Alzheimer-like Changes in Microtubule-associated Protein Tau Induced by Sulfated Glycosaminoglycans

Hyperphosphorylated microtubule-associated protein tau is the major proteinaceous component of the paired helical and straight filaments which constitute a defining neuropathological characteristic of Alzheimer’s disease and a number of other neurodegenerative disorders. We have recently shown that full-length recombinant tau assembles into Alzheimer-like filaments upon incubation with heparin. Heparin also promotes phosphorylation of tau by a number of protein kinases, prevents tau from binding to taxol-stabilized microtubules, and produces rapid disassembly of microtubules assembled from tau and tubulin. Here, we have used the above parameters to study the interactions between tau protein and a number of naturally occurring and synthetic glycosaminoglycans. We show that the magnitude of the glycosaminoglycan effects is proportional to their degree of sulfation. Thus, the strongly sulfated glycosaminoglycans dextran sulfate, pentosan polysulfate, and heparin were the most potent, whereas the non-sulfated dextran and hyaluronic acid were without effect. The moderately sulfated glycosaminoglycans heparan sulfate, chondroitin sulfate, and dermatan sulfate had intermediate effects, whereas keratan sulfate had little or no effect. These in vitro interactions between tau protein and sulfated glycosaminoglycans reproduced the known characteristics of paired helical filament-tau from Alzheimer’s disease brain. Sulfated glycosaminoglycans are present in nerve cells in Alzheimer’s disease brain in the early stages of neurofibrillary degeneration, suggesting that their interactions with tau may constitute a central event in the development of the neuronal pathology of Alzheimer’s disease.

The paired helical filament (PHF) 1 and the related straight filament (SF) are the major components of the neurofibrillary deposits that form a defining neuropathological characteristic of Alzheimer's disease and a number of other neurodegenerative disorders (reviewed in Ref. 1). They are composed of microtubule-associated protein tau, in a hyperphosphorylated state. Mass spectrometry and immunological studies have identified a large number of phosphorylation sites in PHF-tau (2)(3)(4)(5)(6)(7)(8)(9)(10)(11). Some of these sites are not phosphorylated in tau from normal brain, whereas others are phosphorylated to a greater extent in PHF-tau than in tau from normal brain. Many phosphorylated sites are serine/threonine-prolines. Consequently, tau protein can be phosphorylated in vitro at many of these sites by proline-directed protein kinases, such as mitogen-activated protein (MAP) kinase (12)(13)(14), stress-activated protein (SAP) kinases (15,16), glycogen synthase kinase-3 (GSK3) (17)(18)(19), and neuronal cdc2-like kinase (NCLK) (20,21). Moreover, cAMP-dependent protein kinase and Ca 2ϩ /calmodulin-dependent protein kinase II phosphorylate tau at specific sites in vitro, some of which are also phosphorylated in PHF-tau (22)(23)(24). Hyperphosphorylation of tau results in its inability to bind to microtubules and is believed to precede PHF assembly (25)(26)(27). However, it is unclear whether hyperphosphorylation of tau is either necessary or sufficient for PHF formation.
We have recently shown that a phosphorylation-independent interaction between recombinant tau and sulfated glycosaminoglycans leads to the formation of Alzheimer-like filaments under physiological conditions in vitro (28,29). Three repeatcontaining tau isoforms gave rise to paired helical-like filaments, whereas four repeat-containing isoforms formed straight filaments, thus suggesting an explanation for the two tau assemblies present in Alzheimer's disease brain. We also showed that heparin prevents tau from binding to microtubules and promotes microtubule disassembly. Heparan sulfate and hyperphosphorylated tau have been found to co-exist in nerve cells in Alzheimer's disease brain at the earliest known stages of neurofibrillary pathology (28). Moreover, phosphorylation of tau by cdc28, cAMP-dependent protein kinase, GSK3, and several SAP kinases is markedly stimulated by heparin (11, 15, 30 -32). Taken together, these findings suggest that an interaction between tau protein and sulfated glycosaminoglycans may be a central event in the development of the neurofibrillary pathology of Alzheimer's disease.
In the present study we show that the degree of sulfation of glycosaminoglycans is of crucial importance for their ability to induce tau filament formation, to prevent tau from binding to microtubules, and to promote microtubule disassembly. Of those tested, dextran sulfate, pentosan polysulfate, and heparin are the most sulfated, followed by dermatan sulfate, heparan sulfate, chondroitin sulfate, and keratan sulfate. Hyaluronic acid and dextran are not sulfated (see Table I for degrees of sulfation). We also show that the phosphorylation of tau by MAP kinase, NCLK, and GSK3␤ is markedly stimulated in the presence of heparin, at heparin concentrations lower than those required for tau filament formation. Phosphorylation of tau by MAP kinase, NCLK, and GSK3␤ is also stimulated by heparan sulfate, dextran sulfate, and pentosan polysulfate, but not by dextran and hyaluronic acid, with the magnitude of stimulation of tau phosphorylation being proportional to the degree of glycosaminoglycan sulfation. Tau phosphorylation by MAP kinase, NCLK, and GSK3 is also stimulated in the presence of nucleic acids and tubulin. Nucleic acids had little effect on the binding of tau to microtubules. However, incubation of tau with tRNA led to the formation of filaments, in confirmation of a recent report (33).
These results raise the possibility that an interaction between tau protein and negatively charged polymers with a sugar backbone, as found in sulfated glycosaminoglycans and nucleic acids, results in a conformational change in tau that induces polymerization of tau molecules via the microtubulebinding repeats of individual tau molecules, resulting in the formation of filaments like those present in Alzheimer's disease and other neurodegenerative disorders.
Microtubule Binding and Assembly-For microtubule binding, recombinant htau40 (4 M, 0.18 mg/ml) was incubated with different concentrations of glycosaminoglycans and nucleic acids (10, 50, 100, 500, and 1,000 g/ml) in assembly buffer (80 mM PIPES, 1 mM MgCl 2 , 1 mM EGTA, 1 mM dithiothreitol, 1 mM GTP, pH 6.8) for 10 min at 37°C, then added to 10 M taxol-stabilized microtubules and incubated for a further 20 min. Following ultracentrifugation, aliquots of supernatants (free tau) and pellets (microtubule-bound tau) were subjected to SDSpolyacrylamide gel electrophoresis. Protein concentrations were estimated by scanning the gels with a Molecular Dynamics computing densitometer (Model 300 A), and were expressed as percentage of tau bound to microtubules in the absence of glycosaminoglycans and nucleic acids (taken as 100%). For microtubule assembly, recombinant htau40 (2 M) was incubated with tubulin (10 M) in assembly buffer at 37°C. After 5 min, glycosaminoglycans and nucleic acids (100 g/ml) were added and incubated for a further 5 min. Polymerization and depolymerization of microtubules were monitored by measuring the turbidity at 350 nm.

Effects of Glycosaminoglycans on Phosphorylation of Tau by
NCLK-Recombinant htau40 was incubated for various times (ranging from 10 min to 24 h) with 5 units/ml recombinant reconstituted NCLK, in the presence or absence of 50 g/ml heparin. After 24 h tau incorporated 3.8 mol of phosphate/mol of protein in the absence of heparin and 11.4 mol of phosphate/ mol of protein in the presence of heparin (Fig. 1). The stimulation of tau phosphorylation by heparin was apparent throughout the incubation period and was particularly evident at early time points (Fig. 1). A dose-response curve showed that the effect was maximal at 30 g/ml heparin. Phosphorylation of tau by NCLK in the presence of heparin produced the epitopes of phosphorylation-dependent anti-tau antibodies which recognize (S/T)P sites in tau, as shown in Fig. 2 for antibody AT8 which recognizes tau phosphorylated at Ser-202 and Thr-205 (8). Moreover, tau was immunoreactive for antibody 12E8 which recognizes the phosphorylated non-(S/T)P sites Ser-262 and/or Ser-356 (9) (Fig. 2). Phosphorylation of htau24S262A, htau24S356A, and htau24S262AS356A by NCLK plus heparin showed that Ser-262, but not Ser-356 was phosphorylated (Fig.  2). The effects of glycosaminoglycans (50 g/ml) other than heparin on tau phosphorylation by NCLK were investigated after 18 h of incubation (Fig. 3). Dextran sulfate produced a large effect, resulting in an approximately 3-fold stimulation of tau phosphorylation. Heparan sulfate and pentosan polysulfate also produced a large effect, with smaller effects for dermatan sulfate, chondroitin sulfate, and keratan sulfate. Hyaluronic acid, dextran, and poly-L-glutamic acid were without a significant effect. Addition of 50 g/ml nucleic acids also stimulated tau phosphorylation by NCLK, with tRNA producing a larger effect than DNA (Fig. 3). Incubation of tau with NCLK in the presence of 20 M tubulin led to an approximately 50% stimulation of tau phosphorylation (Fig. 3).
Effects of Glycosaminoglycans on Phosphorylation of Tau by GSK3␤ and MAP Kinase-Recombinant htau40 was incubated for 18 h with 1 unit/ml GSK3␤ purified from skeletal muscle or 1 units/ml activated recombinant p42 MAP kinase, in the presence or absence of 50 g/ml glycosaminoglycans, nucleic acids, and 20 M tubulin. As shown before, phosphorylation of tau by GSK3␤ was strongly stimulated by heparin. Dextran sulfate had a similar effect, with an almost 5-fold stimulation of phosphorylation (Fig. 3). Addition of heparan sulfate, chondroitin sulfate, pentosan polysulfate, and dermatan sulfate resulted in a 2.0 -3.0-fold stimulation of tau phosphorylation by GSK3␤, whereas keratan sulfate, hyaluronic acid, dextran, and poly-Lglutamic acid were without a significant effect. RNA and DNA produced a 2.5-3.0-fold stimulation of tau phosphorylation by GSK3␤; an effect of similar magnitude was obtained upon addition of tubulin (Fig. 3). Phosphorylation of htau40 by MAP kinase was stimulated 1.5-3.0-fold by heparin, heparan sulfate, pentosan polysulfate, dextran sulfate, and RNA (Fig. 3). Chondroitin sulfate, dermatan sulfate, keratan sulfate, hyaluronic acid, dextran, poly-L-glutamic acid, and DNA were without a significant effect. Incubation of tau with GSK3␤ or MAP kinase in the presence of 20 M tubulin led to a 3-fold stimulation of tau phosphorylation (Fig. 3).
Effects of Glycosaminoglycans on the Binding of Tau to Microtubules-Recombinant htau40 was incubated with different concentrations (50, 100, 250, 500, and 1,000 g/ml) of glycosaminoglycans or nucleic acids and added to taxol-stabilized microtubules, followed by ultracentrifugation to separate unbound tau from microtubule-bound tau. The addition of heparin, dextran sulfate, and pentosan polysulfate resulted in a dose-dependent inability of tau to bind to microtubules, with IC 50 values of approximately 100 g/ml (Fig. 4). Heparan sulfate and dermatan sulfate produced a smaller, but significant, reduction which amounted to 40% at 1 mg/ml (Fig. 4). Only small reductions in the ability of tau to bind to microtubules were observed upon addition of RNA and DNA. Hyaluronic acid, chondroitin sulfate, keratan sulfate, and poly-L-glutamic acid had no significant effect, even at high concentrations (Fig.  4).
Effects of Glycosaminoglycans on Tau-promoted Microtubule Stability-Recombinant htau40 was incubated with tubulin and microtubule assembly monitored by an increase in turbidity. After 5 min, when assembly was maximal, glycosaminoglycans or nucleic acids (100 g/ml) were added and microtubule disassembly monitored for a further 5 min by a decrease in turbidity. Addition of heparin, pentosan polysulfate, dextran sulfate, and DNA caused rapid and complete microtubule disassembly (Fig. 5). Heparan sulfate and dermatan sulfate had an intermediate effect (Fig. 5). RNA and poly-L-glutamic acid produced only a small effect on microtubule disassembly, whereas addition of hyaluronic acid, chondroitin sulfate, keratan sulfate, and dextran had no significant effect (Fig. 5).
Glycosaminoglycan-induced Assembly of Tau into Paired Helical-like Filaments-Incubation of the three repeat-containing tau isoform htau37 with glycosaminoglycans led to bulk assembly into twisted filaments with a morphology similar to the PHFs from Alzheimer's disease brain (Table II, Figs. 6 and 7). As observed before (28,29), incubation of four repeat-containing tau isoforms with glycosaminoglycans gave straight fila- ments with a morphology similar to the SFs from Alzheimer's disease brain (data not shown). Incubation of htau37 with heparin of molecular masses ranging from 3 to 30 kDa gave large numbers of twisted filaments (Fig. 6, A and B). A second class of filament appearing as thinner, wavy structures was also observed (shown in Fig. 6C). These may correspond to half-twisted filaments, as they are sometimes seen extending from the ends of twisted filaments. This would imply that the filaments formed in vitro are two-stranded, like Alzheimer filaments (37). The relative numbers of each type of filament were strongly dependent on the tau:heparin ratios, with a 2-fold change in the ratios being sufficient to switch from a predominance of half-filaments to a preponderance of twisted PHF-like filaments. Filaments with very similar morphologies were observed, when heparan sulfate was used instead of heparin (Fig. 6C). However, the total number of filaments was always higher with heparin. Addition of 1-10 M ZnCl 2 significantly increased the number of twisted filaments formed following addition of heparan sulfate to htau37 (Fig. 6D). Addition of 10 mM MgCl 2 led to a small increase in the number of twisted filaments, whereas CaCl 2 and AlCl 3 were without effect. Chondroitin sulfate and dermatan sulfate induced the assembly of tau into filaments with a morphology similar to those formed after addition of heparin or heparan sulfate (Fig.  7, A and B). However, the number of filaments was small relative to that observed after addition of heparin. Keratan sulfate, hyaluronic acid, dextran, and poly-L-glutamine failed to induce filament formation (Table II). This contrasted with the results obtained using the highly sulfated synthetic glycosaminoglycans dextran sulfate and pentosan polysulfate which induced the formation of very large numbers of short twisted tau filaments (Fig. 7, C and D, Table II). In contrast to the filaments obtained with naturally occurring sulfated glycosaminoglycans, these filaments were much shorter, consisting mostly of pieces less than one turn. We also examined the effects of tRNA and both single-stranded and double-stranded DNA on tau filament formation (Table II). Whereas DNA of various sizes failed to induce filament formation reproducibly, addition of tRNA to the three-repeat tau isoform htau37 resulted in the formation of twisted filaments with a similar morphology to PHFs (Fig. 8A). Addition of tRNA to the fourrepeat tau isoform htau40 gave straight filaments (Fig. 8B). DISCUSSION Abnormal tau filaments in the form of PHFs and SFs constitute one of the defining neuropathological characteristics of  Alzheimer's disease. We have recently shown that addition of heparin to full-length recombinant tau induces bulk assembly of tau into filaments that closely resemble the PHFs and SFs of Alzheimer's disease (28,29). Similar results have been reported with partial tau sequences encompassing the microtubule-binding repeat region (38). We also showed that tau is a heparin-binding protein and that heparin inhibits binding of tau to taxol-stabilized microtubules and induces rapid disassembly of tau-stabilized microtubules (28,29). Previous experiments have shown that phosphorylation of tau by a number of protein kinases is stimulated by heparin (11, 15, 30 -32). In the present study we have carried out an analysis of the effects of different glycosaminoglycans on stimulation of tau phosphorylation by the proline-directed protein kinases NCLK, GSK3␤, and MAP kinase, on the ability of tau to bind to microtubules, on disassembly of tau-stabilized microtubules, and on assembly of tau-containing filaments. We have found that the degree of sulfation is an important variable that determines the Alzheimer-like interactions of glycosaminoglycans with tau protein.
Thus, dextran sulfate was the most potent of the glycosaminoglycans tested, whereas dextran was without effect. In general, dextran sulfate, pentosan polysulfate, and heparin were the most effective, followed by heparan sulfate, chondroitin sulfate, and dermatan sulfate. Keratan sulfate, hyaluronic acid, and dextran were the least effective. Some of the differences in the effects of the moderately sulfated glycosaminoglycans cannot be accounted for by sulfation alone, indicating that sequence differences between individual glycosaminoglycans also play a role. Heparin markedly stimulated the phosphorylation of tau by NCLK, resulting in a 3-fold stimulation of phosphate incorpo-ration after an incubation of 18 h. The effect of heparin was most marked at early time points and it was maximal at 30 g/ml. Similar results were obtained with MAP kinase, adding NCLK and MAP kinase to the list of protein kinases whose ability to phosphorylate tau is stimulated by heparin. This list also includes cdc28, GSK3, cAMP-dependent protein kinase, SAPK1␥, SAPK3, and SAPK4 (11, 15, 30 -32). Phosphorylation of tau by NCLK in the presence of heparin produced the epitopes of phosphorylation-dependent anti-tau antibodies that recognize (S/T)P sites, in accordance with the known substrate specificities of NCLK. Unexpectedly, tau also became immunoreactive with antibody 12E8 which recognizes the non-(S/T)P sites Ser-262 and/or Ser-356 located in the microtubule-binding repeat region (9). By using tau mutants we could show that Ser-262, but not Ser-356, became phosphorylated. Hyperphosphorylation of tau at multiple (S/T)P sites and at Ser-262 is a pathological hallmark of PHF-tau. It has been suggested that phosphorylation of tau at Ser-262 plays an important role in regulating the ability of tau to bind to microtubules (39). Several non-proline-directed protein kinases have been shown to phosphorylate tau at Ser-262 (24,40,41). The present findings indicate that in the presence of sulfated glycosaminoglycans the proline-directed NCLK phosphorylates tau at multiple (S/ T)P sites, as well as at Ser-262.
The study of the influence of glycosaminoglycans on phosphorylation of tau by NCLK, MAP kinase, and GSK3 showed that heparin, dextran sulfate, pentosan polysulfate, and heparan sulfate were the most effective, whereas keratan sulfate, hyaluronic acid, dextran, and poly-L-glutamic acid had little or no effect. Chondroitin sulfate and dermatan sulfate had intermediate effects when tau was phosphorylated by NCLK and GSK3␤, but no effect when tau was phosphorylated by MAP kinase.
RNA stimulated the phosphorylation of tau by NCLK, GSK3␤, and MAP kinase, whereas DNA stimulated the phosphorylation of tau by NCLK and GSK3␤, suggesting that nucleic acids bind to tau in a similar manner to sulfated glycosaminoglycans. Tubulin similarly stimulated phosphorylation of tau by NCLK, GSK3␤, and MAP kinase. A previous study showed that tubulin stimulates phosphorylation of tau by GSK3␤ (42). We show that this is also true of phosphorylation by NCLK and MAP kinase. Tau binds through its positively charged repeat region to the negatively charged carboxyl terminus of tubulin (43,44). This may result in a conformational change in tau that renders some phosphorylation sites more accessible to protein kinases. The stimulation of tau phosphorylation by tubulin may represent a mechanism for regulating the binding of tau to microtubules in nerve cells.
The degree of glycosaminoglycan sulfation also influenced the ability of tau to bind to microtubules. Heparin, dextran sulfate, and pentosan polysulfate had a strong negative effect, with intermediate effects for heparan sulfate and dermatan sulfate. Hyaluronic acid, chondroitin sulfate, keratan sulfate, dextran, poly-L-glutamic acid, DNA, and tRNA had little or no effect. DNA had a strong negative effect toward tau-promoted microtubule assembly, similar to heparin, dextran sulfate, and pentosan polysulfate. Heparan sulfate, dermatan sulfate, and tRNA had intermediate effects, whereas hyaluronic acid, chondroitin sulfate, keratan sulfate, dextran, and poly-L-glutamic acid had little or no effect.
Heparin induced bulk assembly of three-repeat recombinant tau into twisted filaments similar to PHFs. As noted before (28), the molar tau:heparin ratio was of crucial importance, being optimal at approximately 4:1. When suboptimal, a preponderance of thin, wavy filaments was observed, with only few twisted filaments. The morphology of the thin, wavy filaments suggests that they may correspond to half-twisted filaments. Under non-optimal tau:heparin ratios, no filaments were formed. Dextran sulfate and pentosan polysulfate, two highly sulfated synthetic analogues of glycosaminoglycans, induced the formation of large numbers of filaments when incubated with the three-repeat tau isoform htau37 at a tau:glycosaminoglycan ratio of approximately 4:1. Unlike the filaments formed after addition of heparin, these filaments were short, suggesting that a high degree of glycosaminoglycan sulfation may affect the ratio of nucleation to growth. Incubation of threerepeat tau with heparan sulfate at a molar ratio of 4:1 led to the formation of twisted filaments with a similar morphology to the filaments formed after addition of heparin. The number of heparan sulfate-induced filaments was smaller, in keeping with its lower degree of sulfation. Addition of ZnCl 2 significantly stimulated the formation of tau filaments induced by heparan sulfate. Addition of MgCl 2 had a small stimulatory effect, whereas addition of NaCl, CaCl 2 , and AlCl 3 was without effect. The effective zinc concentration was 1-10 M, in the same range as that which has been shown to stimulate A␤ amyloid aggregation (45). Zinc is present at high concentrations in cerebral cortex and hippocampal formation (46), two brain regions which are particularly prone to develop the neuropathology of Alzheimer's disease. Moreover, high concentrations of zinc have been found in cycad-derived flour which has been linked to the high incidence of the amyotrophic lateral sclerosis/Parkinsonism-dementia complex on the island of Guam, a condition characterized by an extensive fibrillary tau pathology (47,48). Incubation of the three-repeat tau isoform htau37 with chondroitin sulfate and dermatan sulfate led to the formation of small numbers of twisted filaments with a similar morphology to those formed after addition of heparin and heparan sulfate. No filaments were formed after addition of keratan sulfate, hyaluronic acid, dextran, and poly-L-glutamic acid.
A recent study has shown the formation of tau filaments after incubation of recombinant tau with tRNA (33). We have observed a similar effect, with the formation of twisted filaments after incubation of the three-repeat tau isoform htau37 with tRNA. Similar to our previous results with sulfated glycosaminoglycans, the four-repeat tau isoform htau40 gave straight filaments when incubated with tRNA. By acridine orange staining, RNA has been shown to be sequestered in the neurofibrillary lesions of Alzheimer's disease (49).
Sulfated glycosaminoglycans and RNA share a repeat sugar backbone and negative charges in the form of sulfates or phosphates. This may be of relevance for the mechanism underlying tau filament formation. Tau protein is thought to be an extended molecule with little secondary structure which becomes partially structured upon binding to microtubules (50). Binding of sulfated glycosaminoglycans or RNA to tau may induce or stabilize a conformation of tau that brings the microtubulebinding repeats of individual tau molecules in close proximity, creating sites which favor polymerization of tau into filaments.
Immunohistochemical studies have shown the presence of heparan sulfate in nerve cells in Alzheimer's disease brain in the early stages of neurofibrillary degeneration which are characterized by the presence of hyperphosphorylated tau (28,(51)(52)(53). Chondroitin sulfate and dermatan sulfate proteoglycans have also been found in association with the neurofibrillary pathology of Alzheimer's disease (54,55). Sulfated glycosaminoglycans stimulate tau phosphorylation at lower concentrations than those required for tau filament formation. The pathological presence of heparan sulfate within the cytoplasm of nerve cells, perhaps as a result of leakage from membranebound Golgi compartments (56), could first lead to hyperphosphorylation of tau, resulting in its inability to bind to microtubules. At higher heparan sulfate concentrations tau could then assemble into PHFs and SFs.