Within various structures of tau aggregates, the microtubule-binding domain emerges as a central nucleus governing tau’s assembly and stability through a network of intramolecular and intermolecular interactions, encompassing hydrophobic interactions, hydrogen bonding and salt bridges7,8,9,10,11,12. This domain of tau has been proposed to be intricately associated with the aggregation process of Aβ. Thus, we rationally selected various tau fragments within the microtubule-binding domain (Fig. 1b) to determine their influence on Aβ’s aggregation behavior. We also evaluated the cytotoxicity tendency of Aβ species following incubation with the abovementioned tau fragments under both extracellular and intracellular environments, and investigated their molecular-level interactions. Our work unveils a compelling revelation: tau segments characterized by balanced hydrophobic and hydrophilic properties possess the capacity to profoundly alter the aggregation pathways of Aβ. This modulation occurs through direct interactions and, as a consequence, mitigates the cytotoxicity induced by Aβ. These findings advance our understanding of the complicated interplay between amyloidogenic proteins in AD pathogenesis and provide insights into developing therapeutic strategies targeting protein-protein interactions in amyloidogenesis.
Tau can undergo proteolysis by various types of proteases producing tau fragments that can aggregate into NFTs under pathological conditions. The fibrillar aggregates of tau and Aβ are characterized by the β-sheet formation through hydrophobic and electrostatic interactions (Supplementary Figs. 1 and 2). The microtubule-binding domain of tau carrying hydrophobic and hydrophilic regions is implied to modify Aβ aggregation by disrupting intramolecular and intermolecular interactions among Aβ peptides. The microtubule-binding domain of tau (K18), which is composed of four repeat domains, was calculated for its hydrophobic and hydrophilic characteristics using the hydropathicity analysis (Fig. 1b,c). The result indicates that K18 comprises both hydrophobic and hydrophilic portions, with the highest hydrophobicity in R2 and R3 containing two hexapeptides, PHF6* and PHF6, respectively. Predictions using WALTZ, TANGO and AGGRESCAN verified that the PHF6* and PHF6 regions, which have been previously identified as crucial hydrophobic cores for tau aggregation, play an essential role in the assembly process of tau (Fig. 1c and Supplementary Fig. 3). As expected, among individual tau fragments, the hydrophobic property was exhibited in R2, R3, PHF6* and PHF6, while R1 and R4 did not show this feature. Based on these results, we chose seven tau fragments, including K18, R1-R4, PHF6* and PHF6 (Fig. 1b), to explore our abovementioned hypothesis. Using these selected tau fragments, we investigated their effects on the aggregation and cytotoxicity of Aβ, with the identification of the critical structural motifs of tau and their crucial interactions with Aβ (vide infra).
To determine the impact of tau fragments on the aggregation behavior of Aβ, synthetic Aβ (trifluoroacetic acid form; Supplementary Fig. 4) was used in this study. We monitored the kinetics of amyloid fibrillation by the fluorescent thioflavin-T (ThT) assay that can analyze the formation of β-sheet-rich aggregates (Fig. 2, Extended Data Fig. 1 and Supplementary Table 1). Aβ displayed a typical sigmoidal curve with lag, elongation and plateau phases on incubation, indicative of the nucleation-dependent amyloid fibril formation. Similar sigmoidal increases in the ThT intensity were also observed in the amyloidogenesis of α-synuclein and lysozyme.
Sigmoidal fit analyses revealed the lag time (t) of 0.71 (±0.03) h and the half-time (t) of 0.97 (±0.02) h (Supplementary Table 1). Aβ aggregation was facilitated at concentrations ranging from 1 to 7.5 μM, resulting in decreased t and t values (Supplementary Fig. 5 and Supplementary Table 2). By contrast, no marked change in the aggregation kinetics was observed above 7.5 μM of Aβ. The slope derived from the logarithmic plot of half-time versus initial monomer concentration within the range of 1 to 7.5 μM is roughly -0.3, indicating that secondary nucleation has nearly reached saturation while elongation is approaching saturation. As a consequence, at concentrations above 7.5 μM, the aggregation rate becomes less dependent on monomer concentration. Unlike previous findings, which reported concentration-dependent secondary nucleation at low concentrations of recombinant Aβ in 20 mM sodium phosphate buffer, pH 7.4, 200 μM EDTA and 0.02% sodium azide at 37 °C under quiescent conditions with a slope of -1.2 (ref. ), synthetic Aβ under our experimental conditions (for example, 20 mM HEPES, pH 7.4, 150 mM NaCl, 37 °C and constant agitation) reached rapid saturation of multistep nucleation at its low concentrations. This observation suggests that the collision frequency and interactions among Aβ peptides possibly increase under our experimental conditions, which contributes to early saturation. Note that we cannot rule out the potential influence of trifluoroacetic acid on Aβ aggregation.
The presence of K18, R2 and R3 at varying concentrations prolonged the lag phase of Aβ aggregation in a dose-dependent manner (Fig. 2b,c and Supplementary Table 1), indicating that they can retard Aβ aggregation. The concentration of K18 from 10 to 100 μM noticeably extended t and t values during Aβ aggregation with longer incubation, highlighting a distinct delay in the β-sheet-rich aggregate formation in K18-treated Aβ. Figure 2b shows that the Aβ-to-K18 ratios of 1:1, 1:5 and 1:10 reached saturated levels of β-sheet-rich aggregates after 5, 9 and 18 h, respectively. The rate of forming the ThT-reactive β-sheet-rich structure was inversely correlated with the concentration of K18. Similarly, treatment with R2 increased t and t values in a concentration-dependent manner, extending the lag phase from roughly 10 to 50 min compared to Aβ only. This was accompanied by the reduced ThT fluorescence intensity, suggesting the slower and suppressed β-sheet-rich aggregate production. In contrast, while R3 initially enhanced the t and t values of Aβ aggregation, these values decreased at higher concentrations. At a 1:1 ratio of Aβ to R3, the fluorescence intensity was lower than that of Aβ alone, whereas at a 1:5 ratio it was comparable. The ThT fluorescence intensity varied sequentially when treated with 10 equiv. of R3, implying the modulation of aggregation kinetics without changing the overall number of β-sheet-rich aggregates. In the case of R1, R4, PHF6* and PHF6, no notable effects on the t and t values of Aβ aggregation were observed, although they enhanced the ThT fluorescence intensity by around 19-50% (Extended Data Fig. 1 and Supplementary Table 1). Note that we used unmodified PHF6 for this study, which is distinct from PHF6 with the N- and C-terminal protection previously reported to facilitate Aβ fibrillization.
To further assess the contribution of the hydrophobic regions, PHF6* and PHF6, within K18 to the modulation of Aβ aggregation, we carried out aggregation studies using K18 mutants, including K18ΔPHF6*, K18ΔPHF6 and K18ΔPHF6*ΔPHF6. These K18 mutants were prepared through DNA cloning and protein expression in Escherichia coli BL21 (DE3) RIL cells, followed by purification and characterization (Supplementary Scheme 1 and Supplementary Figs. 6-8). K18ΔPHF6*, K18ΔPHF6 and K18ΔPHF6*ΔPHF6 exhibited inhibitory effects on Aβ aggregation, although to varying degrees (Extended Data Fig. 2). The t and t values for Aβ aggregation increased in proportion to the concentration of K18ΔPHF6*, relative to those of Aβ only. This inhibitory effect led to a delay in producing ThT-reactive β-sheet-rich aggregates. After 1.5 h of incubation, the ThT fluorescence intensity of Aβ only rapidly reached saturation, while it remained low in the samples treated with K18ΔPHF6*. Similar effects were observed in the samples added with K18ΔPHF6. For K18ΔPHF6*ΔPHF6, which lacks both hydrophobic regions, the impact on Aβ aggregation was presented but less potent even at higher concentrations, relative to K18ΔPHF6* and K18ΔPHF6. These findings highlight that the presence of the two hydrophobic regions, PHF6* and PHF6, within K18 enhances the ability to modulate Aβ aggregation. Note that tau fragments themselves could not produce ThT-reactive aggregates (Fig. 2b and Extended Data Figs. 1a and 2a). 1,4-Dithiothreitol (DTT) was added to the samples containing K18, K18 mutants, R2 and R3 to prevent the disulfide bond formation between Cys291 and Cys322 in R2 and R3, respectively. DTT did not affect Aβ aggregation (Supplementary Fig. 9 and Supplementary Table 1).
Together, these results show that K18, R2, R3 and K18 mutants can modulate the nucleation-dependent fibrillization of Aβ, depending on the Aβ-to-tau fragment stoichiometry. R1, R4, PHF6* and PHF6, however, could not noticeably modify Aβ aggregation. The inhibitory effect was also reported when full-length tau was incubated with Aβ (ref. ). In addition, previous studies showed that adding K18 seeds to monomeric Aβ did not significantly affect Aβ aggregation kinetics but reduced the ThT fluorescence intensity. Conversely, Aβ aggregates promoted the aggregation of K18 (ref. ). These overall findings demonstrate that tau segments with both hydrophobic and hydrophilic characteristics are more effective in altering Aβ aggregation, compared to tau fragments primarily with either a hydrophobic or hydrophilic feature.
To investigate the aggregates of Aβ or tau fragments produced on incubation, we first probed the samples using gel electrophoresis with western blotting (gel-western blot) using an anti-Aβ antibody (6E10, ref. ) and anti-tau antibodies (anti-4R for K18 and R2; anti-tau 316-355 (ref. ) for R3). We analyzed the samples from several incubation time points based on the aggregation kinetics associated with the lag, elongation and plateau phases (Fig. 3a,b and Supplementary Figs. 10-13). Aβ aggregates with molecular weights of around 20 and 100 kDa were indicated by the 6E10 antibody after incubating Aβ with 1, 5 and 10 equiv. of K18 for 0.7 and 4 h (Supplementary Fig. 10b). These bands were apparent in the presence of 10 equiv. of K18. After 8 hours of incubation, Aβ aggregates exhibiting the intense smearing ranging from around 4 to 270 kDa were predominantly monitored in the 1:10 Aβ-to-K18 stoichiometry (Fig. 3b and Supplementary Fig. 10b). Such aggregates in the samples treated with 1 and 5 equiv. of K18 were gradually formed during 24 h incubation. We also investigated the effect of Aβ on tau aggregation using an anti-4R antibody. The monomeric form of K18 as a major species, with its dimer as a minor species, was revealed both in the presence and absence of Aβ for 0.7 to 12 h incubation (Supplementary Fig. 10b). After 16 and 24 hours of incubation, K18 aggregates, including tetramers and hexamers, appeared from the samples containing 10 equiv. of K18 with and without Aβ, while the relatively weak smearing above around 100 kDa was only exhibited in the sample containing both Aβ and K18 (Fig. 3b and Supplementary Fig. 10b). Notably, at the 1:1 Aβ-to-K18 ratio, the band corresponding to the monomer of K18 disappeared, implying the interaction with Aβ or its oligomerization. Collectively, these results signify that the aggregates observed by both anti-Aβ and anti-tau antibodies may be a mixture of homogenous and heterogenous assemblies composed of Aβ, K18 or both.
The change in the molecular weight distribution of Aβ on incubation with R2 or R3 was observed by the 6E10 antibody. The gel’s signal intensity at around 9 kDa appeared at a short incubation time (Supplementary Figs. 11b and 12b). In addition, the intense band at around 13 kDa and the weak smearing ranging from around 20 to 270 kDa were shown at longer incubation time points (Fig. 3b and Supplementary Figs. 11b and 12b). In contrast to K18, the aggregation of R2 and R3 themselves was not monitored by anti-4R and anti-tau 316-355 antibodies, respectively (Supplementary Figs. 11b and 12b). Thus, the abovementioned aggregates were formed by the self-assembly of Aβ. No notable impact of R1, R4, PHF6* and PHF6 on the molecular weight distribution of Aβ on incubation was indicated (Supplementary Fig. 13).
The morphological change of Aβ aggregates produced with tau fragments was visualized by transmission electron microscopy (TEM). A mixture of amorphous structures and short and thin fibrils was detected in the samples of Aβ incubated with 10 equiv. of K18, R2 or R3, distinct from amyloid fibrils in the sample of Aβ only (Fig. 3c). The samples containing Aβ with R1, R4, PHF6* or PHF6 displayed amyloid fibrils (Supplementary Fig. 14). In the case of K18, R2 and R3 themselves, spherical or amorphous aggregates were monitored on incubation (Fig. 3c), while R1, R4, PHF6* and PHF6 were not self-assembled under the parameters of our study (Supplementary Fig. 14). In addition, the self-aggregates of K18, R2 and R3 were quantitatively monitored through turbidity and light scattering assays. The turbidity values obtained on incubation of the samples containing K18, R2 or R3 at various concentrations increased by around 27-213%, 22-161% or 72-134%, respectively, compared to their nonincubated forms (Supplementary Fig. 15). This aggregation propensity was further supported by light scattering experiments, although the relatively low increase in scattering intensity suggests the formation of small oligomers rather than mature amyloid fibrils. This limited aggregation is likely due to the presence of DTT under our experimental conditions, which prevents disulfide bond formation in K18, R2 and R3, thereby restricting the generation of larger aggregates. These results indicate that K18, R2 and R3 remain largely soluble and form small aggregates, such as oligomers, under our experimental conditions. Note that the self-assembly of K18, R2 and R3 may still be driven by disulfide bond formation, even with the addition of DTT (t = 10 h, pH 7.5 at 20 °C). Overall, our investigations corroborate that K18, R2 and R3 can modify Aβ aggregation, forming relatively small-sized aggregates. Moreover, we highlight that tau fragments with both hydrophobic and hydrophilic portions can influence Aβ aggregation more noticeably than relatively hydrophilic R1 and R4 and hydrophobic PHF6* and PHF6.
To evaluate the cytotoxicity of Aβ species with tau fragments, we performed two experiments: their administration under (I) extracellular and (II) intracellular conditions (Fig. 3d). For experiment I, the samples with different Aβ-to-tau fragment ratios were added into the growth media of human neuroblastoma SH-SY5Y cells, and cell survival was determined by the 3-(4,5-dimethylthiaol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. The treatment of Aβ incubated with various concentrations of K18 did not recover the cytotoxicity observed by Aβ only (Fig. 3e). This result was not anticipated based on K18’s notable effect on Aβ aggregation (vide supra), but it may be attributed to its toxicity. In the case of R2 and R3, as their concentration increased, the viability of cells added with Aβ was improved by around 15% and 23%, respectively. R2 and R3 themselves exhibited minimal toxicity (around 90% of cell survival). As expected from the aggregation studies, R1, R4, PHF6* and PHF6 could not affect the Aβ-related toxicity (Extended Data Fig. 3). Moving forward, we performed microinjection studies (experiment II shown in Fig. 3d) to probe how tau fragments (K18, R2 and R3, which can modify Aβ aggregation) could affect the toxicity associated with intracellular Aβ. The administration of Aβ treated with 10 equiv. of K18, R2 or R3 inside the cells enhanced cell viability by around 5-11%, compared to that of Aβ only (Fig. 3f). Tau fragments themselves presented around 85% of cell survival under our microinjection conditions. Collectively, these results and observations illustrate that K18, R2 and R3 alter the toxicity triggered by Aβ to different extents.
To better understand how tau fragments affect Aβ aggregation, we investigated their interactions with Aβ by electrospray ionization-mass spectrometry (ESI-MS) (Fig. 4, Extended Data Fig. 4 and Supplementary Figs. 16-18). Monomeric, dimeric and trimeric Aβ with multiple charge states were monitored without tau fragments (Supplementary Fig. 16). These peaks identified in the mass-to-charge ratio (m/z) spectra were deconvoluted into the mass spectra (kDa), and their relative abundance was determined by integrating the peaks in the deconvoluted mass spectra. The relative abundance of monomeric, dimeric and trimeric species from the sample of Aβ only is 81.3 (±0.3), 16.6 (±0.7) and 2.1 (±0.7)%, respectively.
When 1, 5 and 10 equiv. of K18 were incubated with Aβ, we found the hetero-dimer composed of Aβ and K18 with various charge states (Supplementary Fig. 17a). The peak intensity was amplified by increasing the concentration of K18. Note that the presence of Aβ-K18 aggregates with the different Aβ-to-K18 stoichiometry cannot be excluded, as the hetero-assemblies may not be detected due to the limited resolution of the instrument and overlapping peaks. To confirm the composition of the hetero-dimer, we further conducted ESI with tandem MS (ESI-MS) and used collision-induced dissociation energy to fragment the peak of interest (Supplementary Fig. 17b). The ESI-MS analysis of the peak at 1,847.8 m/z indicated the hetero-dimer composed of +2-charged Aβ and +8-charged K18. In addition, the relative abundance of Aβ unbound and bound to K18 from the deconvoluted mass peaks was calculated to investigate whether the degree of K18 binding to monomeric Aβ varies with the concentration of K18. The level of the Aβ-K18 complex was found to be 18.1 (±6.1)% at the 1:1 Aβ-to-K18 ratio and increase to 44.4 ( ± 13.7)% with 10 equiv. of K18 (Fig. 4b and Supplementary Table 3). We also determined the saturation coefficient (S) using the equation described in the experimental section, which represents the binding equilibrium and analyzes the degree of binding. The S value at the 1:10 Aβ-to-K18 stoichiometry was around 2.4 or 1.3 times higher than that at the 1:1 or 1:5 ratio, respectively, indicating that the hetero-dimer formation is favored with supra-stoichiometric K18 (Supplementary Table 4). This finding suggests that the binding between Aβ and K18 is influenced by their stoichiometry, and the 1:10 ratio presents remarkable hetero-dimer formation.
Multiple heterogeneous complexes were observed in the sample containing Aβ and R2, such as Aβ-R2, Aβ-(R2), (Aβ)-R2 and (Aβ)-(R2) (Fig. 4a and Supplementary Fig. 17). Heterogeneous trimers and larger assemblies, such as Aβ-(R2), (Aβ)-R2 and (Aβ)-(R2), could be constructed through the interactions between Aβ and R2 monomers, their individual dimers or both. For example, the (Aβ)-(R2) complexation can occur in four distinct manners, including two Aβ monomers with two R2 monomers, one Aβ dimer with one R2 dimer, two Aβ monomers with one R2 dimer and one Aβ dimer with two R2 monomers. Note that distinguishing whether two Aβ monomers or one Aβ dimer are involved in heterogeneous assemblies composed of Aβ and R2 is challenging, as is for R2, since the molecular weights and m/z ratios of their two monomers and one dimer are identical under our ESI-MS conditions. Increasing the concentration of R2 enhanced its overall binding toward 1 and 2 equiv. of Aβ, with the highest abundance of the Aβ-R2 and (Aβ)-(R2) complexes (Fig. 4b and Supplementary Table 3). The S values between R2 and one Aβ monomer or between R2 and two Aβ monomers or one Aβ dimer increased proportionally with the concentration of R2, which was similarly observed with K18. R2 exhibited a binding preference for two Aβ monomers or one Aβ dimer over one Aβ monomer under the same stoichiometry conditions. For example, at the 1:10 Aβ-to-R2 ratio, the S value between R2 and two Aβ peptides was around 1.7 times higher than that for monomeric Aβ (Supplementary Table 4). Like R2, R3 formed various adducts with one or two Aβ peptides and displayed a preference for binding to 2 equiv. of Aβ, as supported by a higher S value observed for two Aβ monomers or one Aβ dimer compared to one Aβ monomer under Aβ-to-R3 stoichiometric conditions (Fig. 4b, Supplementary Fig. 17 and Supplementary Table 4). These results show that R2 and R3 preferentially bind to two Aβ monomers or one Aβ dimer over a single Aβ monomer.
Various heterogeneous complexes were also detected in the samples of Aβ containing different concentrations of R4 (Extended Data Fig. 4, Supplementary Fig. 18 and Supplementary Tables 3 and 4). Different from R2 and R3, which displayed a notable preference for 2 equiv. of Aβ over 1 equiv. of Aβ, R4 demonstrated a relatively greater affinity for one Aβ monomer with the S value ranging from 0.25 (±0.01) to 0.46 (±0.03) than that for two Aβ peptides (from 0.12 (±0.01) to 0.32 (±0.01)). In the case of a hexapeptide, when 1 equiv. of PHF6* was treated with Aβ, two forms, that is, Aβ-PHF6* and (Aβ)-PHF6*, were generated, while the Aβ-(PHF6*) complex also emerged at the 1:5 and 1:10 Aβ-to-PHF6* ratios. Compared to K18, R2, R3 and R4, all S values of PHF6* to one and two Aβ peptides were relatively low. These MS investigations reveal that R2 and R3 show a greater tendency to interact with two Aβ monomers or one Aβ dimer over one Aβ monomer, in contrast to R4 and PHF6*. Our findings corroborate that K18, R2 and R3 with hydrophobic and hydrophilic properties modify Aβ aggregation through their direct interactions with Aβ showing different tau fragment-to-Aβ ratios.
The detailed binding events between tau fragments and Aβ in solution were analyzed using isothermal titration calorimetry (ITC) (Supplementary Fig. 19). To prevent the undesired self-aggregation of Aβ that could interfere with calorimetric analyses, we conducted ITC experiments under salt-free buffer conditions to minimize this interference. Titration of K18 into the solution of Aβ resulted in a spontaneous exergonic reaction with the negative free energy change (ΔG), characterized by thermodynamically favorable negative enthalpy (ΔH) and positive entropy changes (ΔS) (ΔG = - 25.4 (±0.3) kJ mol; ΔH = - 4.7 (±1.2) kJ mol; TΔS = 20.7 (±1.5) kJ mol) (Supplementary Fig. 19). The positive entropy change is likely attributable to the release of water molecules from Aβ on interaction with K18, indicating a central role of hydrophobic interactions. The Aβ-binding affinity of K18 was evaluated by the dissociation constant (K) of 20.3 ( ± 2.9) μM. Furthermore, the molar binding ratio (n) of K18 to Aβ, which is 0.4 (±0.1), suggests the potential formation of hetero-assemblies involving one K18 with two or three Aβ peptides. The feasibility of these hetero-adducts validates that K18 and Aβ can build higher-order hetero-aggregates, in addition to their hetero-dimer found in our ESI-MS data despite experimental limitations.
R2, R3, R4 and PHF6* showed negative ΔG values from -25.9 (±0.4) to -22.8 (±1.7) kJ mol on titration, similar to K18, signifying their spontaneous association with Aβ (Supplementary Fig. 19a). Among them, repeat domain fragments, namely, R2, R3 and R4, exhibited thermodynamically unfavorable endothermic binding isotherms, while PHF6* indicated a weak exothermic reaction (Supplementary Fig. 19b). R2, R3, R4 and PHF6* engaged in hydrophobic interactions with Aβ, leading to positive ΔS, but this entropic contribution must outweigh the enthalpic effect for the repeat domains to guide a spontaneous reaction. The K values of R2, R3 and R4 to Aβ ranged from 17.6 (±2.8) to 34.8 (±8.3) μM, whereas PHF6* displayed the relatively much weaker binding affinity with the K value of 63.1 (±45.0) μM under our experimental conditions. The reaction stoichiometric values of 0.5 (±0.1) and 0.5 (±0.2) for R2 and R3, respectively, indicated that they could interact with two Aβ peptides. Conversely, R4 primarily formed a hetero-dimer at the 1:1 Aβ-to-R4 stoichiometry, while PHF6* formed a hetero-dimer or hetero-trimer with 1:1 or 1:2 Aβ-to-PHF6* ratios, respectively. These results in solution complement our ESI-MS data in the gas phase, signifying that R2 and R3 preferentially assemble hetero-adducts with 2 equiv. of Aβ rather than 1 equiv. of Aβ, while R4 and PHF6* predominantly build a hetero-dimer with one Aβ monomer. Therefore, tau fragments displayed relatively weak binding affinities to Aβ within the micromolar range; however, K18, R2 and R3, characterized by a balanced combination of hydrophobic and hydrophilic features, exhibited varying stoichiometric interactions with more than 2 equiv. of Aβ, unlike R4 and PHF6*. This indicates that tau fragments form diverse hetero-complexes with Aβ depending on their physicochemical properties. Moreover, our findings demonstrate that interactions of tau fragments with two or more Aβ monomers can noticeably regulate Aβ aggregation behavior, rather than simply engaging with one Aβ monomer.
The molecular-level interactions of K18, R2 and R3 with Aβ were further analyzed by two-dimensional H-N selective optimized flip angle short-transient heteronuclear multiple quantum coherence nuclear magnetic resonance (2D H-N SOFAST-HMQC NMR) spectroscopy. Substantial chemical shift perturbations (CSPs) were observed at Asp7, Val12, Gln15, Lys16, Phe20, Val24, Lys28, Ile31 and Val36 of N-labeled Aβ by treatment with a supra-equimolar concentration of K18 (Fig. 5). Most of these amino acid residues exist in Aβ’s hydrophobic self-recognition and C-terminal regions (Fig. 5a) that are critical for its aggregation. In addition, the peak intensities of all amino acid residues were reduced, indicating that the addition of K18 could produce NMR-invisible Aβ aggregates (Fig. 5b). Similar to K18, the alteration in the CSPs of N-labeled Aβ was observed in the N-terminal, self-recognition and C-terminal regions following the treatment with R2 or R3; however, the affected amino acid residues and their quantities were different. These results validate that K18, R2 and R3 can mostly interact with the hydrophobic self-recognition and C-terminal regions of Aβ, which may direct their ability to modify Aβ aggregation.
To further identify the amino acid residues of K18 that can be affected through the interactions with Aβ, we performed 2D H-N band-selective excitation short-transient transverse relaxation optimized spectroscopy (BEST-TROSY) NMR experiments of N-labeled K18 encompassing the 244-372 residues. N-labeled K18 was obtained from the E. coli Rosetta (DE3) strain by subcloning the K18 gene into the pET-15b vector and incubating it in N-supplemented media. The Thr263, Glu264, Lys281, Cys291, Lys294, Val300, Val339, Lys353, Val363 and Glu372 residues located in the R1, R2 and R4 domains of tau exhibited pronounced CSPs on incubation with a supra-equimolar concentration of Aβ (Fig. 6). When Aβ was added to N-labeled K18, significant CSPs were not detected in two hexapeptide regions, PHF6* and PHF6. Note that, unfortunately, several peaks corresponding to amino acid residues in the R3 region, including Pro312, Gly323, Ser324, Asn327, His329, His330, Lys331, Pro332 and Gly335, could not be resolved; thus, the interaction between amino acid residues associated with R3 and Aβ could be inconclusive. Overall, our 2D NMR studies suggest that tau fragments affect the N-terminal, self-recognition and C-terminal regions of Aβ to different extents. The sites, involved in Aβ self-assembly, are directly and indirectly affected by K18, R2 and R3, which supports their impact on Aβ aggregation.
