Gel‐like inclusions of C‐terminal fragments of TDP‐43 sequester stalled proteasomes in neurons

Report19 April 2022Open Access Transparent process Gel-like inclusions of C-terminal fragments of TDP-43 sequester stalled proteasomes in neurons Henrick Riemenschneider Henrick Riemenschneider German Center for Neurodegenerative Diseases (DZNE), Munich, Munich, Germany Contribution: Conceptualization, Formal analysis, ​Investigation, Visualization, Methodology, Writing - original draft, Writing - review & editing Search for more papers by this author Qiang Guo Qiang Guo orcid.org/0000-0003-3520-5439 Department of Molecular Structural Biology, Max Planck Institute of Biochemistry, Martinsried, Germany State Key Laboratory of Protein and Plant Gene Research, School of Life Sciences and Peking-Tsinghua Center for Life Sciences, Peking University, Beijing, China Contribution: Conceptualization, Data curation, Formal analysis, ​Investigation, Visualization, Methodology, Writing - original draft, Writing - review & editing Search for more papers by this author Jakob Bader Jakob Bader orcid.org/0000-0002-6575-0609 Department of Proteomics and Signal Transduction, Max Planck Institute for Biochemistry, Martinsried, Germany Contribution: Conceptualization, Data curation, Formal analysis, ​Investigation, Methodology Search for more papers by this author Frédéric Frottin Frédéric Frottin orcid.org/0000-0002-2756-7838 Department of Cellular Biochemistry, Max Planck Institute for Biochemistry, Martinsried, Germany Institute for Integrative Biology of the Cell (I2BC), Université Paris-Saclay, CEA, CNRS, Gif-sur-Yvette, France Contribution: Formal analysis, Methodology Search for more papers by this author Daniel Farny Daniel Farny German Center for Neurodegenerative Diseases (DZNE), Munich, Munich, Germany Contribution: Methodology Search for more papers by this author Gernot Kleinberger Gernot Kleinberger German Center for Neurodegenerative Diseases (DZNE), Munich, Munich, Germany Contribution: Methodology Search for more papers by this author Christian Haass Christian Haass orcid.org/0000-0002-4869-1627 German Center for Neurodegenerative Diseases (DZNE), Munich, Munich, Germany Chair of Metabolic Biochemistry, Faculty of Medicine, Biomedical Center (BMC), Ludwig-Maximilians-Universität Munich, Munich, Germany Munich Cluster of Systems Neurology (SyNergy), Munich, Germany Contribution: Supervision, Writing - review & editing Search for more papers by this author Matthias Mann Matthias Mann orcid.org/0000-0003-1292-4799 Department of Proteomics and Signal Transduction, Max Planck Institute for Biochemistry, Martinsried, Germany Contribution: Funding acquisition Search for more papers by this author F. Ulrich Hartl F. Ulrich Hartl orcid.org/0000-0002-7941-135X Department of Cellular Biochemistry, Max Planck Institute for Biochemistry, Martinsried, Germany Munich Cluster of Systems Neurology (SyNergy), Munich, Germany Contribution: Supervision, Funding acquisition, Writing - review & editing Search for more papers by this author Wolfgang Baumeister Wolfgang Baumeister Department of Molecular Structural Biology, Max Planck Institute of Biochemistry, Martinsried, Germany Contribution: Funding acquisition Search for more papers by this author Mark S Hipp Mark S Hipp orcid.org/0000-0002-0497-3016 Department of Cellular Biochemistry, Max Planck Institute for Biochemistry, Martinsried, Germany Department of Biomedical Sciences of Cells and Systems, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands School of Medicine and Health Sciences, Carl von Ossietzky University Oldenburg, Oldenburg, Germany Contribution: Conceptualization, Formal analysis, Supervision, Funding acquisition, ​Investigation, Writing - review & editing Search for more papers by this author Felix Meissner Felix Meissner orcid.org/0000-0003-1000-7989 Department of Proteomics and Signal Transduction, Max Planck Institute for Biochemistry, Martinsried, Germany Department of Systems Immunology and Proteomics, Medical Faculty, Institute of Innate Immunity, University of Bonn, Germany Contribution: Supervision Search for more papers by this author Rubén Fernández-Busnadiego Corresponding Author Rubén Fernández-Busnadiego [email protected] orcid.org/0000-0002-8366-7622 Department of Molecular Structural Biology, Max Planck Institute of Biochemistry, Martinsried, Germany Institute of Neuropathology, University Medical Center Göttingen, Göttingen, Germany Cluster of Excellence “Multiscale Bioimaging: from Molecular Machines to Networks of Excitable Cells” (MBExC), University of Göttingen, Göttingen, Germany Contribution: Conceptualization, Formal analysis, Supervision, Funding acquisition, ​Investigation, Writing - original draft, Writing - review & editing Search for more papers by this author Dieter Edbauer Corresponding Author Dieter Edbauer [email protected] orcid.org/0000-0002-7186-4653 German Center for Neurodegenerative Diseases (DZNE), Munich, Munich, Germany Munich Cluster of Systems Neurology (SyNergy), Munich, Germany Graduate School of Systemic Neurosciences (GSN), Ludwig-Maximilians-University Munich, Munich, Germany Contribution: Conceptualization, Formal analysis, Supervision, Funding acquisition, ​Investigation, Visualization, Writing - original draft, Writing - review & editing Search for more papers by this author Henrick Riemenschneider Henrick Riemenschneider German Center for Neurodegenerative Diseases (DZNE), Munich, Munich, Germany Contribution: Conceptualization, Formal analysis, ​Investigation, Visualization, Methodology, Writing - original draft, Writing - review & editing Search for more papers by this author Qiang Guo Qiang Guo orcid.org/0000-0003-3520-5439 Department of Molecular Structural Biology, Max Planck Institute of Biochemistry, Martinsried, Germany State Key Laboratory of Protein and Plant Gene Research, School of Life Sciences and Peking-Tsinghua Center for Life Sciences, Peking University, Beijing, China Contribution: Conceptualization, Data curation, Formal analysis, ​Investigation, Visualization, Methodology, Writing - original draft, Writing - review & editing Search for more papers by this author Jakob Bader Jakob Bader orcid.org/0000-0002-6575-0609 Department of Proteomics and Signal Transduction, Max Planck Institute for Biochemistry, Martinsried, Germany Contribution: Conceptualization, Data curation, Formal analysis, ​Investigation, Methodology Search for more papers by this author Frédéric Frottin Frédéric Frottin orcid.org/0000-0002-2756-7838 Department of Cellular Biochemistry, Max Planck Institute for Biochemistry, Martinsried, Germany Institute for Integrative Biology of the Cell (I2BC), Université Paris-Saclay, CEA, CNRS, Gif-sur-Yvette, France Contribution: Formal analysis, Methodology Search for more papers by this author Daniel Farny Daniel Farny German Center for Neurodegenerative Diseases (DZNE), Munich, Munich, Germany Contribution: Methodology Search for more papers by this author Gernot Kleinberger Gernot Kleinberger German Center for Neurodegenerative Diseases (DZNE), Munich, Munich, Germany Contribution: Methodology Search for more papers by this author Christian Haass Christian Haass orcid.org/0000-0002-4869-1627 German Center for Neurodegenerative Diseases (DZNE), Munich, Munich, Germany Chair of Metabolic Biochemistry, Faculty of Medicine, Biomedical Center (BMC), Ludwig-Maximilians-Universität Munich, Munich, Germany Munich Cluster of Systems Neurology (SyNergy), Munich, Germany Contribution: Supervision, Writing - review & editing Search for more papers by this author Matthias Mann Matthias Mann orcid.org/0000-0003-1292-4799 Department of Proteomics and Signal Transduction, Max Planck Institute for Biochemistry, Martinsried, Germany Contribution: Funding acquisition Search for more papers by this author F. Ulrich Hartl F. Ulrich Hartl orcid.org/0000-0002-7941-135X Department of Cellular Biochemistry, Max Planck Institute for Biochemistry, Martinsried, Germany Munich Cluster of Systems Neurology (SyNergy), Munich, Germany Contribution: Supervision, Funding acquisition, Writing - review & editing Search for more papers by this author Wolfgang Baumeister Wolfgang Baumeister Department of Molecular Structural Biology, Max Planck Institute of Biochemistry, Martinsried, Germany Contribution: Funding acquisition Search for more papers by this author Mark S Hipp Mark S Hipp orcid.org/0000-0002-0497-3016 Department of Cellular Biochemistry, Max Planck Institute for Biochemistry, Martinsried, Germany Department of Biomedical Sciences of Cells and Systems, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands School of Medicine and Health Sciences, Carl von Ossietzky University Oldenburg, Oldenburg, Germany Contribution: Conceptualization, Formal analysis, Supervision, Funding acquisition, ​Investigation, Writing - review & editing Search for more papers by this author Felix Meissner Felix Meissner orcid.org/0000-0003-1000-7989 Department of Proteomics and Signal Transduction, Max Planck Institute for Biochemistry, Martinsried, Germany Department of Systems Immunology and Proteomics, Medical Faculty, Institute of Innate Immunity, University of Bonn, Germany Contribution: Supervision Search for more papers by this author Rubén Fernández-Busnadiego Corresponding Author Rubén Fernández-Busnadiego [email protected] orcid.org/0000-0002-8366-7622 Department of Molecular Structural Biology, Max Planck Institute of Biochemistry, Martinsried, Germany Institute of Neuropathology, University Medical Center Göttingen, Göttingen, Germany Cluster of Excellence “Multiscale Bioimaging: from Molecular Machines to Networks of Excitable Cells” (MBExC), University of Göttingen, Göttingen, Germany Contribution: Conceptualization, Formal analysis, Supervision, Funding acquisition, ​Investigation, Writing - original draft, Writing - review & editing Search for more papers by this author Dieter Edbauer Corresponding Author Dieter Edbauer [email protected] orcid.org/0000-0002-7186-4653 German Center for Neurodegenerative Diseases (DZNE), Munich, Munich, Germany Munich Cluster of Systems Neurology (SyNergy), Munich, Germany Graduate School of Systemic Neurosciences (GSN), Ludwig-Maximilians-University Munich, Munich, Germany Contribution: Conceptualization, Formal analysis, Supervision, Funding acquisition, ​Investigation, Visualization, Writing - original draft, Writing - review & editing Search for more papers by this author Author Information Henrick Riemenschneider1,†, Qiang Guo2,3,†, Jakob Bader4, Frédéric Frottin5,6, Daniel Farny1, Gernot Kleinberger1, Christian Haass1,7,8, Matthias Mann4, F. Ulrich Hartl5,8, Wolfgang Baumeister2, Mark S Hipp5,9,10, Felix Meissner4,11, Rubén Fernández-Busnadiego *,2,12,13 and Dieter Edbauer *,1,8,14 1German Center for Neurodegenerative Diseases (DZNE), Munich, Munich, Germany 2Department of Molecular Structural Biology, Max Planck Institute of Biochemistry, Martinsried, Germany 3State Key Laboratory of Protein and Plant Gene Research, School of Life Sciences and Peking-Tsinghua Center for Life Sciences, Peking University, Beijing, China 4Department of Proteomics and Signal Transduction, Max Planck Institute for Biochemistry, Martinsried, Germany 5Department of Cellular Biochemistry, Max Planck Institute for Biochemistry, Martinsried, Germany 6Institute for Integrative Biology of the Cell (I2BC), Université Paris-Saclay, CEA, CNRS, Gif-sur-Yvette, France 7Chair of Metabolic Biochemistry, Faculty of Medicine, Biomedical Center (BMC), Ludwig-Maximilians-Universität Munich, Munich, Germany 8Munich Cluster of Systems Neurology (SyNergy), Munich, Germany 9Department of Biomedical Sciences of Cells and Systems, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands 10School of Medicine and Health Sciences, Carl von Ossietzky University Oldenburg, Oldenburg, Germany 11Department of Systems Immunology and Proteomics, Medical Faculty, Institute of Innate Immunity, University of Bonn, Germany 12Institute of Neuropathology, University Medical Center Göttingen, Göttingen, Germany 13Cluster of Excellence “Multiscale Bioimaging: from Molecular Machines to Networks of Excitable Cells” (MBExC), University of Göttingen, Göttingen, Germany 14Graduate School of Systemic Neurosciences (GSN), Ludwig-Maximilians-University Munich, Munich, Germany † These authors contributed equally to this work *Corresponding author. Tel: +49 551 3960745; E-mail: [email protected] *Corresponding author. Tel: +49 89 440046510; E-mail: [email protected] EMBO Reports (2022)23:e53890https://doi.org/10.15252/embr.202153890 PDFDownload PDF of article text and main figures. Peer ReviewDownload a summary of the editorial decision process including editorial decision letters, reviewer comments and author responses to feedback. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions Figures & Info Abstract Aggregation of the multifunctional RNA-binding protein TDP-43 defines large subgroups of amyotrophic lateral sclerosis and frontotemporal dementia and correlates with neurodegeneration in both diseases. In disease, characteristic C-terminal fragments of ~25 kDa ("TDP-25") accumulate in cytoplasmic inclusions. Here, we analyze gain-of-function mechanisms of TDP-25 combining cryo-electron tomography, proteomics, and functional assays. In neurons, cytoplasmic TDP-25 inclusions are amorphous, and photobleaching experiments reveal gel-like biophysical properties that are less dynamic than nuclear TDP-43. Compared with full-length TDP-43, the TDP-25 interactome is depleted of low-complexity domain proteins. TDP-25 inclusions are enriched in 26S proteasomes adopting exclusively substrate-processing conformations, suggesting that inclusions sequester proteasomes, which are largely stalled and no longer undergo the cyclic conformational changes required for proteolytic activity. Reporter assays confirm that TDP-25 impairs proteostasis, and this inhibitory function is enhanced by ALS-causing TDP-43 mutations. These findings support a patho-physiological relevance of proteasome dysfunction in ALS/FTD. Synopsis TDP-25, a C-terminal fragment of TDP-43 found in ALS and FTD patients, forms cytoplasmic inclusions with gel-like properties in primary neurons. Proteasome enrichment and impaired proteostasis support the relevance of proteasome dysfunction in ALS/FTD. neuronal cytoplasmic TDP-25 inclusions adopt an amorphous gel-like state without detectable fibrils stalled proteasomes are present in the inclusions multiple ALS-causing mutations further increase proteasomal impairment Introduction TDP-43 aggregation is the disease-defining pathological hallmark in > 90% of patients with amyotrophic lateral sclerosis (ALS) and ~45% of patients with frontotemporal dementia (FTD) (Gao et al, 2017; Prasad et al, 2019; Tziortzouda et al, 2021). However, we still have a limited understanding of the native structure of TDP-43 inclusions and their role in disease (Gao et al, 2017; Chien et al, 2021). Predominantly, cytoplasmic neuronal inclusions of TDP-43 correlate strongly with regional neuron loss in spinal cord, motor cortex, or frontal-temporal cortical regions (Mackenzie et al, 2013). These inclusions are enriched in 25-35 kDa C-terminal fragments of TDP-43 that contain a glycine-rich low-complexity region (Neumann et al, 2006; Igaz et al, 2008). Rare autosomal dominant ALS-causing mutations in TDP-43 cluster in this region, although their pathomechanism is still poorly understood, and they have only modest effects in vitro and in knockin models (Prasad et al, 2019). However, overexpression of the inclusion-forming 25 kDa C-terminal fragment ("TDP-25") triggers neurodegeneration in mice even without disease-associated mutations (Walker et al, 2015). Distinct biophysical mechanisms can drive inclusion formation in the context of neurodegenerative diseases: (i) formation of highly insoluble amyloid fibrils with cross β-sheet conformation (Eisenberg & Jucker, 2012), (ii) liquid-liquid phase separation (LLPS) into highly dynamic liquid-like droplets (Gomes & Shorter, 2019), which may solidify and adopt amyloid-like conformations under pathological conditions (Kato et al, 2012; Patel et al, 2015; Qamar et al, 2018). By adopting aberrant conformations, aggregated proteins may engage in toxic cellular interactions (Olzscha et al, 2011; Hipp et al, 2019). Rapid advances in cryo-electron microscopy (cryo-EM) elucidated the structure of amyloid fibrils formed by various disease-associated aggregating proteins purified from patient tissue at near-atomic resolution, most recently including TDP-43 (Creekmore et al, 2021; Arseni et al, 2022). Cryo-EM revealed a unique amyloid conformation of TDP-43 fibrils purified from patients, with little resemblance to fibrillar structures derived from recombinant TDP-43 C-terminal region (at pH 4) and shorter synthetic peptide fragments (Guenther et al, 2018; Cao et al, 2019; Li et al, 2021; Arseni et al, 2022). Full-length TDP-43 has also been reported to form liquid droplets through LLPS in vitro (Conicella et al, 2016), and RNA-free TDP-43 can assemble into so-called anisosomes consisting of a liquid TDP-43 shell and a HSP70 core in the nucleus, suggesting that TDP-43 condensates solidify into fibrils in disease (Yu et al, 2021; Arseni et al, 2022). ALS/FTD-causing mutations in C9orf72, OPTN, SQSTM1, TBK1, UBQLN2, and VCP are linked to the ubiquitin-proteasome system (UPS) and autophagy, which can clear aggregated and phase-separated proteins, suggesting the proteostasis system is of particular importance at least in genetic ALS/FTD (Gitcho et al, 2009; Deng et al, 2011; Hipp et al, 2019). This view is reinforced by the observation that proteasome inhibition leads to the formation of TDP-43 anisomes (Yu et al, 2021). Cryo-electron tomography (cryo-ET) is a powerful method to visualize neurotoxic inclusions within their native cellular environment complementing the analysis of purified fibrils. Cryo-ET studies have revealed a striking heterogeneity of ultrastructure and pathomechanisms for neuronal inclusion proteins (Bauerlein et al, 2020). In particular, we have used cryo-ET to show that amyloid-like poly-GA inclusions linked to C9orf72 ALS/FTD disrupt neuronal proteostasis by sequestering proteasomes in a rare transition state, suggesting the cyclic conformational changes required for substrate processing (Collins & Goldberg, 2017) are altered by the association with the inclusions (Guo et al, 2018). Here, we aimed to elucidate gain-of-function mechanisms and the structure of cytoplasmic TDP-43 aggregates found in sporadic and most genetic ALS/FTD cases. We focused on the aggregation-prone TDP-25 fragment (residues 220-414 of full-length human TDP-43) (Neumann et al, 2006; Zhang et al, 2009) using a pipeline of cryo-ET, proteomics, and functional assays. Neuronal TDP-25 inclusions show amorphous morphology and gel-like biophysical properties. The inclusions are enriched in stalled 26S proteasomes similar to poly-GA inclusions, suggesting inhibition of the UPS is a common pathomechanism in ALS/FTD. Results and Discussion TDP-25 inclusions are amorphous and enriched in 26S proteasomes Expression of GFP-tagged TDP-25 in rat primary neurons resulted in abundant cytoplasmic inclusions phosphorylated at disease-specific sites (Fig 1A) (Hasegawa et al, 2008). To explore the pathomechanisms enhanced by known pathogenic mutations, we additionally used a TDP-25 variant containing eight mutations (G290A, G294V, G298S, A315T, M337V, G348C, N352S, and A382T) that individually cause ALS (Prasad et al, 2019). Wild-type and mutant GFP-TDP-25 formed inclusions predominantly in the cell soma that were of similar appearance by light microscopy and showed a comparable degree of disease-specific phosphorylation (TDP-25 wild-type: 95.8 ± 2.9% vs. TDP-25 mutant: 96.2 ± 3.1%, mean ± SD from n = 3 biological replicates) (Fig 1A). While GFP-TDP-43 was almost completely soluble in RIPA buffer, a large fraction of wild-type and mutant GFP-TDP-25 was only solubilized upon sequential extraction of the RIPA-insoluble material with 2% SDS (Fig 1B) indicative of stronger intermolecular interactions in aggregating TDP-25. Solubility of wild-type and mutant TDP-25 was not significantly different (Fig 1B). Figure 1. TDP-25 forms amorphous inclusions enriched in proteasomes Primary rat hippocampal neurons were transduced with constructs encoding for GFP-tagged TDP-25 (amino acids 220-414 of TDP-43) on day 5 in vitro and cultured for 8 additional days (DIV5+8). Immunofluorescence of GFP-TDP-25 wild-type and variant containing eight ALS-causing mutations shows disease-related phosphorylation at serines 403/404. Counterstain to label the neuronal cytoskeleton (MAP2) and nuclei (DAPI). Scale bar = 10 µm. Hippocampal neurons transduced with GFP-TDP-25 constructs (DIV5+8) or GFP-TDP-43 (DIV5+4 due to higher toxicity) were sequentially extracted with RIPA buffer, followed by 2% SDS buffer and analyzed via immuoblotting. Left panel shows representative immunoblot for GFP and loading control calnexin. Upper arrow marks GFP-TDP-43 band, lower arrow marks GFP-TDP-25 band. Right panel shows quantification of the ratio of the respective GFP densiometric signal in SDS extract to RIPA extract. Barplots showing means ± SD from n = 3 independent experiments. TDP-43 (0.039 ± 0.029, mean ± CI) vs. TDP-25 wt (1.22 ± 0.296) vs. TDP-25 mut (0.928 ± 0.564): F(2,6) = 51.4, ***P = 0.000168, η² = 0.94. TDP-43 vs. TDP-25 wt: ***P = 1.67*10−4, TDP-43 vs. TDP-25 mut: ***P = 7.983*10−4, TDP-25 wt vs. TDP-25 mut: P = 0.1182, One-way ANOVA with Tukey's post-hoc test. Tomographic slice of an aggregate within a GFP-TDP-25 wild-type transduced neuron (DIV5+8). Colored boxes show a series of higher magnification tomographic slices of representative protein complexes detected in the tomogram. Green boxes show side views of single-capped, yellow boxes show ring-like cross-sections of 26S proteasomes. Red dotted line segments aggregate area. Scale bar = 200 nm. 3D rendering of the aggregate shown in (C). Amorphous aggregate material is labeled in red, proteasomes in violet, ribosomes in yellow, TRiC/CCT chaperones in green, mitochondria (Mt) in white, endoplasmic reticulum (ER) in pink and other vesicles (V) in light yellow. The irregular aggregate structures were approximately segmented using a threshold-based approach for visualization purposes. Average structures of proteasomes, ribosomes, and TRiC/CCT were pasted at the locations and orientations determined by template matching. See also Movie EV1. Subtomogram averaging of macromolecules in GFP-TDP-25 inclusions reveals the proteasome structure at ~20 Å resolution (see Fig EV1). The positions of Rpn5/PSMD11 and Rpn6/PSDM12 are indicated. Prominent extra densities in the substrate binding region are colored in pink in the 3D rendering. Classification based on the conformation of the regulatory particle in TDP-25 inclusions compared with non-transduced control neurons (Asano et al, 2015) and poly-GA inclusions (Guo et al, 2018). GS: Ground State, SPS: Substrate Processing State. Due to the uncertainties inherent to the classification procedure, it is possible that a small fraction of particles adopted other conformations. Download figure Download PowerPoint To investigate the downstream consequences of TDP-25 aggregation, we analyzed neuronal GFP-TDP-25 inclusions in situ using cryo-ET (Bauerlein et al, 2020). To that end, neurons were cultured, transduced with GFP-tagged TDP-25 constructs (wild-type and mutant) and vitrified on EM grids. Vitrified grids were imaged by cryo-light microscopy to determine the subcellular location of TDP-25 inclusions. Correlative light-electron microscopy (CLEM) was performed to locate back the inclusions in a dual-beam cryo-focused ion beam/scanning electron microscope (cryo-FIB/SEM). Thin (100–200 nm) lamellae were prepared at those positions. Finally, lamellae were transferred to a cryo-transmission electron microscope, where they were imaged at high magnification. Wild-type TDP-25 inclusions appeared amorphous and seemingly lacked fibrillar structure, although they were clearly demarcated within the cytoplasm (Fig 1C and D, Movie EV1). These findings are in stark contrast to previously investigated poly-Q, poly-GA, and α-synuclein neuronal inclusions, which show amyloid-like conformation in neurons in our cryo-ET pipeline (Bauerlein et al, 2017; Guo et al, 2018; Trinkaus et al, 2021). Introducing the eight ALS-causing mutations had no obvious effect on the morphology of TDP-25 inclusion (Fig EV1A). Click here to expand this figure. Figure EV1. Proteasomes are enriched in TDP-25 inclusions Tomographic slice of an aggregate within a neuron transduced with GFP-TDP-25 mutant construct (DIV5+8). Red dotted line segments aggregate area. Scale bar = 200 nm. Workflow of subtomogram averaging and classification. Subtomograms were identified using a low-resolution single-capped proteasome as template. All proteasomes were firstly classified into single-capped or double-capped. To further analyze the conformation of the regulatory particles, all proteasomes were cut in silico between the β-rings of the core particle, resulting in two independent particles for double-capped ones. 50% of the particles were assigned to the single and double-capped classes, respectively. Cut out regulatory particles were merged and subjected to a further round of classification. All regulatory particles were assigned to substrate processing conformations. However, due to the uncertainties inherent to the classification procedure, it is possible that a small fraction of particles adopted other conformations. Gold-standard Fourier shell correlation curve of the proteasome structure showing a resolution of 20 Å. Molecular mapping in three tomograms of GFP-TDP-25 wild-type inclusions in transduced neurons (DIV5+8). Regions containing GFP-TDP-25 are outlined in red. For the whole tomogram, proteasomes (purple), TRiC (cyan) and ribosomes (yellow) are mapped to their original positions and orientations using the information from template matching and subtomogram averaging. The numbers of proteasomes and ribosomes detected in the tomograms are plotted versus the volume of the tomogram. Scale bar = 200 nm. Molecular mapping in a tomogram recorded on a non-transduced control neuron, with proteasomes (purple), ribosomes (yellow) and TRiC (cyan) plotted as in Fig EV1D. This tomogram contains 24 proteasomes, 1,222 ribosomes and 73 TRiC molecules. Scale bar = 200 nm. Download figure Download PowerPoint Similar to our findings in C9orf72 poly-GA inclusions (Guo et al, 2018), we detected ring-like structures (Fig 1C) accumulating within TDP-25 inclusions. Sub-tomogram averaging of these objects (Fig EV1B) converged to a proteasome structure at ~20 Å resolution (Fig 1E, Fig EV1C). Interestingly, an extra density that was not accounted for by proteasomal subunits was present on the proteasome regulatory particle, possibly reflecting substrates or adaptor proteins (Fig 1E). Whereas ribosomes were largely excluded from TDP-25 inclusions, we found an approximately eightfold enrichment of proteasomes compared with proteasome concentration in control neurons (Asano et al, 2015; Guo et al, 2018; Figs 1D and EV1D and E). Strikingly, classification based on the conformation of 19S regulatory particles revealed that virtually all proteasome particles within the inclusions were in substrate processing states (Fig 1E and F). In comparison, only 20 and 37% of proteasomes were in substrate processing states in control neurons and poly-GA inclusions, respectively (Asano et al, 2015; Guo et al, 2018; Fig 1F). It should be noted that, due to the uncertainties inherent to the classification procedure, we cannot rule out that a small fraction of particles adopted other conformations. The little to no detectable amount of ground state proteasomes suggests that proteasomes inside TDP-25 inclusions are stalled, as proteasome function requires cyclic transition through activated and ground states (Collins & Goldberg, 2017). Amorphous TDP-25 inclusions have gel-like properties Given the amorphous appearance of TDP-25 inclusions, we asked whether they may represent phase-separated liquid droplets. Thus, we analyzed the mobility of GFP-TDP-25 in neuronal inclusions using fluorescence recovery after photobleaching (FRAP) in comparison with known liquid and solid reference proteins, that is, nucleolar NPM1 and inclusion-forming poly-Q, respectively (Bauerlein et al, 2017; Frottin et al, 2019). In contrast to TDP-25, GFP-NPM1 fluorescence recovered within seconds after bleaching, consistent with high mobility of the protein within the liquid-like nucleolus (Fig 2A and B). This clearly argues against a liquid droplet character of TDP-25 inclusions. Some neurons showed diffuse cytoplasmic TDP-25 expression without inclusion formation. Photobleaching of such diffuse TDP-25 resulted in very quick recovery suggesting that inclusion formation greatly reduces mobility (Fig EV2). Moreover, nuclear full-length TDP-43 showed a much higher mobile fraction than TDP-25 inclusions (Fig EV2), in line with its good solubility (Fig 1B). However, compared with fibrillar poly-Q (huntingtin exon 1 containing 97 glutamines fused to GFP; Htt97Q-GFP), a classical amyloid previously shown to interact with ER membranes (Bauerlein et al, 2020), TDP-25 mobility was much higher (~10% vs. ~25% recovery at 22 min, Fig 2C and D). Figure 2. TDP-25 inclusions are neither liquid-like nor solid-like A–D. Primary rat hippocampal neurons were transduced with GFP-TDP-25 variants, GFP-NPM1 or Htt97Q-GFP and analyzed by fluorescence recovery after photoble