Inhibition of Heparin-induced Tau Filament Formation by Phenothiazines, Polyphenols, and Porphyrins*

Tau protein is the major component of the intraneu-ronal filamentous inclusions that constitute defining neuropathological characteristics of Alzheimer’s disease and other tauopathies. The discovery of tau gene mutations in familial forms of frontotemporal dementia has established that dysfunction of the tau protein is sufficient to cause neurodegeneration and dementia. Here we have tested 42 compounds belonging to nine different chemical classes for their ability to inhibit he-parin-induced assembly of tau into filaments in vitro . Several phenothiazines (methylene blue, azure A, azure B, and quinacrine mustard), polyphenols (myricetin, epicatechin 5-gallate, gossypetin, and 2,3,4,2 spin column, as described (33). Column-ex- cluded tau was quantified, and the levels of tau-bound compounds were determined by measuring the absorbance of hematin at 400 nm, myri- cetin at 500 nm, and azure A at 670 nm. The binding of azure A, myricetin, and hematin to tau filaments was assessed by incubating 2 (cid:3) M of Sarkosyl-insoluble tau with 20 (cid:3) M of compound in 30 m M Tris-HCl, pH 7.5, for 12 h at 37 °C, followed by ultracentrifugation. Bound reagent was solubilized with 100 (cid:3) l of Me 2 SO, and the levels of tau-bound compounds were determined by measuring the absorbance as above. The chosen wavelengths were established as being characteristic of each compound by using scanning spectra from 200 to 750 nm. incubation of tau and heparin in the absence ( Con. ) or the presence of compounds (THBP, exifone, hematin, methylene blue, and APMT). SDS-stable, high molecular weight tau ( bracket and arrowhead ) was detected in the supernatants of samples incubated with inhibitory compounds (THBP, exifone, hematin, and methylene blue) but not in the supernatants of controls or samples incubated with the noninhibitory compound APMT. The presence of high molecular weight tau was accompanied by a reduction in the levels of monomeric tau in the pellet ( arrow ). A typical experiment is shown. Similar results were obtained in three separate experiments.

Alzheimer disease (AD) 1 is characterized by the presence of two kinds of filamentous aggregates in the brain, known as neuritic plaques and neurofibrillary lesions (1). The extracellular plaque deposits are made of ␤-amyloid, a 40 -42-amino acid proteolytic fragment of the amyloid precursor protein. The identification of the genetic defects that cause early onset AD has established that A␤ production and aggregation are central to the development of AD (2)(3)(4)(5). The intracellular neurofibrillary lesions consist of paired helical filaments and straight filaments, which are made of the microtubule-associated protein tau in a hyperphosphorylated state (1). The temporal and spatial accumulation of hyperphosphorylated tau protein correlates with nerve cell loss and the severity of dementia.
Tau inclusions, in the absence of extracellular deposits, are characteristic of progressive supranuclear palsy, corticobasal degeneration, Pick's disease, argyrophilic grain disease, and frontotemporal dementia and parkinsonism linked to chromosome 17 (FTDP-17) (6). The identification of mutations in the tau gene in FTDP-17 has established that dysfunction of tau protein is central to the neurodegenerative process (7)(8)(9). Currently, more than 30 different mutations in the tau gene have been described in over 100 families with FTDP-17.
Fibrils formed from A␤ display a prototypical cross-␤ structure characteristic of amyloid (10). The same structure has been demonstrated for many other filaments deposited in the extracellular space in systemic and organ-specific amyloidoses (11), including prion protein deposits in Gerstmann-Strä ussler-Scheinker syndrome and some cases of Creutzfeldt-Jakob disease (12). Considerably less work has been done on filaments that form intracellularly. Filaments assembled from ␣-synuclein were shown by various techniques to possess cross-␤ structure (13), as were the synthetic filaments formed from exon 1 of huntingtin with 51 glutamines (14). More recently, tau filaments, both those extracted from human brain and those assembled in vitro from recombinant protein, were also shown to have a cross-␤ structure (15). It is therefore appropriate to consider the tauopathies a form of brain amyloidosis.
The conversion of a small number of soluble peptides and proteins into insoluble filaments is believed to be the central event in the etiology of the most common neurodegenerative diseases (16,17). Consequently, many current therapeutic strategies are aimed at inhibiting filament formation and at promoting filament clearance. They include the use of antibodies, synthetic peptides, heat shock proteins, and chemical compounds. Among these, small organic molecules have been ex-tensively tested for their ability to inhibit filament formation in vitro, particularly in relation to A␤ deposition (18), the formation of protease-resistant forms of the prion protein (19), and the aggregation of huntingtin (20). By contrast, there are only a few reports of therapeutic strategies aimed at inhibiting tau filament formation. In one study, the formation of aggregates made of truncated tau protein was inhibited by a number of phenothiazines, although the filamentous nature of the aggregates was not demonstrated (21). Phenothiazines have also been shown to inhibit the conversion of soluble prion protein into the protease-resistant form (22)(23)(24)(25). In a second study, arachidonic acid-induced tau filament formation was found to be inhibited by a benzothiazole compound that did not affect A␤ or ␣-synuclein assembly (26). Finally, a recent study reported the inhibition of heparin-induced tau filament formation by a number of compounds, including daunorubicin hydrochloride and 2,3,4-trihydroxybenzophenone (THBP) (27).
Here we have investigated the effects of 42 compounds belonging to nine different chemical classes on heparin-induced tau filament formation. The ability of these compounds to inhibit the assembly of A␤ was investigated in parallel. Filament formation was assessed using electron microscopy, thioflavin S (ThS) fluorescence, and the formation of Sarkosyl-insoluble tau protein. Several compounds belonging to the phenothiazine, polyphenol, and porphyrin classes were found to inhibit tau filament formation by binding to soluble, SDS-stable tau oligomers.

MATERIALS AND METHODS
Expression and Purification of Tau Protein-The 412-amino acid isoform of human brain tau (28) was expressed from cDNA clone htau46 in Escherichia coli BL21(DE3), purified as described (29), and dialyzed against 30 mM Tris-HCl, pH 7.5. Following separation by reverse phase high pressure liquid chromatography (Aquapore RP300 column), the absorbance at 215 nm was measured and compared with that of tau protein of known concentration, to give the concentration of the freshly purified protein.
Tau Filament Assembly and Inhibitor Testing-Purified recombinant tau (1 mg/ml) and heparin (0.1 mg/ml) were incubated at 37°C for 72 h in 50 l of 30 mM Tris-HCl, pH 7.5, containing 0.1% sodium azide (30), in the presence or absence of the compounds listed in Table I. The different compounds were used at concentrations ranging from 20 nM to 200 M. For a semi-quantitative assessment of filament formation, electron microscopy was used. Aliquots (2 l) of assembly mixtures were placed on collodion-coated 300-mesh copper grids, stained with 2% sodium phosphotungstate, and micrographs recorded at a nominal magnification of ϫ30,000 on a JEOL 1200EXII electron microscope. For a quantitative assessment of filament formation, ThS fluorescence and levels of Sarkosyl-insoluble tau were assessed. For ThS fluorescence, aliquots (20 l) of assembly mixtures were removed at various time points and brought to 300 l with 5 M ThS (Sigma) in 20 mM MOPS, pH 6.8. Fluorimetry was performed using a Hitachi F4000 fluorescence spectrophotometer (set at 440 nm excitation/521 nm emission) (31,32). Sarkosyl-soluble and -insoluble tau was prepared as follows. Aliquots (25 l) of assembly mixtures were removed, added to 125 l of 30 mM Tris-HCl, pH 7.5, containing 1% Sarkosyl, and left at room temperature for 1 h. The mixtures were then spun at 150,000 ϫ g for 20 min. The supernatants (Sarkosyl-soluble tau) were removed, and the pellets (Sarkosyl-insoluble tau) were resuspended in 25 l of SDS sample buffer containing 0.1% 2-mercaptoethanol, prior to SDS-PAGE. Following staining with Coomassie Brilliant Blue, the intensities of the Sarkosyl-insoluble tau bands were quantified by scanning densitometry, as described (29). Each compound was tested at least three times at each concentration, and the results were expressed as % of tau assembly in the absence of any compound, taken as 100. Statistical analysis was carried out by unpaired t test using Kai plot software, and the results were expressed as the means Ϯ S.D. IC 50 values were calculated for each compound by using the levels of Sarkosyl-insoluble tau.
A␤ Filament Assembly and Inhibitor Testing-Monomeric A␤-(1-40) (1 mM, Peptide Institute Inc., Osaka, Japan) in distilled water was diluted with phosphate-buffered saline, pH 7.5, to a final concentration of 15 M and incubated at 37°C for 24 h in 40 l of phosphate-buffered saline, containing 5 M thioflavin T, in the presence or absence of compounds (0.3-40 M). Fluorimetry was performed using a Biolu-min960 fluorescence spectrophotometer (set at 450 nm excitation/485 nm emission, Amersham Biosciences) (31), and IC 50 values were calculated for each compound.
Disassembly of Tau Filaments-Sarkosyl pellets of recombinant tau assemblies were washed and resuspended at 4 M (0.2 mg/ml) in 30 mM Tris-HCl, pH 7.5. Following the addition of compounds (concentration range of 20 nM to 200 M), the assemblies were incubated for 24 h at 30°C, followed by a 20-min centrifugation at 150,000 ϫ g. Supernatants and pellets were resuspended in SDS sample buffer containing 2-mercaptoethanol and analyzed by SDS-PAGE. Each experiment was repeated at least three times.
Microtubule Binding and Assembly Assays-To examine the effects of inhibitors of tau filament assembly on the ability of tau to bind to microtubules, recombinant tau protein at 2 M (0.1 mg/ml) was incubated with various concentrations of inhibitors (0.2-20 M) in assembly buffer (80 mM Pipes, 1 mM EGTA, 0.2 mM MgCl 2 , 1 mM dithiothreitol, 1 mM GTP, pH 6.8) for 30 min at 37°C. Taxol-stabilized microtubules (1 mg/ml) were added, and the mixture was incubated for 20 min at 37°C. Following a 20-min spin at 150,000 ϫ g, aliquots of the supernatants (tau not bound to microtubules) and pellets (tau bound to microtubules) were subjected to SDS-PAGE and stained with Coomassie Brilliant Blue. Each experiment was repeated three times. To study the effects of inhibitors of tau filament assembly on the ability of tau to promote microtubule assembly, recombinant tau protein was incubated with the inhibitors (0.2-20 M) as described above. Tubulin (20 M, Cytoskeleton) was added, and the mixture was incubated for 20 min at 37°C, as described previously (29). The assembly of tubulin into microtubules was monitored over time by a change in turbidity at 350 nm. Each experiment was repeated three times. Several compounds gave a weak absorbance at 350 nm, which was subtracted.
Binding of Tau to Heparin-Sepharose-Recombinant tau protein (1 mg/ml) was incubated with heparin-Sepharose (AF-heparin Toyopearl 650 M) at 37°C for 1 h in the presence or absence of the various compounds (0.2-200 M). Bound tau was eluted sequentially with 30 mM Tris-HCl, pH 7.5, containing 0.1, 0.3, and 0.5 M NaCl and run on SDS-PAGE.
Preparation and Characterization of Monomeric Tau, Filamentous Tau, and Aged Soluble Tau-Freshly prepared protein was the source of monomeric tau. Sarkosyl-insoluble pellets of tau assembly mixtures were the source of filamentous tau. Aged soluble tau was defined as the protein remaining in the supernatant following the ultracentrifugation of tau assembly mixtures incubated with heparin for 72 h. Limited protease digestion was used for further characterization. Monomeric tau, aged soluble tau, and filamentous tau were incubated with 20 and 200 g/ml trypsin for 30 min at 37°C, and the digests were analyzed by SDS-PAGE. Each experiment was repeated at least three times. In some experiments, aged soluble tau was fractionated by gel filtration chromatography using a TSK 3000SWXL column in 50 mM phosphate buffer, pH 7.0, containing 0.15 M NaCl. Column fractions were analyzed by SDS-PAGE in the presence or absence of 2-mercaptoethanol. As a control, heparin was added to freshly prepared tau and loaded onto the column without further incubation.
Interaction of Compounds with Monomeric Tau, Aged Soluble Tau, and Filamentous Tau-Monomeric, aged soluble, and filamentous tau were incubated with the various compounds (2-20 M) for 12 h at 37°C, followed by the addition of sample buffer containing 2-mercaptoethanol and SDS-PAGE. The binding of compounds to monomeric and oligomeric tau was investigated by incubating 2 M protein purified by gel filtration with 20 M azure A, myricetin, and hematin in 30 mM Tris, pH 7.5, for 12 h at 37°C. Bound and unbound compounds were separated using a Sephadex G-25 spin column, as described (33). Column-excluded tau was quantified, and the levels of tau-bound compounds were determined by measuring the absorbance of hematin at 400 nm, myricetin at 500 nm, and azure A at 670 nm. The binding of azure A, myricetin, and hematin to tau filaments was assessed by incubating 2 M of Sarkosyl-insoluble tau with 20 M of compound in 30 mM Tris-HCl, pH 7.5, for 12 h at 37°C, followed by ultracentrifugation. Bound reagent was solubilized with 100 l of Me 2 SO, and the levels of taubound compounds were determined by measuring the absorbance as above. The chosen wavelengths were established as being characteristic of each compound by using scanning spectra from 200 to 750 nm.

Inhibition of Tau Filament
Assembly-42 compounds belonging to nine different chemical classes (listed in Table I) were tested for their ability to inhibit the heparin-induced assembly of tau into filaments. By electron microscopy, a large reduction in the number of tau filaments was observed in the presence of several phenothiazines (methylene blue, lacmoid, azures A-C, and thionin), polyphenols (exifone, gossypetin, 2,3,4-trihydroxybenzophenone, purpurogallin, 2,3,4,2Ј,4Ј-pentahydroxybenzophenone, myricetin, and epicatechin 3-gallate), and porphyrins (hemin chloride, hematin, ferric dehydroporphyrin IX, and phthalocyanine). Other phenothiazines (chlorpromazine hydrochloride, perphenazine, promazine hydrochloride, acetopromazine maleate, and propionylpromazine hydrochloride) and polyphenols (apigenin, epicatechin and 2,2-dihydroxybezophenone) did not affect filament formation. Benzothiazoles, rifamycins, polyene macrolides, anthracyclines, sulfonated dyes, and methyl yellow were also without effect. A representative example of each class of compound is shown in Fig. 1A. The same compounds were tested using the ThS fluorescence assay (Fig. 1B). A reduced fluorescence was observed in the samples incubated with methylene blue, exifone, and hemin chloride, in agreement with the electron microscopic findings. In addition, tau filament formation was quantified by measuring the levels of Sarkosyl-insoluble tau. As shown in Fig. 1, C and D, little or no Sarkosyl-insoluble tau was detected in tau assembly mixtures incubated in the presence of methylene blue, exifone, or hemin chloride, consistent with the findings by electron microscopy and ThS fluorescence.
We next determined the IC 50 values of phenothiazines, polyphenols, and porphyrins for inhibiting tau filament assembly by quantifying the levels of Sarkosyl-insoluble tau. Inhibition of tau filament formation was concentration-dependent, as illustrated in Fig. 1E for polyphenol compounds. The IC 50 values of phenothiazines, polyphenols, and porphyrins for inhibiting tau filament assembly are given in Table II, where they are compared with the ability of the same compounds to inhibit A␤ aggregation. A strong inhibitory effect on tau filament assembly (IC 50 values of 1-2 M) was observed with myricetin, ferric dehydroporphyrin IX, epicachetin 3-gallate, methylene blue, azure B, and gossypetin. Other compounds, such as 2,3,4,2Ј,4Јpentahydroxybenzophenone, azure A, exifone, purpurogallin, and azure C were less strongly inhibitory (IC 50 values of 2-10 M). A weak inhibitory effect (IC 50 values of 10 -70 M) was observed with hematin, thionin, 2,3,4-trihydroxybenzophenone, hypericin, pseudohypericin, lacmoid, and phthalocyanine. No inhibition (IC 50 values of Ͼ200 M) was observed with apigenin, epicatechin, chlorpromazine hydrochloride, perphenazine, promazine hydrochloride, acetopromazine, and propionylpromazine. Overall, compounds that strongly inhibited tau filament assembly were also potent inhibitors of A␤ aggregation (Table II).
Disassembly of Tau Filaments-To investigate whether compounds can disassemble tau filaments, Sarkosyl-insoluble tau was incubated at 37°C for 24 h with each of the compounds listed in Table I. Following ultracentrifugation, the pattern of tau bands and their relative levels in supernatant and pellet were assessed by SDS-PAGE. Representative findings with a polyphenol, a phenothiazine, and a porphyrin are given in Fig.  2, A and B. Sarkosyl pellets of recombinant tau assemblies incubated in the absence of compound served as the control. Approximately 20% of tau was recovered in the supernatant and 80% in the pellet, indicating some solubilization of tau filaments, even in the absence of compounds (Fig. 2, A and B).
Upon incubation with THBP, monomeric tau in supernatant and pellet was reduced, and an increase in SDS-stable, high molecular weight (HMW) tau was observed at the top of the gel ( Fig. 2A). By contrast, the noninhibitory polyphenols apigenin, epicatechin, and dihydrobenzophenone did not affect the running of tau on SDS-PAGE. Following incubation with methylene blue, the band corresponding to monomeric tau was weaker in the supernatant and absent from the pellet. Instead, several HMW tau bands appeared (Fig. 2, A and B). A similar band shift was observed with all the phenothiazines that inhibited tau filament formation. No shift was detected with phenothiazines devoid of inhibitory activity (data not shown). The incubation of tau with phthalocyanine led to a concentration-dependent reduction in monomeric tau in the pellet and a corresponding increase in the supernatant (Fig. 2, A--C), suggestive of the disassembly of tau filaments. This was confirmed by electron microscopy, where a reduction in filament numbers was observed with phthalocyanine but not with THBP or methylene blue (Fig. 2D). Other porphyrin compounds, such as hematin, showed a similar depolymerizing effect at high concentration (data not shown).
Microtubule Binding and Assembly Activities of Tau in the Presence of Inhibitory Compounds-Taxol-stabilized microtubules were incubated with tau and the inhibitory compounds exifone, methylene blue, and hematin. Bound and unbound tau were separated by ultracentrifugation and analyzed by SDS-PAGE. The three compounds did not significantly affect the ability of tau to bind to microtubules (Fig. 3, A and B). The effects of exifone, methylene blue, and hematin on the ability of tau to promote microtubule assembly were also investigated. Tau and inhibitory compound were mixed together, followed by the addition of tubulin and the monitoring of microtubule assembly. A representative experiment is shown in Fig. 3C, and the means Ϯ S.D. (n ϭ 3) of the optical densities at 1 min are shown in Fig. 3D. Inhibitory compounds failed to influence the ability of recombinant tau to promote microtubule assembly.
Heparin Binding of Tau in the Presence of Inhibitory Compounds-Recombinant tau with or without added exifone, methylene blue, and hematin was incubated with heparin-Sepharose, and bound tau eluted with 0.1, 0.3, and 0.5 M NaCl. The binding of tau was not affected by the presence of inhibitory compounds (Fig. 4A). Moreover, similar amounts of tau protein were eluted with 0.3 M NaCl (Fig. 4B) and 0.5 M NaCl (data not shown) from control samples and from samples with added inhibitor.
Interaction of Inhibitory Compounds with Sarkosyl-soluble and Sarkosyl-insoluble Tau-Recombinant tau and heparin were incubated at 37°C for 72 h in the absence or the presence of the inhibitory compounds THBP, exifone, methylene blue, and hematin or the noninhibitory compound APMT. Sarkosylsoluble (Fig. 5A) and Sarkosyl-insoluble (Fig. 5B) fractions were prepared and analyzed by SDS-PAGE. In the presence of inhibitory compounds, HMW tau oligomers were present in the Sarkosyl-soluble fraction. No HMW tau was detected in Sarkosyl-soluble tau incubated in the absence of inhibitors or in the presence of APMT (Fig. 5A). The level of HMW tau increased with increasing concentrations of inhibitory compounds and correlated inversely with the amount of Sarkosyl-insoluble tau.
Interaction of Inhibitory Compounds with Monomeric Tau, Aged Soluble Tau, and Filamentous Tau-THBP, exifone, methylene blue, hematin, and APMT were incubated for 12 h at 37°C with monomeric, aged soluble, and filamentous tau. As judged by SDS-PAGE, these compounds did not affect the running pattern of monomeric tau (Fig. 6A). By contrast, when aged soluble tau was incubated with inhibitory compounds THBP, exifone, methylene blue, and hematin, tau bands of Ͼ150-kDa apparent molecular mass appeared (Fig. 6B). No such bands were seen following the incubation of aged soluble tau with APMT. Incubation of tau filaments with THBP and methylene blue resulted in a marked decrease in monomeric tau and an increase in tau that accumulated at the top of the gel (Fig. 6C, arrowhead). Similar changes were observed upon the addition of exifone and hematin (Fig. 6C).
Fractionation and Characterization of Aged Soluble Tau-To characterize the molecular species in aged soluble tau that bind A, semiquantitative electron microscopic assessment of tau filaments formed in the absence (Control) or the presence of compounds. Filament numbers similar to controls were observed in the presence of APMT, rifamycin B, mycostatin, daunorubicin, Ponceau SS, and methyl yellow. By contrast, only small numbers of filaments were seen in the presence of exifone, hemin chloride, and methylene blue. B, quantitation of tau filament formation using ThS fluorescence. A significant reduction in fluorescence was observed in the presence of exifone, hemin chloride, and methylene blue (* ϭ p Ͻ 0.001, n ϭ 6, unpaired t test). Results are given as means Ϯ S.E. and expressed as % fluorescence of controls (taken as 100%). Statistical analysis was done by unpaired t test using Kai plot software. C-E, quantitation of tau filament formation using levels of Sarkosyl-insoluble tau. C and D, Coomassie Brilliant Blue staining and quantitation by densitometry. A significant reduction in Sarkosyl-insoluble tau was observed in the presence of exifone, hemin chloride, and methylene blue (* ϭ p Ͻ 0.001, n ϭ 6, unpaired t test). E, concentration-dependent inhibition of tau filament formation by six different polyphenols. Sarkosyl-insoluble tau was analyzed following a 72-h incubation of tau and heparin with exifone inhibitory compounds and give rise to SDS-stable HMW protein, we incubated recombinant tau with heparin for 72 h and fractionated the supernatant of the assembly mixture by gel filtration chromatography, followed by SDS-PAGE (Fig. 7A). Freshly prepared tau run in the presence of heparin was recovered around the position of bovine gamma globulin (Fig.  7A). Following a 72-h incubation with heparin, aged soluble tau eluted as a single peak in the void volume (Fig. 7A). By SDS-PAGE, in the absence of reducing agent, only HMW tau was detected in the material eluting in the void volume (Fig. 7B). However, in the presence of 2-mercaptoethanol, this material ran at the position of monomeric tau. Limited protease digestion was used for further characterization of monomeric, aged soluble, and filamentous tau. As shown in Fig. 7C, aged soluble and filamentous tau showed a similar resistance to digestion with 20 and 200 g/ml trypsin. By contrast, monomeric tau was completely digested under these conditions.
Binding of Inhibitory Compounds to Monomeric Tau, Aged Soluble Tau, and Filamentous Tau-To investigate the binding of inhibitory compounds to tau, we incubated monomeric tau (material eluting at 150-kDa apparent molecular mass on gel filtration), oligomeric tau (material eluting in the void volume following gel filtration of recombinant tau incubated with hep-arin for 72 h), and filamentous tau (Sarkosyl-insoluble material) with azure A, myricetin, and hematin. These compounds were chosen, because their concentrations can be determined spectrophotometrically. The phenothiazine compound azure A bound to filamentous tau, with only little binding to either monomeric or aged soluble tau (Fig. 8A). Myricetin, a potent polyphenol inhibitor of tau filament formation, bound more strongly to filamentous and aged soluble tau than to the monomeric protein (Fig. 8B). This contrasted with the porphyrin compound hematin that bound equally well to monomeric, aged soluble, and filamentous tau (Fig. 8C). DISCUSSION Thioflavin S fluorescence, levels of Sarkosyl-insoluble tau, and electron microscopic examination were used to investigate the effects of nine classes of compounds on the heparin-induced assembly of tau into filaments and on the disassembly of filaments. All three methods measured the amount of tau filament assembly and identified the same inhibitory compounds. As a benzothiazole, thioflavin S could not be used for testing inhibition by this class of compounds.
Several phenothiazines (methylene blue, azure A, azure B, and quinacrine mustard), polyphenols (myricetin, epicatechin 3-gallate, gossypetin, and 2,3,4,2Ј,4Ј-pentahydroxyphenone), and the porphyrin ferric dehydroporphyrin IX inhibited tau filament formation with IC 50 values that were in the low micromolar range. Other compounds inhibited tau filament formation less potently or were without effect. In contrast to the significant number of compounds that inhibited tau assembly, only the porphyrin phthalocyanine was able to disassemble tau filaments. Phthalocyanine also inhibited tau assembly, albeit not very potently (IC 50 ϭ 67 M). All experiments used heparin as a co-factor for inducing filament assembly (30). In order to rule out a potential influence of heparin, we studied the binding of tau to heparin-Sepharose in the presence of inhibitory compounds. None of the compounds interfered with the binding of tau to heparin. The binding of one member of each class of inhibitory compound to monomeric, oligomeric, and filamentous tau was also investigated. The phenothiazine azure A bound to filamentous tau, with only little binding to either monomeric or oligomeric protein. This suggests that azure A binds preferentially to tau in a cross-␤ conformation. The polyphenol myricetin bound more strongly to oligomeric and filamentous tau than to monomeric protein, whereas the porphyrin compound hematin bound equally well to monomeric, oligomeric, and filamentous tau.
The repeat region of tau is important, not only for filament formation but also for the ability to interact with microtubules (6). Therefore, we investigated the ability of tau to promote microtubule assembly in the presence of inhibitory compounds. None of the inhibitors influenced tau-promoted microtubule assembly, in line with previous findings indicating that distinct conformations are involved in physiological tau folding and pathological tau aggregation (34).
Inhibition of A␤ fibril formation was also investigated, and in general a good correlation was observed between the relative potencies of phenothiazines, polyphenols, and porphyrins in inhibiting the assembly of A␤ and tau. This confirms and extends previous work describing the inhibition of A␤ fibril formation by a number of polyphenols and porphyrins (35)(36)(37). It also establishes that some phenothiazine compounds are potent inhibitors of A␤ fibril formation. Similar mechanisms may thus underlie A␤ and tau assembly, and a given compound might be effective against both pathologies However, we also observed that compounds belonging to several chemical classes inhibited A␤ fibril formation but not tau assembly. Several of these compounds have been shown previously to inhibit A␤ fibril formation (38 -40). Benzothiazoles similar to the ones tested here are being developed as amyloid-imaging agents (41). They have been shown to bind A␤ deposits with higher affinity than tau inclusions, in agreement with the present findings. In contrast, a recent study (26) has reported the inhibition of arachidonic acid-induced tau filament formation by a benzothiazole compound that did not affect the assembly of either A␤ or ␣-synuclein. The same compound has also been reported to inhibit neurodegeneration in a lamprey model of tauopathy (42). Furthermore, some benzothiazoles have been shown to inhibit the assembly of exon 1 of huntingtin (20). It remains to be determined whether the heparin-induced assem- FIG. 2. Disassembly of tau filaments  by phthalocyanine. A and B, tau filaments incubated in the absence (Control) or the presence of compounds THBP, methylene blue, and phthalocyanine were ultracentrifuged, and the resulting supernatants and pellets were analyzed by SDS-PAGE. The positions of monomeric and aggregated tau are indicated by arrows and arrowheads, respectively. Note that phthalocyanine effected an increase in monomeric tau in the supernatant and a reduction in monomeric and aggregated tau in the pellet. C, concentrationdependent increase in monomeric tau in the supernatant and decrease in the pellet following incubation of tau filaments with phthalocyanine. D, electron microscopic analysis of tau filaments incubated in the absence (Control) or the presence of compounds (THBP, methylene blue, and phthalocyanine).
bly of tau can be inhibited by these compounds. None of the compounds tested here inhibited tau filament formation without inhibiting A␤ fibril formation.
Analysis of the Sarkosyl-soluble fraction of tau assemblies by SDS-PAGE under reducing conditions revealed large amounts of HMW tau in the presence of inhibitory compounds. HMW tau also appeared upon incubation of inhibitory compounds with preformed aged soluble and filamentous tau but not following incubation with monomeric tau. Aged soluble tau was further characterized by gel filtration chromatography in the absence of inhibitory compounds, where it eluted in the void volume of the column, with monomeric tau eluting at the position of gamma globulin, corresponding to an apparent molecular mass of 150 kDa. Abnormal elution profiles of this kind (tau is ϳ50 kDa) are typical of natively unfolded proteins (43). By SDS-PAGE, under nonreducing conditions, the material eluting in the void volume consisted entirely of HMW tau. However, under reducing conditions, it ran at the position of monomeric tau, indicative of disulfide bond-dependent oligomerization. Aged soluble tau exhibited partial protease resistance similar to that of filamentous tau. These findings indicate that oligomeric tau may be an intermediate in the pathway from monomeric to filamentous tau and suggest that the compounds investigated here may be inhibitory toward tau filament assembly by virtue of their ability to bind to and stabilize disulfidelinked tau oligomers. More work is required to establish the precise molecular nature of these tau oligomers. Previous work has suggested that the heparin-induced assembly of tau may involve multimerization prior to filament formation (44). The mechanisms underlying the formation of HMW tau in the presence of inhibitory compounds remain to be identified. Particularly large amounts of HMW tau formed when aged soluble tau was incubated with polyphenols, including THBP and exifone. Recent work (45)(46)(47) has reported inhibition of ␣-synuclein aggregation and prion protein formation by a number of polyphenols, including catechol-containing compounds and flavonoids. These studies have indicated that the inhibition of ␣-synuclein assembly by catecholamines and flavonoids was the result of covalent modifications of oligomeric protein by quinones. Like catechol-containing compounds and flavonoids, the polyphenols that inhibited tau filament formation had two adjacent phenolic -OH groups, whereas noninhibitory compounds lacked these. It is therefore conceivable that similar covalent modifications of oligomeric tau by polyphenols may have caused the formation of HMW tau that was stable on SDS-PAGE under reducing conditions. A recent study (27) reported that THBP was unable to inhibit heparin-induced filament formation of four-repeat tau, while potently inhibiting the assembly of three-repeat protein. We found that THBP was equally strongly inhibitory toward three-and four-repeat tau (data not shown). In contrast to our findings, the above study also reported inhibition of tau filament formation by the anthracycline compound daunorubicin hydrochloride. The reasons for these discrepancies are unknown but may be related to the different assay conditions used. Unlike the present study, Pickhardt et al. (27) tested the ability of their compounds to inhibit the assembly of four-repeat tau at high temperature and under strongly reducing conditions.
Members of the phenothiazine class of compounds were shown previously to inhibit the aggregation of truncated tau protein (21). We extended these findings to the full-length protein, the filament formation of which was inhibited by a number of phenothiazines that bound filamentous tau in preference to the monomeric and oligomeric proteins. When comparing the structures of active and inactive phenothiazines, with the exception of quinacrine mustard, compounds lacking a side chain at position 9 of the phenothiazine ring were found to be strongly inhibitory, whereas those with a side chain were either weakly inhibitory or had no effect. Phenothiazines have also been shown to inhibit the conversion of soluble prion protein into the protease-resistant form (22)(23)(24)(25). However, unlike the present study, phenothiazines with a side chain were found to be inhibitory.
Inhibition of tau filament formation by porphyrin compounds is demonstrated here for the first time. Incubation with porphyrins resulted in the formation of HMW tau that was stable by SDS-PAGE under reducing conditions, although the levels were lower than for tau incubated with polyphenols and phenothiazines. Unlike the latter, porphyrins bound equally well to monomeric, oligomeric, and filamentous tau. Phthalocyanine, a porphyrin, was the only compound able to disassemble tau filaments. It has been shown previously to inhibit fibril formation of ␣-synuclein by interacting with the monomeric protein (33). In addition, phthalocyanine has been found to exhibit anti-scrapie activity, probably through inhibition of the formation of protease-resistant prion protein (48,49). It remains to be seen whether endogenous porphyrins may play a role in protecting against tau filament formation in brain. Increased levels of heme oxygenase-1, the enzyme that catalyzes the degradation of heme to biliverdin and iron, have been detected in AD brain, where heme oxygenase-1 co-localized with the neurofibrillary pathology (50).
It remains to be determined whether the oligomeric tau species that formed upon binding of inhibitory compounds protect against neurodegeneration in vivo. This can be investigated by the administration of inhibitory compounds to transgenic mice that exhibit the essential molecular and cellular features of the human tauopathies, including the formation of abundant filaments made of hyperphosphorylated tau protein and nerve cell loss (51,52). Of the inhibitory compounds identified here, members of the phenothiazine class are known to cross the blood-brain barrier and to be relatively nontoxic (53). Future experiments using inhibitory compounds may throw light on the as yet unresolved question of what constitutes the toxic tau species. In conclusion, the identification of phenothiazine, polyphenol, and porphyrin inhibitors of tau filament formation will form a starting point for the development of mechanism-based therapies for the tauopathies.