A152T tau allele causes neurodegeneration that can be ameliorated in a zebrafish model by autophagy induction

Abstract

Mutations in the gene encoding tau (MAPT) cause frontotemporal dementia spectrum disorders. A rare tau variant p.A152T was reported as a risk factor for frontotemporal dementia spectrum and Alzheimer's disease in an initial case-control study. Such findings need replication in an independent cohort. We analysed an independent multinational cohort comprising 3100 patients with neurodegenerative disease and 4351 healthy control subjects and found p.A152T associated with significantly higher risk for clinically defined frontotemporal dementia and progressive supranuclear palsy syndrome. To assess the functional and biochemical consequences of this variant, we generated transgenic zebrafish models expressing wild-type or A152T-tau, where A152T caused neurodegeneration and proteasome compromise. Impaired proteasome activity may also enhance accumulation of other proteins associated with this variant. We increased A152T clearance kinetics by both pharmacological and genetic upregulation of autophagy and ameliorated the disease pathology observed in A152T-tau fish. Thus, autophagy-upregulating therapies may be a strategy for the treatment for tauopathies.

Keywords: autophagy; neurodegeneration; proteasome; tauopathy.

Figure 1

MAPT p.A152T carrier frequencies and associated odds ratio for the different disease cohorts. The total number of individuals and p.A152T carriers for each of the disease cohorts and controls is shown in the table (left), with odds ratios and nominal P-values depicted in the forest plot (right). Overall refers to the combined neurological disease patient samples. In the forest plot, squares represent the estimate odds ratio and are drawn proportional to the weight of the sample and lines represent 95% confidence intervals.

Figure 2

Phenotypic characterization of Dendra-tau transgenic zebrafish. (A) Representative images of fish with pan-neuronal expression of WT-tau and A152T-tau. In all clutches of WT-tau offspring, larvae were normal. In contrast, A152T-tau fish showed abnormal phenotypes in ∼50% of each clutch. Differing degrees of abnormal curvature of the dorsal spine was observed in A152T-tau fish (percentages refer to the number of the larvae within each severity range per clutch; observations based on more than 30 individual clutches per transgenic line). Scale bar = 1 mm. (B) Motor neuron morphology was analysed by confocal microscopy on live fish and abnormalities found only in A152T-tau fish including truncation, abnormal pathfinding and aberrant branching (ramifications) (i) WT-tau; (ii–iv) A152T-tau, with normal (ii), moderate (iii) and severe (iv) phenotypes. (v) Schematic view of a single somatic motor neuron (mn) unit from the spinal cord (sc) to the yolk extension (ye) and its ramifications divided into dorsal (arrowhead 1) and medial (arrowhead 2). Arrowhead 3 labels normal axonal length reaching the yolk extension. Panels vi–x correspond to magnified images of i–iv highlighting normal (white arrowheads) or abnormal (black arrowheads) ramifications and length according to axonal scheme represented in v [vi corresponds to WT-tau (i); vii corresponds to normal phenotype A152T-tau (ii); viii corresponds to moderate phenotype A152T-tau (iii); ix and x correspond to severe phenotype A152T-tau (iv)]. Scale bar = 50 µm. See also Supplementary Fig. 2. (C) Quantification of the branching defects observed in motor neurons (MN) according to scheme in B(v) at 3 dpf. Abnormalities at any specified point (arrowheads 1, 2 or 3) were counted as abnormal (five segmental motor neuron units within the yolk sac extension region of the trunk, anterior to the urogenic opening, were counted for n = 18 fish per group; graph represents mean ± SD; two-tailed t-test, ^**P < 0.01 and ^***P < 0.001 versus WT-tau). (D) Escape response defects were observed in A152T-tau but not WT-tau fish at 6 dpf (three independent experiments in triplicate, n = 20/group shown as mean ± standard error; ^***P < 0.001 versus negative siblings by two-tailed t-test). (E) Fish expressing A152T-tau show significantly higher levels of total human tau protein at 3 dpf (mean ± SEM, n = 3 independent clutches in triplicate, ^***P < 0.001 versus WT-tau by two-tailed t-test). (F) Levels of Dendra-tau protein (∼120 kDa) could be equally detected by western blot with either Dendra or Tau5 antibodies (mean ± standard error of four clutches in triplicate, by two-tailed t-test). (G) The higher levels of Dendra-tau together with the morphological and motility defects observed in A152T-tau fish are not the result of higher levels of expression of the transgene (mRNA levels). [G(i)] Quantification of the mRNA expression levels of Dendra and Gal4 by quantitative RT-PCR at 24 hpf (pre-phenotype) shows variability in the expression of Dendra between different clutches of WT-tau fish (three independent clutches labelled 1–3) and A152T-tau fish (three independent clutches labelled 4–6). Analysis was performed on groups containing n = 10 fish. [G(ii)] Phenotypic assessment of larvae at 3.dpf from the same clutches analysed in [G(i)] shows abnormal phenotypes in all clutches of A152T-tau fish regardless of the expression level of Dendra-tau. The similar levels of expression of Dendra in clutches 2 (WT-tau) and 5 (A152T-tau) at 24 hpf resulted in abnormal phenotypes only in A152T-tau expressing fish (sev = severe, mod = moderate, according to morphological phenotype score presented in A). (H) Quantification of the mRNA expression levels of Dendra and Gal4 from 10 individual fish collected from clutches 2 (WT-tau) and 5 (A152T-tau). When siblings from these clutches were analysed as pooled samples, equal Dendra mRNA expression levels were observed at 24 hpf. When mRNA expression levels of Dendra or Gal4 were measured in single WT-tau and A152T-tau fish at 24 hpf (pre-phenotype), no significant differences were observed. [H(i)]. At 3 dpf [H(ii)] WT-tau and A152T-tau individual fish with moderate and severe phenotypes had equivalent levels of Dendra and Gal4 mRNA expression. However, A152T-tau fish which were morphologically normal were found to have significantly lower levels of Dendra mRNA expression [H(ii)].

Figure 3

Phosphorylation state of tau in Dendra-tau transgenic zebrafish. Pathological hyperphosphorylation and conformational changes are hallmarks of tauopathies. (A) Both WT- and A152T-tau expressing larvae showed positive immunostaining for the hyperphosphorylation markers AT270 (residue Thr181), AT8 (residues Ser202/Thr205) and PHF1 (residues Ser396/Ser404) in cryosections, from 24 hpf onwards. [A(i)] Schematic overview, DAPI and Dendra images of transverse sections through the spinal cord used for phosphorylated tau detection at 24 hpf (SC = spinal cord, n = notochord, m = muscle block and Y = yolk sac). Scale bar = 50 μm. [A(ii)] Fluorescent images of Dendra-tau (green) and phospho-tau antibodies (red) show positive single neurons stained for AT270, AT8 and PHF1 within the spinal cord in both WT-tau and A152T-tau fish at 24 hpf. Scale bar = 10 μm. See Supplementary Fig. 5. (B) Western blot for phosphorylation markers AT270, AT8 and PHF1 in whole fish lysates at 3 dpf. The levels of phospho-tau were significantly increased in A152T-tau fish compared to WT-tau fish relative to the loading control, actin (mean ± SEM of 10 independent clutches; two-tailed t-test, ^***P < 0.001 versus WT-tau). (C) The higher phosphorylation levels were also observed in A152T-tau fish relative to total tau (Dendra-tau) levels (mean ± SEM, n = 24 fish/group of 10 independent clutches for PHF1 and eight independent clutches for AT8; two-tailed t-test, ^***P < 0.001 versus WT-tau).

Figure 4

Tau aggregation and cell death in Dendra-tau transgenic zebrafish. (A) Levels of sarkosyl-soluble and insoluble tau reflect accumulation of the insoluble form only in A152T-tau fish at 6 dpf. The level of total tau was analysed by immunoblotting using Tau5 antibody (four independent clutches for WT-tau and A152T-tau). (B) Antibody staining for the conformational marker MC1 in cryosections across the eye of WT-tau and A152T-tau fish. No staining was observed in either WT-tau or A152T-tau fish at 3 dpf (left), whereas only A152T-tau presented positive staining for conformational changes at 6 dpf (right). Scale bar = 100 μm. (C) Western blot for active Caspase 3 (Casp3) (quantified below), indicative of increased cell death in fish expressing the A152T variant (mean ± SEM of nine independent clutches; Student-Newman-Keuls one-way ANOVA, *P < 0.01 versus negative, ^##P < 0.01 versus WT-tau). (D) The increased cell death in A152T-tau fish was confirmed by quantification of TUNEL labelling on transverse sections (mean ± SD; n = 5 fish, from a minimum of five sections; Student-Newman-Keuls one-way ANOVA, ^***P < 0.001 versus negative; ^###P < 0.001 versus WT-tau). See also Supplementary Fig. 4. (E) Morphologically abnormal A152T-tau fish showed increased cell death (quantification of TUNEL-positive nuclei) compared to morphologically normal A152T- or WT-tau fish (mean ± SD; Student-Newman-Keuls one-way ANOVA, ^**P < 0.01 and ^***P < 0.001 versus WT-tau). Representative images in Supplementary Fig. 7.

Figure 5

Tau clearance in vivo and autophagy function. Clearance kinetics of photoconverted Dendra-tau measured in neurons of WT-tau and AT152T-tau fish. Measurement of the intensity of the red Dendra-tau signal over time reflects the clearance or degradation of tau protein. (A) Representative images of photoconverted red Dendra-tau signal comparing a single neuron from WT-tau and A152T-tau fish at three different timepoints: immediately after photoconversion (0 h), 24 h and 48 h after photoconversion. (B) Quantification of red Dendra-tau intensity in photoconverted neurons in the spinal cord of WT-tau and A152T-Tau transgenic fish (representative images shown in Supplementary Fig. 8C). The percentage of residual photoconverted red Dendra-tau was measured over 48 h measured at 12-h intervals. Dendra-tagged A152T-tau clears at a significantly lower rate than WT-tau. (n = 30/group shown as mean ± SD; Student-Newman-Keuls one-way ANOVA, ^**P < 0.01 and ^***P < 0.001 versus WT-tau). (C–F) Western blots for LC3-II, a well-characterized marker of autophagosome number, demonstrate that there are no differences in the levels of this protein between WT-tau and A152T-tau fish either at 24 hpf (pre-phenotype; C and D) or 72 hpf (post-phenotype; E and F). (E and F) Measurements of LC3-II levels in the presence or absence of ammonium chloride provides a method for measuring autophagic flux. No differences were observed between the two transgenic lines at 3 dpf, suggesting that autophagy functions normally in both WT-tau and A152T-tau fish (graph represents mean ± SD of four independent clutches per group for E and F and three for C and D; two-tailed t-test). (G and H) Clearance kinetics of Dendra-tau was measured in the presence or absence of ammonium chloride. Treatment with ammonium chloride blocks autophagic flux and delays the clearance of both WT-tau and A152T-tau to the same extent, indicating that flux occurs at the same rate in these two lines (mean ± SD, n = 62 neurons/group; Student-Newman-Keuls one-way ANOVA, ^**P < 0.01 and ^***P < 0.001 versus untreated group). Note in G and H, the ‘WT-tau + NH₄Cl’ (denoted by black squares and black dashed line) overlaps with the ‘A152T-tau’ line (denoted by grey triangles and grey solid line). The graphs in G and H are presented with a different line in the foreground and background to aid interpretation.

Figure 6

Analysis of proteasome function in Dendra-tau transgenic zebrafish. (A and B) Clearance kinetics of Dendra-tau in the presence or absence of MG132. Treatment with MG132 blocks proteasome function and delays the clearance of WT-tau (A) but had no effect on the clearance of A152T-tau (B) (n = 65 neurons/group; mean ± SD; Student-Newman-Keuls one-way ANOVA, ^**P < 0.01 and ^***P < 0.001 versus untreated group). Note in A, the ‘WT-tau + MG132’ (denoted by black squares and black dashed line) overlaps with the ‘A152T-tau untreated’ line (denoted by grey circles and grey solid line). In B, the ‘A152T-tau untreated’ line (denoted by the black triangles and solid black line) overlaps with the ‘A152T-tau + MG132’ line (denoted by black squares and dashed black line). (C) Proteasome activity in lysates of WT-tau and A152T-tau fish was measured using the synthetic Suc-LLVY-AMC substrate. A152T-tau fish showed significantly reduced chymotrypsin-like activity compared to WT-tau fish (n = 20 fish from six independent clutches per group in triplicate; mean ± SEM; two-tailed t-test: *P < 0.05; ^**P < 0.01 and ^***P < 0.001 versus WT-tau). When treated with MG132, proteasome function is blocked in all genotypes [i.e. non-transgenic fish (−ve), WT-tau and A152T-tau], hence the grey lines overlap. (D and E) Western blots for the α1–7 proteasome subunits showed no difference between WT-tau and A152T-tau fish indicating that the proteasome is structurally normal and equally abundant in both groups (lysates are the same as used for C; all bands are specific and upper band was used for quantification; mean ± SEM n = 6/group; two-tailed t-test). (F and G) Transient expression of UbG76V-GFP (Dantuma et al., 2000) was used to measure proteasome function in vivo. In non-transgenic and WT-tau transgenic fish, the GFP signal is detected at a low level under basal conditions but is increased in the presence of MG132. However, in A152T-tau fish the basal level of GFP is higher than in non-transgenic and WT-tau fish and treatment with MG132 fails to increase the detected levels of GFP (specific band found at 27 kDa, asterisk shows unspecific band in F; mean ± SEM of three clutches per group, n = 20 fish per clutch in duplicate; two-tailed t-test, *P < 0.05 versus untreated group, ^ΦP < 0.05 versus untreated WT-tau group, ^##P < 0.001 versus untreated negative siblings). (H and I) Higher levels of endogenous ubiquitinated proteins were detected in A152T-tau fish compared to WT-tau fish (minimum two independent clutches per group, each analysed in triplicate, mean ± SEM; two-tailed t-test, *P < 0.05 and ^**P < 0.01 versus untreated group, ^ΦΦΦP < 0.001 versus untreated WT-tau and ^###P < 0.001 versus untreated negative siblings).

Figure 7

Modulation of A152T-tau clearance and pathology by upregulation of autophagy. (A and B) Treatment of A152T-tau fish with the autophagy inducers clonidine (A), rapamycin or rilmenidine (B), accelerated clearance kinetics of A152T-tau (mean ± SD of n ≥ 40 neurons/group; Student-Newman-Keuls one-way ANOVA, *^/#P < 0.05; ^**/##P < 0.01 ^***P < 0.001 versus untreated). (C) Treatment with rapamycin (rap), clonidine (clo) or rilmenidine (ril) also ameliorated morphological defects in A152T-tau transgenic fish (n = 6 independent experiments, 20 fish/group, mean ± SD; Student-Newman-Keuls one-way ANOVA, *P < 0.05; ^**P < 0.01 and ^***P < 0.001 versus DMSO). (D) Representative image showing rapamycin treatment reduced the proportion of abnormal A152T-tau fish. Scale bar = 3.5 mm. (E) Quantification of the escape response measured in individual A152T-tau fish 3 dpf, treated with either DMSO, rapamycin, clonidine or rilmenidine. (n = 5 independent experiments in triplicate, 15/group shown as mean ± SD; two-tailed t-test: *P < 0.05, ^**P < 0.01 and ^***P < 0.001 versus DMSO). The treatment with autophagy upregulators improved the escape response deficit in A152T-tau fish. (F and G) Induction of autophagy by the addition of rapamycin or rilmenidine to A152T-tau fish also reduced levels of phosphorylated tau at residues Ser202/Thr205 (AT8) and Ser396/Ser404 (PHF1) relative to total tau levels (n = 3 independent experiments in duplicate, 10/group shown as mean ± SEM; two-tailed t-test: *P < 0.05 and ^***P < 0.001 versus DMSO). (H and I) Treatment with autophagy upregulators results in less cell death. Rilmenidine treatment reduced the levels of active caspase-3 (n = 4 independent experiments in triplicate, 10/group shown as mean ± SEM; two-tailed t-test: *P < 0.05 versus DMSO). (J–N) Injection of an expression vector encoding zebrafish atg5 into A152T-tau fish embryos resulted in over-expression of Atg5 protein at 2 dpf (J and K) (high and low exposure of the same blot presented; mean ± SD, n = 6 independent clutches; two-tailed t-test, *P < 0.05 versus control). (J and L) The increase in Atg5 protein correlated with increase in LC3II, a well-characterized reporter for autophagosome number, demonstrating that autophagy was upregulated in Atg5-injected fish (mean ± SEM, n = 8 independent clutches; two-tailed t-test, ^***P < 0.001 versus control). (M and N) A consistent reduction in the number of offspring with morphological defects was observed in Atg5-injected fish compared to control (uninjected) siblings. The percentage of normal fish changes from 52.43% ± 6.01 to 87.79% ± 4.87 after Atg5 injection and consequently, the proportion of abnormal fish diminishes from 47.57% ± 6.01 to 12.21% ± 4.87 (mean ± SEM of seven independent clutches; two-tailed t-test, ^***P < 0.001 versus control).