Targeted degradation of aberrant tau in frontotemporal dementia patient-derived neuronal cell models

Abstract

Tauopathies are neurodegenerative diseases characterized by aberrant forms of tau protein accumulation leading to neuronal death in focal brain areas. Positron emission tomography (PET) tracers that bind to pathological tau are used in diagnosis, but there are no current therapies to eliminate these tau species. We employed targeted protein degradation technology to convert a tau PET-probe into a functional degrader of pathogenic tau. The hetero-bifunctional molecule QC-01-175 was designed to engage both tau and Cereblon (CRBN), a substrate-receptor for the E3-ubiquitin ligase CRL4^CRBN, to trigger tau ubiquitination and proteasomal degradation. QC-01-175 effected clearance of tau in frontotemporal dementia (FTD) patient-derived neuronal cell models, with minimal effect on tau from neurons of healthy controls, indicating specificity for disease-relevant forms. QC-01-175 also rescued stress vulnerability in FTD neurons, phenocopying CRISPR-mediated MAPT-knockout. This work demonstrates that aberrant tau in FTD patient-derived neurons is amenable to targeted degradation, representing an important advance for therapeutics.

Keywords: biochemistry; chemical biology; degrader; frontotemporal dementia; human; human neuronal cell models; neuroscience; tau.

Figure 1.. Design and working model for a new hetero-bifunctional tau degrader.

(A) A degrader molecule was designed to preferentially recognize disease-associated forms of tau (Module 1), and simultaneously engage with CRBN in the CRL4^CRBN E3 ubiquitin ligase complex (Module 3). (B) Degrader-mediated association of tau with CRL4^CRBN and formation of a ternary complex is predicted to mediate tau ubiquitination and degradation by the proteasome. (C) QC-01–175 (I) was synthesized based on the T807 core scaffold for tau recognition, a thalidomide analog E3 ligand (pomalidomide) for CRBN engagement, and a linker designed for maximum target clearance efficiency (II). QC-03–075 is the inactive analog (III). (D) BLI Streptavidin (SA) biosensor assay to measure recombinant tau protein affinity to small molecules (e.g. T807, QC-01–175). (E) BLI results indicate that, in vitro, QC-01–175 binds to variant forms of tau within the same order of magnitude as T807. Bars represent mean K_D (μM)±SD (n ≥ 3). Figure 1—figure supplement 1 shows representative BLI sensograms and steady-state graphs for tau WT and each variant affinity to QC-01–175 and control compounds, with respective K_D values. Figure 1—figure supplement 2 shows QC-01–175 effect on monoamine oxidase (MAO) activity. The following figure supplements are available for Figure 1.

Figure 1—figure supplement 1.. In vitro characterization of tau-binding affinity to QC-01–175 and control compounds.

(A) Association-dissociation curves for BLI biosensor measurements (64 μM compound is shown, except for PE859 at 50 μM) and (B) BLI steady-state graphs, representing human recombinant biotinylated-tau WT (I-III), A152T (IV-VI) and P301L (VII-IX) in vitro binding affinity to compounds PE859 (I, IV, VII), T807 (II, V, VIII) and QC-01–175 (III, VI, IX). Representative curves of n ≥ 3 are shown.

Figure 1—figure supplement 2.. QC-01–175 effect on MAO activity.

(A) Monoamine oxidase activity assay shows that the inhibitory effect of T807 is greater than that of QC-01–175. Parnate is a known MAO inhibitor (positive control) and pomalidomide is used as a negative control. Data points represent mean values of % of MAO inhibition ± SD (n = 3). (B) IC₅₀ values extrapolated from A.

Figure 2.. Concentration effect of QC-01–175

(A) on tau protein levels of A152T and control neurons. Analysis of total tau (TAU5) and phospho-tau (S396 P-tau) levels upon treatment by western blot (B) and ELISA (C). Analysis of total tau (TAU5) and phospho-tau (S396 P-tau) levels upon treatment with the negative control QC-03–075 (D), by western blot (E) and ELISA (F). Representative western blots are shown (B, E) with mean densitometry quantification (bands corresponding to brackets)±SEM (n = 3). (C, F) For ELISA, data points represent mean tau levels (μg of total protein) normalized to vehicle-treated ± SEM (n = 4). Both assays show QC-01–175 dose-dependent effect on tau levels, with QC-03–075 minimal effect (~10%). (G) IF of A152T neurons treated with vehicle or 10 μM compound, immuno-probed for total tau (K9JA, red), P-tau (PHF-1, red) and the neuronal marker MAP2 (green), scale bar 50 μm. (H–J) Tau ELISA of control neurons treated with QC-01–175, which did not show a dose-dependent effect. (H) 8330–8-RC1 line; (I) MGH2069-RC1 line; (J) CTR2-L17-RC2 line. Data points represent mean tau levels (μg of total protein) normalized to vehicle-treated ± SEM (n = 3). All neurons were differentiated for 6 weeks and treated with compound for 24 hr. Student T-test between each concentration and vehicle-treated tau levels ^nsp> 0.05, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Figure 2—figure supplement 1 depicts the variability of degrader effect across biological replicates, by western blot, with an overall 60% to 90% efficacy. Figure 2—figure supplement 2 shows degrader effect in P301L neurons and compares concentration effect across all lines. Figure 2—source data 1 summarizes the information pertaining to each cell line included in this study. Figure 2—source data 2 includes all ELISA data. The following figure supplements are available for Figure 2.

Figure 2—figure supplement 1.. Variability of the effect of QC-01–175 across biological replicates.

(A– I) Western blot analysis of total tau (TAU5) and S396 P-Tau protein, upon 24 hr treatment with QC-01–175 or the negative controls lenalidomide, T807 and QC-03–075. Assessment of the variability in tau clearance across multiple concentrations and nine biological replicates. Red brackets indicate protein bands that were quantified by densitometry. Experiments done with A152T neurons differentiated for 6 weeks, except for (C, E) where age-matched non-mutant controls (1, 8330–8-RC1 and 2, MGH2069-RC1) were included for comparison. (J) Scatter plot of western blot densitometry, for A152T neurons treated with QC-01–175 (n = 9) or the respective negative controls (n ≥ 5), for 24 hr. Lines represent mean tau densitometry ± SD, relative to vehicle-treated. Student T-test for each concentration/antibody relative to vehicle-treated, ^nsp>0.05, *p≤0.01, **p≤0.001, ***p≤0.0001. Some blot images were cropped for the purpose of this figure only, to exclude samples not included in this manuscript (Silva and Ferguson, Manuscript in preparation, 2019), but all samples in each set were run in the same gel.

Figure 2—figure supplement 2.. Demonstration of QC-01–175 effect in tau-P301L neurons.

(A-B) Concentration effect of QC-01–175 and QC-03–075 on total tau (TAU5) and S396 P-Tau by ELISA, in P301L neurons at 6 weeks of differentiation. Data points represent mean values of tau normalized to total μg of protein, relative to vehicle-treated ± SD (n = 3). Student T-test between each concentration and vehicle-treated tau levels ^nsp> 0.05, *p<0.05, ***p<0.001, ****p<0.0001. (C-E) Overview of concentration effect of QC-01–175 24 hr treatment on (C) total tau and (D) S396 P-Tau across all genotypes. Control (1) 8330–8-RC1, control (2) MGH2069-RC1, control (3) CTR2-L17-RC2. Data points represent mean tau levels by ELISA, normalized to total μg of protein and relative to vehicle-treated ± SD (n ≥ 3). (E) Two-Way ANOVA statistical analysis for (C) and (D).

Figure 3.. Mechanism of QC-01–175 clearance of tau is CRL4^CRBN and UPS-dependent.

Neurons were pre-treated for 6 hr with (A) either CRBN ligand excess lenalidomide or tau ligand excess T807, (B) the NAE inhibitor MLN4924, the autophagy inhibitor Baf.A1, or (C) the proteasome inhibitor carfilzomib; followed by 18 hr treatment with QC-01–175 (or negative control QC-03–075), for a total of 24 hr. Total (TAU5) and P-tau S396 levels were analyzed by western blotting. (A–C) Representative blots are shown. (D–F) Densitometry bars represent tau mean intensity values ± SD (n = 3), relative to vehicle-treated samples. Student T-test of QC-01–175 samples relative to vehicle treated, and the remainder bars show p-value of each pre-treatment relative to QC-01–175 to assess rescue of clearance effect (***p<0.001, **p<0.01, *p<0.05, ^nsp>0.05). A152T neurons were differentiated for 6 weeks. Figure 3—figure supplement 1 includes additional specificity controls for A152T, P301L and control neuronal models. The following figure supplement is available for Figure 3.

Figure 3—figure supplement 1.. Additional specificity controls for QC-01–175-mediated tau clearance.

(A–C) Western blot and densitometry analysis of total tau, S396 P-tau and CRBN upon 24 hr treatment with QC-01–175 or the negative controls QC-03–075, T807 and a thalidomide analog, lenalidomide (1 μM), in (A) A152T, (B) P301L or (C) control-1 8330–8-RC1 neurons. Cropped lanes correspond to samples not included in this manuscript (Silva and Ferguson, Manuscript in preparation, 2019), but all samples in each set were run in the same gel. Densitometry graph bars represent tau or CRBN mean intensity levels ± SD (n = 3), relative to vehicle-treated samples. (D) QC-01–175 had minimal effect on CRBN upon 24 hr treatment. Student T-test of compound-treated samples relative to vehicle-treated ***p<0.001, **p<0.01, *p<0.05, ^nsp>0.05.

Figure 4.. Demonstration of ternary complex formation in A152T neurons upon QC-01–175 treatment, by co-IP and western blot analysis.

Neurons (6-week differentiated) were treated for 4 hr with 1 μM QC-01–175 ± 30 min pre-treatment with proteasome inhibitors (carfilzomib or bortezomib at 5 μM), with the goal of capturing maximum molecular interactions at 4 hr and halting tau clearance. QC-03–075 is a negative control for CRBN binding. (A) Co-IP by DDB1 pulldown and detection of tau in the complex by probing for total tau (TAU5), P-tau^S396/S404 (PHF-1), and ubiquitinated proteins (Ubi-1). (B) Co-IP by tau pulldown (TAU5) and detection of CRL4^CRBN subunits CRBN, CUL4A and DDB1. Western blot of total tau (K9JA) was used as a control. (C) Control western blot analysis with 3% (10 μg) of IP input confirms the effect of QC-01–175 ± proteasome inhibitors on tau and P-tau S396. Red arrows and brackets indicate the predicted bands for each immunoprobed protein (n = 3).

Figure 5.. Comparative analysis of the effect of QC-01–175 at (A)

0.01 μM, (B) 0.1 μM, (C) 1 μM, and (D) 10 μM after 4 hr, 8 hr or 24 hr of treatment. Graph bars represent mean levels of total tau (TAU5) and S396 P-tau protein measured by ELISA, normalized to total μg of protein and to vehicle-treated samples ± SEM (n = 3), in A152T 6-week differentiated neurons. Student T-test for each dose/time is relative to vehicle-treated tau levels ^nsp> 0.05, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Figure 5—figure supplement 1 shows concentration effect curves for QC-01–175 at 4 hr and 8 hr, for all doses tested in A152T and P301L neurons; as well as the 4 hr effect seen by western blot. Figure 5—source data 1 includes all values plotted in the main Figure and supplement. The following figure supplement is available for Figure 5.

Figure 5—figure supplement 1.. Degrader concentration and time effect on tau, in A152T and P301L neurons.

(A–C) Concentration and time effect of QC-01–175 on total tau (TAU5) and S396 P-tau levels, in A152T 6-week differentiated neurons, treated for (A) a short 4 hr interval, (B) an intermediate 8 hr interval, (C) compared to 24 hr treatment. (D) Concentration effect of QC-01–175 4 hr treatment in A152T neurons, by western blot analysis of total tau (TAU5) and P-tau S396 (representative blot is shown) with mean densitometry quantification (bands corresponding to brackets)±SEM (n = 3). (F–H) Concentration and time effect of QC-01–175 on total tau (TAU5) and S396 P-tau levels, in P301L 6-week differentiated neurons treated for (F) 4 hr, (G) 8 hr and (H) 24 hr. (A-C, F-H) Data points represent mean levels of tau protein measured by ELISA, normalized to total μg of protein and to vehicle-treated ± SEM (n = 3). Student T-test between each dose and vehicle-treated tau levels ^nsp>0.05, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 6.. Mass spectrometry-based proteomics to quantify the effect of QC-01–175 treatment on the proteome of A152T neurons.

6-week differentiated neurons were treated for 4 hr with (A) 1 μM of QC-01–175, (B) 1 μM of the negative control QC-03–075, or (C) 10 μM MLN4924 (NAE inhibitor, 30 min pre-treatment) and 1 μM of QC-01–175. Upon degrader QC-01–175 treatment (A), three off-targets were detected within statistical significance, which all belong to known IMiD targets, an effect rescued by the negative control (B) or inhibition of neddylation by MLN4924 (C). Significant hits were assessed by moderated t-test as implemented in the limma package (Ritchie et al., 2015), with the log2 fold change shown on the y-axis, and negative log10 P values on the x-axis (n = 3 for treatment with DMSO, QC-01–175, and QC-03–075, and n = 2 for QC-01–175 + MLN4924).

Figure 7.. QC-01–175 treatment rescued stress vulnerability of A152T neurons.

(A) Aβ(1-42) proteotoxicity causes concentration- and genotype-dependent loss of neuronal vulnerability, affecting preferentially A152T and P301L neurons, with a rescue by MAPT KO. Data points represent mean viability relative to vehicle-treated neurons (100%)±SEM (n ≥ 3); two-way ANOVA statistical analysis relative to non-mutant control-1 neurons (black curve, 8330–8-RC1), ****p<0.0001, ^nsP > 0.05. (B) Assay overview to measure effect of the stressor Aβ(1-42) on neuronal viability (c) and potential rescue by pre-treatment with QC-01–175 (b). Effect of 24 hr treatment with QC-01–175 alone was also tested (a). (C) QC-01–175 (light blue-stripe bar) but not the negative control QC-03–075 (dark blue-stripe bar), rescued viability loss caused by Aβ(1-42) (white-stripe bar) in A152T neurons differentiated for 8 weeks, in a comparable manner to genetic MAPT knockout (black-stripe bar). Graph bars represent mean % viability ± SD, relative to vehicle-treated (white) neurons. T-test ***p≤0.001; ^nsp>0.05 (n = 3). Figure 7—source data 1 includes all values plotted in main Figure 7A and C.

Scheme 1.. Synthesis route for the tau degrader QC-01-175.

Chemical structure 1.. 3-(4-(4-nitropyridin-3-yl)phenyl)propan-1-ol (3)

Chemical structure 2.. 3-(4-(3-((tert-butyldiphenylsilyl)oxy)propyl)phenyl)-4-nitropyridine (4)

Chemical structure 3.. 7-(3-((tert-butyldiphenylsilyl)oxy)propyl)-5H-pyrido[4,3-b]indole (5)

Chemical structure 4.. tert-butyl 7-(3-((tert-butyldiphenylsilyl)oxy)propyl)-5H-pyrido[4,3-b]indole-5-carboxylate (6)

Chemical structure 5.. tert-butyl 7-(3-hydroxypropyl)-5H-pyrido[4,3-b]indole-5-carboxylate (7)

Chemical structure 6.. 3-(5H-pyrido[4,3-b]indol-7-yl)propanoic acid (10)

Chemical structure 7.. QC-01-175

Chemical structure 8.. QC-03-075: Was prepared according to Scheme 1.