect these proteins. Representative experiments are shown. (A) Tau T40 was incubated alone, in the presence of 50 μM heparin, in the presence of 1 or 5 mg/mL α-syn, or in the presence of 5 mg/mL β-syn. α-syn was incubated alone at a concentration of 1 or 5 mg/mL. (B to F) The T39, T43, T44, T37, and T34 isoforms of tau were incubated alone or in the presence of 50 μM heparin, 1 mg/mL α-syn, or 1 mg/mL β-syn. S, supernatants; P, pellets.

To investigate the type of molecular interaction(s) involved in promoting tau and α-syn polymerization and the specificity of this process, we performed parallel experiments with β-synuclein (β-syn), a homolog of α-syn that does not fibrillize, and Δ71-82 α-syn, an artificial α-syn mutant that can not self-fibrillize (21). Coincubation of β-syn or Δ71-82 α-syn with tau T40 did not result in detectable K114 fluorescence (Fig. 1A) or the accumulation of pelletable syn or tau T40 polymers (Figs. 1, B and C, and 2A). Moreover, incubation of the amyloidogenic peptide Aβ(1–40) with tau T40 did not induce tau fibrillization and did not have a synergistic effect on the formation of amyloid fibrils (fig. S4), demonstrating specificity of the effects of α-syn on tau.

Because six alternatively spliced tau isoforms are expressed in the CNS [T44 (0N/3R), T43 (0N/4R), T37 (1N/3R), T34 (1N/4R), T39 (2N/3R), and T40 (2N/4R)] (19, 20), we examined the effects of mixing α-syn with each one. K114 fluorometry and formation of pelletable polymers were monitored after a 96-hour incubation of α-syn (1 mg/mL), conditions at which α-syn does not readily self-polymerize. Incubation of each tau isoform with α-syn resulted in K114 fluorescence (fig. S3) and in the accumulation of pelletable polymers of tau and α-syn (Fig. 2 and fig. S3). In contrast, coincubation of β-syn with each tau isoform failed to induce polymerization of tau or β-syn (Fig. 2 and fig. S3).

We conducted negative-staining electron microscopic analysis (EM) to confirm the formation of amyloid fibrils. This revealed the presence of abundant 10- to 15-nm–wide filaments from samples incubated with both tau and α-syn, but no filaments in the tau plus β-syn samples (data not shown). To assess the composition of the fibrils formed from mixing α-syn and tau, we performed immuno-EM with antibodies to α-syn and tau and secondary antibodies conjugated to either 5-nm or 10-nm colloidal gold particles (Fig. 3). Whereas many fibrils were labeled solely with antibodies to α-syn (Fig. 3A), bundled fibrils labeled with both α-syn and tau antibodies were also demonstrated (Fig. 3, B and C). Analysis of individual fibrils revealed that most were homopolymers, because most filaments were labeled solely with either antibodies to tau (Fig. 3D) or antibodies to α-syn (Fig. 3E), whereas rare fibrils were predominantly labeled with antibodies to α-syn but also displayed scant tau immunoreactivity (Fig. 3F). However, other fibrils were labeled focally in spatially separate domains with antibodies to either α-syn or tau (Fig. 3G), suggesting that these fibrils resulted from the end-to-end annealing of filaments formed solely by either tau or α-syn. Some fibrils labeled by antibodies to tau exhibited the characteristic twisted appearance of tau fibrils isolated from human-disease brains (Fig. 3H).

Fig.3.

Immuno-EM of tau and α-syn fibrils. α-syn (1 mg/mL) and tau T40 (1 mg/mL) were incubated under assembly conditions for 96 hours. Samples were applied to carbon-coated grids, labeled with various primary and colloidal gold conjugated secondary antibodies, and negatively stained with uranyl acetate (22). The following combinations of antibodies were used: (A) mouse antibody to α-syn Syn 505 (5-nm gold) and rabbit antibody to tau 17026 (10-nm gold); (B) mouse antibody to α-syn Syn 506 (5-nm gold) and rabbit 17026 (10-nm gold); (C) rabbit antibody to α-syn SNL-4 (5-nm gold) and mouse antibody to tau Tau-1 (10-nm gold); (D to G) mouse Syn 505 (5-nm gold) and rabbit 17026 (10-nm gold); (H) rabbit SNL-4 (10-nm gold) and mouse antibody to tau 24E12 (5-nm gold). In (G), the brackets emphasize the spatial separation of bound antibodies to tau and α-syn within continuous fibrils. In (H), the fibril labeled solely with antibodies to tau has a twisted appearance. Five- and 10-nm colloidal gold particles are identified with arrowheads and arrows, respectively. Scale bar, 100 nm [(A) to (C), (G) and (H)], 80 nm [(D) and (E)].

To evaluate whether α-syn can induce tau fibrillization in vivo, we analyzed brain sections from a transgenic mouse model of α-synucleinopathy that was generated by overexpression of human A53T α-syn (10). These mice develop severe motor impairments that coincide with the formation of abundant α-syn inclusions (Fig. 4A). Re-examination of brains from diseased mice with antibodies to tau revealed that a subset (∼25%) also accumulated abundant tau-positive threads, grains (Fig. 4B), and spheroids (Fig. 4B, inset) in affected regions of the pons, midbrain, and spinal cord. Rare perikaryal tau inclusions with the morphology of a pre-tangle also were detected (Fig. 4B, arrowhead). Less frequent tau inclusions also were present in another ∼25% of symptomatic A53T α-syn mice, but none were detected in ∼50% of the diseased mice. Tau inclusions were not detected in control nontransgenic mice or in transgenic mice that expressed equivalent levels of wild-type human α-syn but do not develop α-syn inclusions or disease (10). Double-label immunofluorescence microscopy demonstrated that some pathological inclusions in the A53T α-syn transgenic mice were composed of tau alone, α-syn alone, or both proteins (Fig. 4, C to E).

Fig.4.

Tau and α-syn pathology in α-syn A53T transgenicand α-syn/P301L tau bigenic mice. (A and B) Immunohistochemistry of 6-μm sections from the pons of A53T human α-syn transgenicmice stained with (A) mouse Syn506 and (B) rabbit 17026. PrP, prion protein promoter; Tg, transgenic. The micrograph shows perikaryal and neuritic α-syn and tau inclusions in the pons. Occasional tau-positive spheroids (B, inset) and a rare somatic accumulation of tau consistent with pre-tangle pathology (B, arrowhead) were detected. (C to E) Double-label immunofluorescence of brainstem pathology stained with (C) rabbit 17026 (red) and (D) Syn 506 (green). (E) is an overlay of (C) and (D), where lesions containing both α-syn and tau are identified by arrowheads. [(F) to (T)] Formation of tau and α-syn amyloidogenicinclusions in oligodendrocytes of CNP tauP301L/α-syn bigenicmice. Staining of (F to H) an 18-month-old CNP-tauP301L (line 6) single transgenicmouse, (I to K) an 18-month-old CNP-α-syn (line M2) single transgenic mouse, and (L to T) a 12-month-old CNP-tauP301L/α-syn bigenicmouse. A subset of Thioflavinne-S–positive oligodendrocytes were detected in CNP-tauP301L/α-syn bigenicmice but not in single CNP-α-syn or CNP-tauP301L transgenic mice. Immunofluorescence staining [left column and (S)], Thioflavinne-S staining [green, in (G), (J), (M), and (P)] and merge images (right column) of tissue sections in the basal ganglia of mice are shown. Tau staining (antibody 17026) is shown in (F), (H), (L), (N), (R), and (T), and α-syn staining (antibody Syn 204) is shown in (I), (K), (O), (Q), (S), and (T). Scale bars, 50 μm [(A) and (B)], 40 μm [(C) to (T)].

To further investigate whether α-syn and tau synergistically promote the fibrillization of each other, we examined bigenic mice that express both wild-type human α-syn and P301L mutant tau (tauP301L) driven by the murine 2′,3′-cyclic nucleotide 3′-phosphodiesterase (CNP) promoter (22). In the single transgenic mouse lines, tauP301L (line 6) and human α-syn (line M2) expression was observed in oligodendrocytes (Fig. 4, F and I), but these mice had not developed tau or α-syn inclusions at 18 months of age (Fig. 4, F to K). By contrast, we detected Thioflavinne S positive staining in a subset of oligodendrocytes of bigenic CNP-tauP301L/α-syn mice that express both α-syn and tau as early as 12 months of age (Fig. 4, L to T). The accumulation of these inclusions coincided with the development of a limb-twitch phenotype. Because oligodendrocytes express very low levels of endogenous α-syn and tau proteins (23, 24), the formation of Thioflavinne S pathological inclusion must be due to the interaction of transgenic α-syn and tau proteins that facilitate and promote each other's fibrillization.

Although α-syn and tau normally have different cytoplasmic distributions and biological functions (1, 2, 4), they also share some common properties. For example, both are abundant neuronal proteins (1, 2, 4) that normally adopt an unfolded conformation (21, 25, 26) but polymerize into amyloid fibrils in disease (1, 4, 5). However, it is unclear why α-syn readily self-polymerizes, whereas tau requires cofactors. One possible explanation is that α-syn is a small protein (140 amino acids) (5) and the middle hydrophobic region that drives filament formation (21) is exposed, facilitating intermolecular interactions that lead to fibrillization. In contrast, the repeat region of tau required for filament formation (27) may be “shielded” within flanking regions of the much larger tau proteins (352 to 441 amino acids), or conformational changes required for fibrillization may be less kinetically favorable in tau.

Our data suggest a two-step mechanism (initiation followed by propagation) whereby α-syn and tau interact to promote each other's fibrillization. The initiation step most likely involves the formation of amyloid-like α-syn polymers, because β-syn and Δ71–82 α-syn can neither fibrillize (21) nor drive tau polymerization. Other possible mechanisms to account for how α-syn might induce tau fibril formation include α-syn acting as a pathological chaperone for tau fibrillization (28) or interactions of α-syn with tau through ionic charges in the COOH-terminal region in α-syn (29). However, we consider these scenarios unlikely, because β-syn also has chaperone activity (28) and both β-syn and Δ71–82 α-syn are similar to α-syn domains in composition and distribution of charged amino acids (4, 5), although neither promoted tau fibril formation. Initiation of tau fibrillization most likely requires α-syn polymers that act as amyloidogenic “seeds” (30, 31) or as amyloidogenic chaperones that induce a conformational change in tau. However, such initiating mechanisms are complex and must require more than just an amyloidogenic “core” structure, because other amyloidogenic proteins [such as Aβ (1–40)] are unable to initiate tau fibrillization. Thus, α-syn polymers appear relatively specific in their initiation of tau fibrillization.

We also observed that tau and α-syn synergistically promote and propagate each other's polymerization into fibrils. In the absence of tau, low concentrations of α-syn did not fibrillize, but in the presence of tau,>60% of α-syn assembled into fibrils. Once fibrillization was initiated by α-syn, tau and α-syn preferentially formed homopolymers, although they also associated in the same filament, presumably by end-to-end annealing of homopolymeric tau and α-syn fibrils. These findings suggest that rather than facilitating the complete assembly of tau filaments, α-syn may instead enhance a critical rate-limiting step in the initiation and early stages of fibrillization.

Previous in vitro studies have demonstrated that cofactors such as glycosaminoglycans or nucleic acids (14) can induce the assembly of tau into fibrils, but there is little evidence that this occurs in vivo. Here, we provide in vivo evidence that α-syn polymerization can induce the assembly of tau fibrils in neurons and in oligodendrocytes. In both humans and mice, the expression of pathogenic A53T mutant α-syn produces α-syn filamentous inclusions linked to the formation of fibrillar tau lesions. However, the induction of tau pathology by α-syn lesions is not restricted to the expression of the A53T mutant α-syn, because co-occurrence of tau and α-syn pathology (in the same or separate lesions) can be detected in neurons and oligodendrocytes of many neurodegenerative disorders (16, 17) (fig. S1).

α-Syn likely is but one of several cofactors that induce the formation of pathological tau lesions, because α-syn inclusions are not found in every brain with tau pathology. Other than glycosaminoglycans that have been observed in neurofibrillary tangles of patients with Alzheimer's, most of these cofactors remain to be identified. However, α-syn may still play a role in brains with tau pathology but no obvious α-syn pathology. In this scenario, we speculate that a limited amount of amyloidogenic α-syn fibrils can serve as seeds to initiate tau fibrillization. This residual amount of α-syn may be undetectable or may be degraded after the initiation of tau polymerization. This is consistent with the notion that tau inclusions are more resistant to degradation than α-syn inclusions are, as suggested by the greater abundance of “ghost” or extracellular fibrillary tangles than of extracellular Lewy bodies.

We conclude that α-syn induces the formation of tau fibrils and that both tau and α-syn synergistically effect the polymerization of each other into fibrillar amyloid lesions. These findings provide insights into mechanisms that underlie the formation of pathological inclusions in neurodegenerative diseases, and they suggest that therapeutic agents that directly or indirectly inhibit the formation of one form of amyloid might be effective on several of these neurodegenerative disorders.

Supporting Online Material

www.sciencemag.org/cgi/content/full/300/5619/636/DC1

Materials and Methods

Figs. S1 to S4

References and Notes

View Abstract

Navigate This Article