Aggregate-selective removal of pathological tau by clustering-activated degraders

Abstract

Selective degradation of pathological protein aggregates while sparing monomeric forms is of major therapeutic interest. The E3 ligase tripartite motif-containing protein 21 (TRIM21) degrades antibody-bound proteins in an assembly state-specific manner due to the requirement of TRIM21 RING domain clustering for activation, yet effective targeting of intracellular assemblies remains challenging. Here, we fused the RING domain of TRIM21 to a target-specific nanobody to create intracellularly expressed constructs capable of selectively degrading assembled proteins. We evaluated this approach against green fluorescent protein-tagged histone 2B (H2B-GFP) and tau, a protein that undergoes pathological aggregation in Alzheimer's and other neurodegenerative diseases. RING-nanobody degraders prevented or reversed tau aggregation in culture and in vivo, with minimal impact on monomeric tau. This approach may have therapeutic potential for the many disorders driven by intracellular protein aggregation.

Fig. 1. R-Nb_F8-2 rapidly degrades aggregated tau species.

(A) Cytosolic TRIM21 exists as an inactive homodimer, with each monomer consisting of a RING E3 ligase domain, B-Box, Coiled-coil, and an antibody Fc binding PRYSPRY domain. (B) The clustering of TRIM21 upon antibody-coated polyvalent substrates allows for RING E3 ligase domain cross-activation and subsequent polyubiquitination. (C, D) TRIM21 RING-Nanobody (R-Nb) degraders were engineered to similarly require RING domain cross-activation to induce target degradation and therefore become active upon binding to (D) homotypic assemblies of antigens but not (C) monomeric antigens. (E) Live cell fluorescent microscopy of TVA cells which express aggregated P301S tau-venus and DOX-inducible degrader, R-Nb_F8-2, with an mCherry reporter. Scale bar, 20µm. (F) Quantification of tau-venus aggregates from live cell imaging of TVA cells expressing DOX inducible R-Nb_F8-2, TRIM21 RING, or F8-2. Images were captured and analyzed every 30 minutes for 20 hours following DOX addition. (G) Capillary based immunoblots of sarkosyl insoluble fractions from TVA cell lines as above treated with or without DOX for 15 hours. Samples were probed for total tau (DAKO) and phospho-tau (pTau) (AT8). (H) Sarkosyl soluble fractions from the same experiment were probed for total tau (DAKO) and actin as a loading control. (I) Sarkosyl soluble samples were further analyzed by diluting 1:4 in PBS and probing with total tau (DAKO and HT7) and pTau (AT8 and pS422) antibodies. (J) Representative super-resolution microscopy images of tau-venus aggregates in lysates from TVA cells expressing DOX inducible R-Nb_F8-2 treated with or without DOX for 15 hours, using anti-tau antibody HT7 for both capture and detection. Scale bar, 1µm (K) Histogram showing the size and count of tau-venus aggregates detected from super-resolution microscopy images. Shaded areas represent mean ± SD. (F) n = 8 biological replicates per condition. (K) n = 5 or 6 technical replicates per condition.

Fig. 2. Selective degradation of aggregated tau by R-Nb is dependent on nanobody affinity and protects cells against seeded tau aggregation.

(A) TVA cells were transfected with R-Nb plasmids employing nanobodies of varying in vitro affinities for tau-venus. F8-2 WT (1332 nM), F8-2_S54L+T127A (683 nM), F8-2_R111C (211 nM), H3-2 (32 nM), vhhGFP4 (1 nM) or Nb139 (p53 targeting, negative control). 24 hours post-transfection the percentage of cells containing aggregated tau-venus was quantified via fluorescence microscopy. (B) TVS cells were transfected with the same R-Nb plasmids, with venus fluorescence quantified by flow-cytometry 24 hours post-transfection. (C) DOX inducible R-Nb_F8-2 TVA cells were lysed 15 hours post-DOX addition, and their proteome compared to untreated (-DOX) cells via mass spectrometry. Significantly altered proteins (q-value <0.01, fold change >2 or <-2) are labeled, alongside TauP301S-Venus. (D) Immunoblot of lysates from DOX inducible R-Nb_F8-2 TVS cells, 24 hours after +/- DOX treatment. Samples probed for tau (Tau12 antibody) and GAPDH. (E) Amounts of tau, normalized to GAPDH, in TVS cells + DOX relative to – DOX amounts. (F) TVS cells pretreated +/- DOX for 24 hours were challenged with heparin assembled P301S tau assemblies. Percentage of cells containing tau-venus aggregates quantified 72 hours later via fluorescent microscopy. (G) Percentage of tau-venus aggregates remaining in TVA cells expressing DOX inducible R-Nb_F8-2, treated with E1/Proteasome/VCP/autophagy inhibitors (solubilized in DMSO) and DOX for 10 hours, compared to TVA without DOX, quantified by fluorescence microscopy. TAK-243, 100 nM; MG132, 125 nM; NMS-873, 5 µM; Bafilomycin A1, 400 nM. (H) Lysates from TVA cells expressing DOX inducible R-Nb_F8-2, or RING/F8-2 only controls, treated +/- DOX for 15 hours, were introduced to TVS cells via Lipofectamine (LF). Seeded aggregation of tau-venus was quantified 72 hours later via fluorescent microscopy. Error bars indicate mean ± SD. (A), (B), (E), (F), (G) and (H) n = 3 biological replicates. (C) n = 3 technical replicates. (A), (B), (G) and (H) one-way analysis of variance (ANOVA) with Tukey’s correction for multiple comparisons. (E) unpaired t test. (F) two-way ANOVA with Šídák’s correction for multiple comparisons. ns, not significant; ***P < 0.001; ****P < 0.0001.

Fig. 3. In vivo degradation of H2B-GFP in the mouse brain using R-Nb_vhhGFP4.

(A) H2B-GFP protein amounts in hippocampi of transgenic H2B-GFP mice unilaterally injected at two months of age with 2x10⁹ genome copies (GC) AAV1/2 (encoding R-Nb_vhhGFP4, TRIM21 RING, or vhhGFP4) compared to non-injected contralateral hippocampi, quantified via GFP ELISA and normalized to hippocampus mass. (B) Comparison of H2B-GFP protein amounts in hippocampi injected with AAV1/2 R-Nb_vhhGFP4 for 10 or 30 days, quantified via ELISA and presented as percentage H2B-GFP protein compared to non-injected contralateral hippocampi. (C) Representative fluorescent microscopy images of hippocampi injected with AAV1/2 encoding R-Nb_vhhGFP4, TRIM21 RING, or vhhGFP4 30 days post-injection. Scale bar, 1mm. (D) GFP intensity in AAV1/2 injected vs non-injected hippocampi was quantified from fluorescent microscopy images and normalized against the average signal in non-injected hippocampi. (A) and (B) each point represents the average of two technical replicates from one mouse. n = 3 or 4 mice. (D) each point represents the average of 4 brain slices from one mouse. n = 3 mice. (A) and (D) paired t test, (B) un-paired t test. ns, not significant; *P < 0.05; ****P < 0.0001.

Fig. 4. Reduction of tau pathology in the aged mouse brain and primary neurons via R-Nb_F8-2.

(A) Representative IF images 30 days post-stereotaxic injection of AAV1/2-CAG-R-Nb_F8-2-T2A-mCherry into the left frontal cortex (two 2x10⁹ GC doses) of 5.5-month-old Tg2541 mice. pTau detected using AT8. Scale bars, 1mm. (B) Quantification of AT8 coverage in AAV1/2 injected vs contralateral frontal cortex hemispheres, 10 days post-injection. (C) Western blot of injected and contralateral cortical hemisphere lysates 10 days after injection with R-Nb_F8-2 AAV1/2. Samples ran in duplicate and probed for pTau, total tau, and loading control with AT100, BR134 and CypB antibodies respectively. (D) Representative IF images of primary Tg2541 mouse neuron cultures, “seeded” via addition of 50 nM heparin-assembled P301S tau assemblies 7 days after pre-treatment with AAV1/2-hSyn-R-Nb_F8-2-T2A-mCherry (20,000 GC/cell). pTau aggregates detected 7 days post-seeding via anti-pTau pS422 staining, with neurons identified via MAP2 staining. Scale bar, 100µm. (E) Quantification of pS422 tau puncta normalized against MAP2 coverage from IF images. Seeded neurons were pre-treated with AAV1/2 encoding R-Nb_F8-2, TRIM21 RING (T21R) or F8-2 only. (F) Lysates from tau seeded neurons pre-treated with R-Nb_F8-2 AAV1/2 (various GC doses) were introduced to TVS cells via Lipofectamine (LF). Tau-venus aggregation was quantified 72 hours later via fluorescent microscopy. (G) Relative amounts of pS422 tau puncta (normalized against MAP2 coverage) in tau seeded neurons pre-treated with catalytically dead or active R-Nb_F8-2 AAV1/2 (20,000 GC/cell), compared to non-transduced neurons. (H) Primary C57BL/6 mouse neuron cultures were transduced with AAV1/2-hSyn-0N4R P301S tau-venus (20,000 GC/cell), seeded 7 days later with heparin assembled P301S tau assemblies, then transduced 3 days post-seeding with R-Nb_F8-2 AAV1/2 (20,000 GC/cell), after which images were captured via live cell imaging. Scale bar, 20µm. Error bars represent mean ± SD. (B) n = 3 mice, 4 technical repeats per point. (E) and (G) n = 4 - 8 biological repeats, 2 technical repeats per point. (F) n = 4 biological repeats. (B) Paired t test. (E), (F) and (G) ANOVA with Tukey’s correction for multiple comparisons. ns, not significant; *P < 0.05; ***P < 0.001; ****P < 0.0001.

Fig. 5. Global CNS expression of R-Nb_F8-2 reduces tau pathology in mice.

(A) Timeline of short (10 day) and long (2 month) experimental protocol using R-Nb_F8-2 AAV_9P31 in Tg2541 mice. In both scenarios mice were treated with a single dose of 1x10¹¹ GC of R-Nb_F8-2 AAV_9P31 via tail vein injection. (B) Representative IF images of frontal cortices stained with AT8 from aged mice treated with or without R-Nb_F8-2 AAV_9P31 for 10 days. Scale bar, 1mm (C) Quantification of AT8 coverage from IF images in (B). (D) Representative IF images of spinal cords from the same mice, stained for AT8. Scale bar, 0.5mm (E) Quantification of AT8 staining in spinal cord sections (D). (F) Sarkosyl soluble and insoluble fractions were prepared from whole brains of aged mice treated with or without 1x10¹¹ GC R-Nb_F8-2 AAV_9P31 for two months. Insoluble fractions were blotted for pTau (pS422 and AT8) and total tau (Tau12), with soluble fractions blotted for Tau12 and mCherry, with CypA as a loading control. Each lane represents an individual mouse. (G) Amounts of seeded tau aggregation in TVA cells treated with brain homogenates from aged mice treated with or without R-Nb_F8-2 AAV_9P31 for two months. Error bars represent mean ± SD. (C) and (E) each point represents the average of 4 brain slices from one mouse. n = 3 mice. (G) Each point represents the average percentage of TVS cells seeded (6 technical replicates) by homogenate from one mouse. n = 4 or 5 mice. Unpaired t test, *P < 0.05.