Scientific Papers

Human tauopathy strains defined by phosphorylation in R1-R2 repeat domains of tau | Acta Neuropathologica Communications


Detergent insoluble AD-tau and PSP-tau seeds induce P301L tau to form aggregates in HEK293T cells

We isolated detergent-insoluble tau seeds using a serial fractionation protocol from brains donated by AD, PSP and age-matched cognitively unimpaired individuals (Additional file 1: Fig. S1a-b, Table S1). Because previous data has shown heterogeneity in among patients [11], we used cortical tau from 6 AD patients and lentiform nuclear tau from 6 PSP patients in our study (Table S1). We confirmed the presence of tau in AD-tau and PSP-tau seed fractions using immunoblotting (Additional file 1: Fig. S1c, g) and ELISA (Additional file 1: Fig. S1d, h). We used a HEK293T cell-based model, which has been used reproducibly in our labs in previous studies [18,19,20,21] to examine the capability of these tau seeds to template 0N/4R WT human tau or 0N/4R P301L tau. We found that detergent-insoluble seeds derived from AD (AD-tau) and PSP (PSP-tau) patients could induce formation of AT8-positive detergent-insoluble tau in cells expressing 0N/4R P301L tau (Additional file 1: Fig. S1e, i). Neither AD-tau nor PSP-tau seeds induced detergent-insoluble tau in cells expressing 0N/4R WT tau or Green Fluorescent Protein (GFP) (Additional file 1: Fig. S1f, j). We next performed an optimization step for AD-tau and PSP-tau seeds in this assay and found that as little as 0.0625 µg of S3 fraction (AD-tau) and 0.312 µg of S3 fraction (PSP-tau) was sufficient to induce AT8-positive detergent-insoluble tau (Additional file 1: Fig. S2a-b) and that transfection with 0.4 µg of P301L tau DNA was optimal to observe seeding (Additional file 1: Fig. S2c-d) with no cell death (Additional file 1: Fig. S2e). The result of these initial studies produced a reliable model system to assess tau seeding using donor-derived tau seeds.

In ADRDs, tau undergoes phosphorylation simultaneously at multiple sites [8] and recent phospho-proteomic analysis suggests that progressive phosphorylation on specific sites precede misfolding and aggregation into NFT [14]. From this study, we identified 6 phosphorylation sites that are modified in pre-symptomatic normal individuals, 7 additional sites that are phosphorylated in a subset of individuals that are pre-symptomatic or definite AD, and 6 more sites that are phosphorylated in definite AD cases (total 19 sites, numbered according to 2N/4R tau: Additional file 1: Fig.S3a). This led us to hypothesize that ablating phosphorylation on these sites could prevent tau from being misfolded in the initial seeding phase. To examine whether phosphorylation on these sites influence self-templating properties of misfolded tau, we introduced Ala substitutions on all 19 sites of phosphorylation spanning across the Proline rich region (PRR), microtubule binding region (MTBR) and C terminal domain, referred to as Complete Phospho-Null (CPN) construct (Additional file 1: Figs. S3a). We also divided these 19 sites into 5 spatially contiguous phospho-substitution sites spanning across human P301L 0N/4R tau sequence (Additional file 1: Fig. S3a). Numbering these 1 through 5, we mutated the Ser/Thr sites on these clusters to phospho-nullifying amino acid Alanine (Ala). The rationale was based on earlier precedence of using such combined mutations to examine microtubule binding, aggregation and seeding properties of tau [6, 20, 22]. We established that the relative levels of tau expression are similar between the different Phospho-Null Variants 1–5 (Additional file 1: Fig. S3a).

To test our hypothesis that ablating phosphorylation would reduce seeding, we first tested whether AD-tau seeds or PSP-tau seeds would template the CPN construct transfected into HEK293T cells, with P301L tau as the positive control and seeds from healthy patients as negative controls (Additional file 1: Fig. S3b-o). We observed that the CPN construct produced insoluble tau when seeded by six different AD-tau (Additional file 1: Fig. S3b-e) and six PSP-tau seed preparations (Additional file 1: Fig. S3i-l). Homogenates from healthy controls did not induce aggregation of P301L tau (Additional file 1: Fig S3c; Patient ID # 1 and 2). As a positive control, unmodified P301L tau was readily seeded by AD-tau seeds and PSP-tau seeds (Additional file 1: Fig S3c, j; #B). Only seeded P301L tau was reactive to AT8 because one of the Ala mutations destroyed the AT8 epitope on the CPN construct (Additional file 1: Fig. S3e, l). We also confirmed that when we tested on individual Phospho-Null Variants 1 through 5, both AD patient-derived tau seeds (Additional file 1: Fig. S3f-h) and PSP patient-derived tau seeds (Additional file 1: Fig. S3m-o) showed equivalent tau seeding efficiency compared to the other variants. These somewhat surprising findings indicate that biochemically mimicking phosphorylation ablation of tau at these 19 early sites does not modulate seeded tau aggregation in cell models.

Phospho-Plus substitutions in the R1-R2 repeat domains alter seeding efficiency of AD-tau

To complement the previous part of this study, we further designed Phospho-Plus tau constructs by mutating Ser/Thr sites to the phospho-mimicking Glutamate (Glu) amino acid. We created the Complete Phospho-Plus (CPP) construct with 19 substitutions as well as individual Variants 1–5, with Glu substitutions as indicated (Fig. 1a). Surprisingly, when we used the CPP construct, we found that none of the AD-tau seeds from 6 different patients induced tau seeding (Fig. 1b–c), which was confirmed by an absence of AT8 signal (Fig. 1d–e).

Fig. 1
figure 1

AD-tau seeds show differential effects on Phospho-Plus tau variants. a. Schematic depiction of Phospho-Plus Ser/Thr → Glu tau variants generated on human 0N/4R P301L mutant tau. All numbers correspond to 2N/4R tau. R1-R4 depict microtubule-binding repeat regions. b–e. Representative immunoblot of the Complete Phospho-Plus (CPP) construct with all 19 sites mutated to Glu and seeded with Patients A-F or two healthy controls (1, 2) shown. P301L tau seeded with Patient B or left unseeded are also shown. Total tau (t-tau) and AT8-tau (p-tau) shown from detergent-soluble (b, d) or insoluble (c, e) cell lysates. f–u. AD-tau ( +) or unseeded (-) HEK293T cells transfected with different tau variants were fractionated into detergent-soluble and detergent-insoluble lysates and probed for total tau (fg, t-tau) or AT8 (n–o, p-tau). Quantitation of % aggregation (insoluble/[soluble + insoluble]*100) for each AD-tau variant are shown (h-m). AT8 levels in detergent insoluble lysates has been normalized and shown (p-u). GAPDH is the loading control for detergent-soluble fraction. Broken lines (hm) depict statistical tests done separately within groups of seeded or non-seeded tau variants. Numbers 1–5 denote the tau Variants; ‘P’, P301L tau; ‘G’, GFP; ‘ND’, not detected. Representative blots for all other patients not depicted here (A, C, D, E and F) are shown in Figure S4.

To refine which phosphorylation sites may be responsible for our observed outcome, we used the five Phospho-Plus Variants containing phospho site Glu substitutions in spatially contiguous clusters (Fig. 1; Additional file 1: Additional file 1: Fig. S4). When we performed the AD-tau seeding assay with these variants, we found that Variants 4 and 5 were less susceptible to seeding than variants 1–3 (Fig. 1f–h). Variant 4 (containing Phospho-Plus mutations in the R1-R2 region) showed lower seeded tau levels in the detergent-insoluble fraction compared to seeded P301L tau when seeded with AD Patient B (Fig. 1h; p < 0.01). Indeed, using multiple donor AD-tau and executing three replicates per patient, Variant 4 appears to be consistently less prone to seeded aggregation (Fig. 1i–m; additional experimental and biological replication blots in Additional file 1: Fig. S4), which indicates a robust loss of self-templating ability of Variant 4 across patients. Using Thioflavin S staining, we confirmed that the levels of tau ‘amyloid’ was indeed reduced in Variant 4 seeded with AD-tau relative to Variant 5 and parent P301L tau seeded with AD-tau from Patient B (Additional file 1: Fig. S5). A similar pattern emerged when soluble and insoluble fractions of AD-tau seeded HEK293T cells were probed with AT8 (Fig. 1n–p; additional experimental and biological replicates: Fig. 1q–u; Additional file 1: Fig. S4). When we normalized AT8 levels in insoluble fraction to total insoluble tau, we confirmed that the seeded Variant 4 selectively showed no detectable phosphorylation on Ser202/Thr205 epitope (Fig. 1p–u, Additional file 1: Fig. S4). Notably, Variant 1 did not show any AT8 reactivity as this isoform contains substitution on the AT8 epitope (Ser202/Thr205).

The levels of expression for the Phospho-Plus tau variants was comparable with each other (Additional file 1: Fig. S4h). Notably, seeding of cluster Variant 5 with AD-tau was variable. Compared to parent P301L tau, seeded Variant 5 resulted in reduced levels of total tau in the detergent-insoluble cellular fractions (Fig. 1f–m; Patient C and Patient F: *p < 0.05). On the other hand, seeds from Patient A showed robust aggregation and seeds from Patients B, D and E showed reduced aggregation (Fig. 1i–m). This would be consistent with previous observations that tau seeding activity displays heterogeneity between different AD donors and this could be related to the phosphorylation profile of seed-competent tau [11, 23]. Interestingly, we noted that Variant 5 consistently showed higher AT8 levels comparable to Variants 2 and 3, as well as parent P301L tau (Fig. 1p–u; p < 0.01 for Patient D), in spite of its overall lower aggregation propensity detected by total tau antibody. This would suggest some dissociation between aggregation propensity and AT8 site phosphorylation when tau is phosphorylated in the C-terminus.

Phospho-Plus PTM isoforms do not significantly alter seeding efficiency of PSP-tau seeds

To examine if the strain differences between AD-tau and PSP-tau [24, 25] are reflected in the self-templating properties of tau, we next examined if these same Phospho-Plus (Ser/Thr → Glu) variants alter the propensity of PSP-tau in forming tau aggregates (Fig. 2). First, we examined the CPP construct with all 19 sites substituted for Glu. Consistent with our data on AD-tau seeding, the CPP construct was not susceptible to PSP-tau seeds derived from 6 different patients as detected by total tau antibody and AT8 antibody (Fig. 2b–e). We then examined how the 5 individual Phospho-Plus Variants respond to PSP-tau seeds from different patients. First, using soluble fractions of seeded and unseeded HEK293T cells, we found little difference in soluble levels of total tau between parent P301L tau and the Phospho-Plus variants, except for Variant 5 (Fig. 2f–m; Additional file 1: Fig. S6: additional blots from other patients). Additionally, analysis of the insoluble cell fraction showed that Variants 1, 2, 3 and 4 showed equivalent levels of total seeded tau using PSP-tau from different patients (Fig. 2f–m: Additional file 1: Fig. S6 for additional patients and replicate blots). Some heterogeneity was noted in this aspect, especially for PSP Patient C seeds which showed lower insoluble tau when seeding Variant 4 (Fig. 2j; p < 0.05). Variant 5 consistently showed reduced seeding efficiency with PSP-tau derived from Patient B (p  < 0.05), Patient C (p  < 0.01), Patient D (p  < 0.05) and Patient E (p  < 0.05), with trend for Patients A and F (Fig. 2h-m). AT8 analysis showed that Variants 2–4 showed comparable soluble and insoluble tau levels, relative to P301L tau (Fig. 2n–u; Additional file 1: Fig. S6). Compared with the other variants, Variant 5 showed higher AT8 burden in spite of lower amounts of total tau in the insoluble fraction, though the data did not reach statistical significance (Fig. 2p–u). This observation is consistent with the data from AD-tau seeding (Fig. 1). Overall, notwithstanding some heterogeneity between patients, our data shows that Phospho-Plus Variant 4 can be seeded by PSP-tau, suggesting that the phosphorylation sites contained in this variant potentially differentiate PSP and AD seeding susceptibility, with phosphorylation mimics in this construct inhibiting AD-tau seeding consistently.

Fig. 2
figure 2

Analysis of seeding efficiency of PSP-tau on Phospho-Plus tau variants. a. Schematic depiction of Phospho-Plus Ser/Thr → Glu tau variants generated on human 0N/4R P301L mutant tau. All numbers correspond to 2N/4R tau. R1-R4 depict microtubule-binding repeat regions. b–e. Representative immunoblot of the Complete Phospho-Plus (CPP) construct with all 19 sites mutated to Glu and seeded with Patients A-F shown. P301L tau seeded with Patient F or left unseeded are also shown. Total tau (t-tau) and AT8-tau (p-tau) shown from detergent-soluble (b, d) or insoluble (c, e) cell lysates. f–u. PSP-tau ( +) or unseeded (−) HEK293T cells transfected with different tau variants were fractionated into detergent-soluble and detergent-insoluble lysates and probed for total tau (f-g, t-tau) or AT8 (n–o, p-tau). Additional replicate blots are shown in Fig. S6. Quantitation of % aggregation (insoluble/[soluble + insoluble]*100) for each variant are shown (h-m). AT8 levels in detergent insoluble lysates has been normalized and shown (p-u). GAPDH is the loading control for detergent-soluble fraction. Broken lines (hm) depict statistical tests done separately within groups of seeded or non-seeded tau variants. Numbers 1–5 denote the tau Variants; ‘P’, P301L tau; ‘G’, GFP; ‘ND’, not detected. Representative blots for all other patients not depicted here (A, C, D, E and F) are shown in Additional file 1: Fig. S6. Relative molecular masses (kDa) are indicated on the left of each blot. N = 3 experimental replicates. 1-way ANOVA with Dunnett’s test; ****P < 0.0001; **p < 0.01, *p < 0.05

Cooperative phosphorylation profiles of R1-R2 Phospho-Plus variants following tau seeding

We wanted to investigate how the phospho-PTM profile on host tau influences the phosphorylation patterns of tau aggregates resulting from seeding [26]. Looking at seeded Phospho-Plus Variant 4 with total tau antibody, we could reproduce our observations that AD-tau does seed Variant 4 less robustly than PSP-tau seeds (Additional file 1: Fig. S7a-d). Using antibodies against phospho-tau epitopes that are distributed in the PRR and C terminal region which were previously identified to be over-represented in AD (Fig. 3a; AT270, AT100, AT180, PHF1, pSer422 and AT8), we found that PSP-tau seeded Variant 4 showed robust phosphorylation on these sites and was identical to P301L tau (Fig. 3b). However, AD-tau seeded Variant 4 showed minimal phosphorylation on these epitopes, especially with no detectable signals for AT8, AT100 and pSer422. Of note, TauC3, a truncated form of tau generated by caspase-3 cleavage at D421 and a precursor of bioactive seeds [27], was present only in PSP-tau seeded Variant 4 but not in AD-tau seeded Variant 4 (Fig. 3b). Because the levels of phosphorylation on AT270, AT8, AT100, AT180, and pSer422 epitopes was equivalent between Variant 5 and P301L tau, this indicates that most of the seeded Variant 5 is highly phosphorylated in spite of total tau levels being lower. This finding suggests that pre-existing phosphorylation on these C terminal epitopes could preferentially drive the emergence of AD-typical phosphorylation pattern following seeding. Notably, we found little to no PHF1 in seeded Variant 5, which is somewhat expected as PHF1 epitope (Ser396/Ser404) overlaps the Phospho-Plus substitutions in Variant 5. We also noted that both AD-tau and PSP-tau resulted in equivalent levels of cleaved TauC3 in seeded Variant 5, but not in seeded Variant 4 seeded with AD-tau (Fig. 3b).

Fig. 3
figure 3

Phosphorylation profile of MTBR-associated tau variants following seeding. a. Schematic depiction of Phospho-Plus Ser/Thr → Glu tau variants generated on human 0N/4R P301L backbone and antibody epitopes used to detect tau phosphorylation. b. Detergent-insoluble fraction of AD-tau ( +), PSP-tau ( +) or unseeded (–) HEK293T cells transfected with different tau variants and GFP were probed for different epitopes. Relative molecular masses are indicated on the left of each blot. Blots showing technical replicate and detergent-soluble fractions are presented in Additional file 1: Fig. S7 a-d

Phospho-PTMs in tau result in altered structure that modify normal function of tau (such as microtubule binding) or add pathological attributes (increased propensity to aggregate) [1, 28]. Because we observed an interesting dichotomy between levels of total tau and phosphorylated tau in Variant 5, we wondered whether Variant 5 was inherently more prone to show reduced microtubule affinity. To assess tau how these Phospho-Plus mutations lead loss of microtubule binding, we performed microtubule binding assay in the presence and absence of paclitaxel, that stabilizes microtubule-tau interaction (Fig. S7i). Both Variant 4 and Variant 5 tau showed lower propensity to bind microtubules, compared to parent P301L tau (Additional file 1: Fig. S7e-f). Additionally, we also confirmed that, in the presence of paclitaxel, both Variant 4 (p < 0.001) and Variant 5 (p < 0.0001) showed lower microtubule binding affinity respectively compared to P301L tau (Additional File 1: Fig. S7g-h).

Ser305 within MTBR-R2 domain is a specific regulator of tau seeding

Having identified that Phospho-Plus substitutions on S262/T263/S289/S305 prevented the ability of AD-tau from seeding P301L tau, we wanted to determine if individual phospho-mimetic epitopes within this variant region could specifically drive this differential seeding outcome. We introduced the phospho-mimetic S305E substitution on P301L tau and we tested the efficiency of AD-tau and PSP-tau in seeding P301L and P301L/S305E tau (Fig. 4). We observed that both AD-tau and PSP-tau could successfully seed P301L tau (Fig. 4a, b; % aggregation quantified in 4c). However, only PSP-tau, but not AD-tau, was able to seed P301L/S305E tau (Fig. 4a, b). This suggests that S305 phosphorylation could be a discriminatory epitope in soluble tau that promotes templating by PSP-tau and is inhibitory for AD-tau templating.

Fig. 4
figure 4

Phospho-Plus mutation on Ser305 abrogate seeding by AD-tau. ac. AD-tau ( +), PSP-tau ( +) or unseeded (–) HEK293T cells transfected with P301L, P301L/S305E tau or GFP were fractionated into detergent-soluble and detergent-insoluble lysates and probed for total tau (t-tau). Quantitation of % aggregation (insoluble/[soluble + insoluble]*100) for each variant are shown (c). 2-way ANOVA with Bonferroni’s test. ***p < 0.001; **** p  < 0.0001



Source link