The tau protein (residues 297-391) forms the core of oligomers and filaments that drive neurodegeneration in Alzheimer’s disease (AD). The mechanisms of pathological assembly and pharmacological inhibition need to be understood before novel treatments can be developed. Molecular dynamics modeling, confirmed by immunochemistry, revealed that self-assembly involves initial capture via residues 337-355 and a conformational switch at proline 332, followed by unfolding and zipper-like assembly into a stable H-bonded structure that occludes epitopes within the core domain. Hydromethylthionine (HMT), a clinically active tau aggregation inhibitor (TAI), works by stabilizing the core monomer in a wrapped-up conformation, which prevents its assembly. A pharmacophore model based on the HMT cryptic pocket was used to design novel active TAIs. These discoveries provide a basis for the rational design of new TAIs for the treatment of AD.
Research Article
Mechanism of core-tau unit assembly and inhibition in Alzheimer’s disease
https://doi.org/10.21203/rs.3.rs-4258009/v1
This work is licensed under a CC BY 4.0 License
Version 1
posted
You are reading this latest preprint version
Discovery of the site and mechanism of action of an inhibitor of tau pathology shown to slow Alzheimer’s disease progression provides a basis for the design of new active drugs.
Tau protein self-assembles to form paired helical filaments (PHFs), which correlate with clinical decline in Alzheimer’s disease (AD).1-4 A tau fragment isolated from the proteolytically stable PHF core5 comprising tau297-391 (dGAE) assembles spontaneously into PHFs in vitro.6-9 Hydromethylthionine (HMT) is a potent inhibitor of this assembly at a protein:HMT stoichiometric ratio of 1:0.1 in vitro.10 HMT has exposure-dependent pharmacological activity on clinical decline and brain atrophy in patients with AD and frontotemporal dementia (FTD) at a dose of 8 mg/day.11,12 We have sought to understand how HMT acts within the PHF core and have used a pharmacophore model based on this understanding to design chemically distinct tau aggregation inhibitors (TAIs).
Although the static structure of the PHF cores in both AD and FTD are available at atomic resolution for residues 304-380,13-15 the mechanisms of assembly are not known. To address this problem, we used molecular dynamics (MD) simulations to determine the mechanism of core monomer incorporation into an existing oligomer. The sequence of occlusion of epitopes predicted by the MD mechanism was confirmed using a newly developed panel of monoclonal antibodies. We next used MD simulation to identify a unique cryptic binding pocket within the tau295-391 monomer that permits the inhibition of aggregation. We report that HMT works by stabilizing the monomer in an alternative conformation that prevents the monomer from participating in assembly. A pharmacophore model based on the HMT pocket was used to design a novel series of TAIs. For the first time, our results provide a basis for understanding the mechanisms of both PHF assembly and its inhibition by novel TAIs for the treatment of AD and related tauopathies.
Modeling of the assembly of PHF-core tau monomers on an oligomeric template
A tau295-391 monomer was constructed using the AD PHF structure (PDB 5O3L) by modeling the additional residues 295-306 and 379-391 (13). Initially, thirty independent 10-ns simulations were performed and analyzed using principal component analysis (PCA) of residue fluctuations to identify common assembly trajectories.16 This permitted the identification of twenty clusters (Table S1) comprising 380 protein conformations. A core pentamer comprising tau306-378 based on PDB 5O3L was used as a template to investigate the mechanism of oligomer elongation (Figure 1 and Supplementary Figure S1). The simulation was performed in two stages: first, to identify the initial anchoring site of the monomer; second, to examine its incorporation into the pentamer using nudged elastic band MD simulations.
The tight hairpin loop of Val337-Gly355 (Figure 1, Frame 1, cyan) constitutes the site of initial contact between the monomer and the assembled PHF model. This sequence contains numerous charged residues, including four acidic (Asp and Glu), five basic (Lys and Arg) and three polar (Ser and Gln) side chains. Attractive electrostatic forces contribute to the initial anchoring of the monomer to the core template. The monomer remains in a partially disordered state following anchoring of the hairpin. Over a 40-ns simulation period, the more freely mobile N- and C-terminal arms slowly unraveled (Figure 1, Frame 44). A key step in commencing the zipper-like formation of the β-sheets17 is the flipping of Pro332 from trans to cis and then back to a trans state18 (Supplementary Figure S2), which allows the residues His329, His330 and Lys331 to rearrange from internal interactions within the monomer to those important in stabilizing the oligomer (Supplementary Figure S3). Next, the N-terminal (Val306-Lys321) and C-terminal (Gly367-Phe378) arms collapse onto the template, driven by electrostatic interactions and the maintenance of a preferential binding conformation through hydrophobic interactions.19 The zipper-like closure is assisted through the formation of hydrogen bonds between Asp314 and Gln307 in the monomer and Lys370, Lys375 and Thr377 in the oligomer. These hydrogen bonding motifs help bring together the two terminal arms while an intricate network of hydrogen bonds between residue side chains and the backbone is formed. In addition, the hydrophobic residues (Ile308, Tyr310 and Pro312) toward the N-terminus form well-packed hydrophobic interactions with C-terminal residues (Leu376 and Phe378) (Supplementary Figure S4).
Immunochemical analysis of the assembly
We developed a panel of single-chain antibody fragments (scAbs) that recognize linear epitopes spanning most of the core tau unit (Figure 1) and used these to investigate epitope availability via dot immunoblotting over the course of the assembly of tau297-391 (dGAE) in vitro (6). According to the predicted folding pathway, epitopes closer to the tight hairpin loop (residues 337-355 recognized by 1D12) should become occluded earlier in the course of assembly than epitopes located in the N- and C-terminal arms of the core unit. The epitopes recognized by CE2 (residues 319-331) and CA4 (residues 355-367) on either side of the hairpin had limited availability, even beginning at time zero, consistent with partial folding and oligomerization in solution (Figure 2). The low CE2 signal at time zero and subsequent loss of immunoreactivity are consistent with a rapid change in the conformation of the monomer, modeled in the first ten frames of assembly (Supplementary Figure S5). This step was followed by a reduction in the distance in the final assembled configuration to 2 Å, after which the epitope became completely inaccessible. The CA4 epitope is partially occluded at time zero, as it is wrapped around residues 321-343. The epitopes for 1G2 (residues 367-379) and E2E8 (residues 379-391) become occluded later during the course of assembly. The immunoreactivity of the 1D12 epitope, which is the primary monomer capture site, is initially high and then decreases by approximately half over the course of assembly. Modeling suggested that this region remains accessible and suitable for antibody binding prior to assembly and, to a lesser extent, following assembly.
Characterization of the HMT-binding pocket
We hypothesized that the ability of HMT to inhibit assembly might be explained by the stabilization of transient cryptic ligand-binding pockets by ligand-induced conformational funnelling.20 Using tau295-391, we identified a single complex (HMT-protein complex 3) (Figure 3, Supplementary Figure S6 and Tables S2 and S3) in which HMT remained tightly bound, and the RMSD between the start of the MD simulation and after 100 ns was 1.2 Å.
Tau295-391 is stabilized by HMT in a compact, folded conformation that lacks any β-strands and is very different from the extended conformation required for the monomer to align with the PHF core template (Figure 3). Within the 1D12 epitope, HMT interacts with Lys347 and Lys343 through π-cation interactions with the aromatic rings of HMT. Within the 1G2 epitope, HMT interacts with Thr373 via a hydrogen bond between the side chain oxygen and the NH in the central ring of HMT. These interactions with HMT bring the 1D12 and 1G2 binding regions of the tau molecule into closer proximity to each other. In the HMT-bound state, the total water-accessible surface area is reduced by 20%, from approximately 10,223 Å2 to approximately 8,014 Å2. This surface area reduction results in a decrease in both polar (22%) and hydrophobic (21%) surface areas, indicating that the driving forces responsible for this conformation are the formation of intramolecular hydrogen bonds and hydrophobic collapse.21
Pharmacophore modeling of the cryptic druggable pocket and its use in identifying novel inhibitors
The HMT binding site was determined to be 70% druggable,22 i.e., to have a favorable fraction of hydrophobic solvent-accessible surface area. We developed a pharmacophore model based on the key residues to assist with the rational design of novel TAIs. The HMT-binding pocket is predominantly hydrophobic in nature (Phe378, Phe346, Val350, Leu315, Ile354, Ile371). A number of residues can potentially form hydrogen bonds with a molecule bound within this pocket, including Lys347, Thr373, Leu315 and the NH of Glu372. The interactions with Lys347, Lys343 and Thr373 have been noted above. We exploited the shape, hydrogen bonding and lipophilic features observed with HMT to design alternative TAIs in the pharmacophore model.
Using computer-aided drug design approaches, we identified a thiazole core as a suitable heterocyclic replacement for the central ring of HMT. A representative range of candidates was tested in a cell-based tau aggregation screening assay.23 In this assay, 1 µM methylthionine (MT) decreased aggregation by 90.4%. Of the 18 compounds tested, compound 17 decreased the level of tau aggregation in the cell assay by 73.4%, with the data for the other compounds shown in the table (Supplementary Figure S7 and Table 1).
A comparison of the inhibitory activities of the tested thiazole derivatives revealed that interactions with Lys343 via hydrogen bonding and p-cation interactions are important for determining potency. The orientation of the sulfonamide oxygen atoms is key to obtaining the desired hydrogen bond with the backbone carbonyl of Lys343. Substituting a methyl group onto amide 1 to produce amide 6 consistently increases potency. A small alkyl substituent is adequate for binding into the lipophilic pocket toward Phe378. The sulfonamide affects the ability of the amide moiety to form a hydrogen bond with the NH of Thr373. Compound 17 fulfills several of the required binding features, including the sulfonamide oxygen hydrogen bonding to the Glu372 backbone NH, the amide carbonyl forming a hydrogen bond to the Lys347 backbone NH and the pyrazole forming a hydrogen bond to the NH of Thr373 and the carbonyl backbone of Leu315. Additionally, the phenyl substituent on the pyrazole neatly binds to the lipophilic pocket, forming a face-edge p stack with Phe378, and is well positioned to interact with Lys343 through p-cation interactions. Therefore, the pharmacophore model was developed on the basis that HMT binding permits identification of compounds unrelated chemically to HMT that are able to inhibit core tau assembly.
Immunochemical confirmation of TAI activity
As a separate confirmation of activity, an aqueous-phase ELISA was used to measure TAI activity via the binding of aqueous-phase tau297-391 (dGAE) to a solid-phase tau297-390 (dGA) substrate. Immunoreactivity was measured before and after the assembly of dGAE in the presence or absence of either HMT or compound 17. All epitopes were available for antibody binding in the soluble dGAE preparation prior to assembly (Figure 4). Following the assembly of dGAE, the CA4 and CE2 epitopes became almost completely unavailable, whereas the availability of the epitopes for 1G2 and 1D12 was substantially reduced, consistent with the immunoblot analyses (Figure 2). When core tau assembly was performed in the presence of HMT, assembly-dependent occlusion of these epitopes did not occur. The CE2 epitope became available to a greater extent in the HMT-treated sample than in the soluble pre-assembly sample, suggesting reversal of partial oligomerization in solution prior to induction of assembly. The 1G2 epitope remained partially occluded in the HMT assembly, consistent with Thr373 (within the 1G2 epitope), being a critical HMT binding site. Epitope exposure was less complete for compound 17. Although comparable to HMT, as measured by 1G2 immunoreactivity, there was only partial exposure of the CA4 and 1D12 epitopes. Whereas HMT binding prevented the conformational changes in the CE2 epitope responsible for the early loss of immunoreactivity, this was not the case for compound 17, a feature that may account for the decreased potency of this compound.
Although the atomic-resolution structure of the core filaments present in AD and FTD is known,13-15 the molecular mechanisms responsible for the pathological assembly of tau are still not fully understood.24-26 A number of precursors and oligomeric states have been identified for some proteins.27-30 However, isolation and further characterization of these structurally heterogeneous and transient forms present a challenge since it is unclear how they could be isolated and made amenable to unbiased structural and biochemical analysis. MD provides a means of exploring extensive conformational spaces and determining the inherent heterogeneity and stability of specific states of a molecule. This provides a basis for structural predictions that can be tested using immunochemistry, medicinal chemistry and biochemistry. We combined these techniques to investigate both the mode of assembly of the PHF core and the pharmacological inhibition of assembly.
The mode of capture of monomeric tau306-378 to form a core oligomer can be summarized in a series of sequential stages. First, the monomer is anchored to the template through the tight hairpin loop formed by residues 337-355; then, Pro332 undergoes a switch in conformation, which permits unfolding of the monomer and zipping of its N- and C-terminal arms with corresponding segments of the existing assembly. These latter events occur rapidly, with residues 319-331 binding in the direction from the C- to N-terminus and residues 355-367 binding in the N- to C-terminal direction. This folding sequence is associated with progressive occlusion and loss of immunoreactivity of a panel of scAbs recognizing epitopes spanning the monomer. Immunoreactivity with epitopes immediately adjacent to the hairpin loop is lost very early, whereas occlusion of epitopes located nearer to the C-terminus occurs later. The proposal that a short fragment comprising residues 306-311 triggers tau assembly31 is inconsistent with the present analysis. Although these short fragments of tau can assemble spontaneously, our model suggests that the mechanism of assembly entails interactions occurring over a greater distance than can be modeled by short peptides in vitro. PHF assembly can be understood as a process of templated monomer capture, unfolding and residue alignment to produce the characteristic β-sheet structure of the PHF core.
HMT works by stabilizing a conformation that competes with the unraveling of the N- and C-terminal domains of the monomer that are required for oligomer elongation. We identified a single complex in which HMT remained tightly bound to the protein structure. HMT stabilizes one of the conformations available among the ensembles of possible conformations of the core tau unit that cannot be revealed. The driving force responsible for the wrapped-up conformation is the formation of intramolecular hydrogen bonds and hydrophobic collapse. This reduces the water-accessible, polar and hydrophobic surface areas required for alignment and accommodation of the monomer onto a preexisting oligomer. We do not know whether a different class of TAIs might be able to act by stabilizing a different conformation of the core tau unit. We can, however, conclude that HMT acts by stabilizing a conformation that is incompatible with assembly.
The identification of the cryptic binding pocket provides a basis for defining a pharmacophore model that can be used to design novel TAIs. The HMT-binding pocket is predominantly hydrophobic in nature, with some sites suitable for forming hydrogen bonds. A set of 18 compounds unrelated to HMT was synthesized to test whether new inhibitors could be rationally designed to optimize some of the medicinal chemical features required for inhibitory activity. Several of these novel compounds were shown to have activity in a cell-based tau assembly assay23, and one of these was compared with HMT in a new aqueous-phase immunoassay examining aggregation-dependent epitope occlusion. HMT was shown to be effective at preventing the occlusion of all the tested antibody recognition sites spanning most of the PHF core. Compound 17 had weaker inhibitory activity in the cell assay and was also more limited in its ability to prevent epitope occlusion. This finding suggested that there is scope for further medicinal chemistry optimization of such compounds.
The PHF core isolated from AD brain tissues is an extremely stable structure that resists proteolysis.5 At first sight, this intractability might suggest that small-molecule TAIs cannot be designed with the ability to act on an intracellular target in the brain at concentrations that are safe and tolerable. However, HMT has been reported to have clinical pharmacological activity at an estimated steady-state brain concentration of approximately 0.1 µM, which is on the same order as the EC50 value of 0.6 µM observed in a cell assay in which high levels of the core tau unit are expressed.23 This study provides an explanation for the surprising potency of HMT as a TAI. We showed that PHF assembly requires the monomer to be flexible and that HMT acts by stabilizing an endogenous assembly incompetent conformation. The discovery of these mechanisms provides a starting point for the rational design of novel TAIs for the treatment of AD and related tauopathies.
Acknowledgments: We thank Claire Hull for her editorial assistance.
Funding: This work was funded by TauRx Therapeutics Ltd., Singapore, and the Genting Diagnostic Test Centre Sdn Bhd., Kuala Lumpur, Malaysia.
Author contributions:
Methodology: MPM, KEM, YKA, DH, RXS, MS, SJK, SP, MA, RL, LKP, and AJRP
Investigation: MPM, KEM, YKA, DH, RXS, MS, SJK, MA, RL, and LKP
Supervision: MPM, SP, AJRP, JMDS, CRH, LCS, CMW
Writing – original draft: MPM, CRH, LCS, CMW
Writing – review & editing: All the authors
Competing interests: CRH, JMDS, and CMW declare that they are officers in TauRx Therapeutics Ltd. MPM, KEM, DH, MS, SJK, SP, MA, RL, LKP, AJRP, CRH, JMDS, LCS, and CMW are named inventors on patents in the field of tau protein in neurodegenerative disorders.
Data and materials availability: All the data are available in the main text or the supplementary materials.
- Wilcock, G. K. & Esiri, M. M. Plaques, tangles and dementia. A quantitative study. J Neurol Sci 56, 343–356 (1982).
- Arriagada, P. V, Growdon, J. H., Hedley-Whyte, E. T. & Hyman, B. T. Neurofibrillary tangles but not senile plaques parallel duration and severity of Alzheimer’s disease. Neurology 42, 631–639 (1992).
- Smith, R. et al. 18F-AV-1451 tau PET imaging correlates strongly with tau neuropathology in MAPT mutation carriers. Brain 139, 2372–2379 (2016).
- Mukaetova-Ladinska, E. B. et al. Staging of cytoskeletal and β-amyloid changes in human isocortex reveals biphasic synaptic protein response during progression of Alzheimer’s disease. Am J Pathol 157, 623–636 (2000).
- Wischik, C. M. et al. Isolation of a fragment of tau derived from the core of the paired helical filament of Alzheimer disease. Proceedings of the National Academy of Sciences 85, 4506–4510 (1988).
- Al-Hilaly, Y. K. et al. Alzheimer’s disease-like paired helical filament assembly from truncated tau protein is independent of disulfide crosslinking. J Mol Biol 429, 3650–3665 (2017).
- Lövestam, S. et al. Assembly of recombinant tau into filaments identical to those of Alzheimer’s disease and chronic traumatic encephalopathy. eLife 11, e76494 (2022).
- Al‐Hilaly, Y. K. et al. Tau (297‐391) forms filaments that structurally mimic the core of paired helical filaments in Alzheimer’s disease brain. FEBS Lett 594, 944–950 (2020).
- Lutter, L. et al. Structural identification of individual helical amyloid filaments by integration of cryo-electron microscopy-derived maps in comparative morphometric atomic force microscopy image analysis. J Mol Biol 434, 167466 (2022).
- Al-Hilaly, Y. K. et al. Cysteine-independent inhibition of Alzheimer’s disease-like paired helical filament assembly by leuco-methylthioninium (LMT). J Mol Biol 430, 4119–4131 (2018).
- Schelter, B. O. et al. Concentration-dependent activity of hydromethylthionine on cognitive decline and brain atrophy in mild to moderate Alzheimer’s disease. Journal of Alzheimer’s Disease 72, 931–946 (2019).
- Shiells, H. et al. Concentration-dependent activity of hydromethylthionine on clinical decline and brain atrophy in a randomized controlled trial in behavioral variant frontotemporal dementia. Journal of Alzheimer’s Disease 75, 501–519 (2020).
- Fitzpatrick, A. W. P. et al. Cryo-EM structures of tau filaments from Alzheimer’s disease. Nature 547, 185–190 (2017).
- Falcon, B. et al. Structures of filaments from Pick’s disease reveal a novel tau protein fold. Nature 561, 137–140 (2018).
- Falcon, B. et al. Tau filaments from multiple cases of sporadic and inherited Alzheimer’s disease adopt a common fold. Acta Neuropathol 136, 699–708 (2018).
- Shkurti, A. et al. pyPcazip: A PCA-based toolkit for compression and analysis of molecular simulation data. SoftwareX 5, 44–50 (2015).
- Perutz, M. F. Glutamine repeats and inherited neurodegenerative diseases: molecular aspects. Curr Opin Struct Biol 6, 848–858 (1996).
- Brandts, J. F., Halvorson, H. R. & Brennan, M. Consideration of the possibility that the slow step in protein denaturation reactions is due to cis-trans isomerism of proline residues. Biochemistry 14, 4953–4963 (1975).
- Ozgur, B. & Sayar, M. Role of hydrophobic/aromatic residues on the stability of double-wall β-sheet structures formed by a triblock peptide. J Phys Chem B 121, 4115–4128 (2017).
- Carlson, H. A. Protein flexibility and drug design: How to hit a moving target. Curr Opin Chem Biol 6, 447–452 (2002).
- Li, H. Prediction of hydrophobic core residues based on network analysis. in 2009 International Symposium on Intelligent Ubiquitous Computing and Education, IUCE 2009 101–104 (2009). doi:10.1109/IUCE.2009.70.
- Cheng, A. C. et al. Structure-based maximal affinity model predicts small-molecule druggability. Nat Biotechnol 25, 71–75 (2007).
- Harrington, C. R. et al. Cellular models of aggregation-dependent template-directed proteolysis to characterize tau aggregation inhibitors for treatment of Alzheimer disease. Journal of Biological Chemistry 290, 10862–10875 (2015).
- Frankel, R. et al. Autocatalytic amplification of Alzheimer-associated Aβ42 peptide aggregation in human cerebrospinal fluid. Commun Biol 2, 365 (2019).
- Oakley, S. S. et al. Tau Filament Self-Assembly and Structure: Tau as a Therapeutic Target. Front Neurol 11, (2020).
- Tornquist, M. et al. Ultrastructural evidence for self-replication of Alzheimer-associated Aβ42 amyloid along the sides of fibrils. Proc Natl Acad Sci U S A 117, 11265–11273 (2020).
- Stroud, J. C., Liu, C., Teng, P. K. & Eisenberg, D. Toxic fibrillar oligomers of amyloid-β have cross-β structure. Proc Natl Acad Sci U S A 109, 7717–7722 (2012).
- Ciudad, S. et al. Aβ(1-42) tetramer and octamer structures reveal edge conductivity pores as a mechanism for membrane damage. Nat Commun 11, 3014 (2020).
- Fichou, Y. et al. The elusive tau molecular structures: Can we translate the recent breakthroughs into new targets for intervention? Acta Neuropathol Commun 7, 31 (2019).
- Darling, A. L. & Shorter, J. Atomic structures of amyloid-β oligomers illuminate a neurotoxic mechanism. Trends Neurosci 43, 740–743 (2020).
- Von Bergen, M. et al. Assembly of τ protein into Alzheimer paired helical filaments depends on a local sequence motif (306VQIVYK311) forming β structure. Proc Natl Acad Sci U S A 97, 5129–5134 (2000).
Table 1. Activity of compounds in a cell-based assembly assay in which full-length tau is captured by tau295-391 and subsequently truncated. The core thiazole moiety is shown in the middle, the amide substituents, labeled 1-6, are shown above, and the sulfonamide substituents, numbered 7-11, are shown at the bottom of the schema. Activity is expressed as the ratio of truncated tau/full-length tau in treated cells to that in untreated cells in three independent experiments using the assay described in Harrington et al., 2015.23 Activity is expressed as the mean value (+ S.E.M.; n=3). Compounds with values <0.5 were regarded as positive tau aggregation inhibitors, whereas those with a value of 1.0 showed no inhibitory activity. For certain compounds, tau aggregation appeared to be enhanced when the ratio was increased to >2.0.
The authors declare potential competing interests as follows: CRH, JMDS, and CMW declare that they are officers in TauRx Therapeutics Ltd. MPM, KEM, DH, MS, SJK, SP, MA, RL, LKP, AJRP, CRH, JMDS, LCS, and CMW are named inventors on patents in the field of tau protein in neurodegenerative disorders.
- MazanetzNaturesupplementary4APRIL2024.docx
Supplementary Materials Materials and Methods 7 Supplementary Figures S1 to S7 3 Supplementary Tables S1 to S3 Supplementary References