Elicit: TDP-43 Biological Pathways (Public)

Skip to main content

TDP-43 Biological Pathways (Public)

Research reportView only

Create alertChat

March 2, 2025

What biological pathways is TDP-43 known to be part of?

TDP-43 functions in seven major biological pathways: RNA metabolism, nucleocytoplasmic transport, stress response, mitochondrial function, synaptic function, inflammatory responses, and cell cycle regulation.

Abstract

Forty studies report that TDP-43 participates in a network of biological pathways. In several papers, TDP-43 regulates RNA metabolism by controlling alternative splicing, cryptic exon inclusion, and mRNA stability. Other studies document that TDP-43 affects nucleocytoplasmic transport by disrupting nuclear pore complex integrity and altering protein–RNA distribution. Reports also describe its role in stress response and protein quality control through modulation of stress granule dynamics, phase separation, and aggresome formation. In addition, TDP-43 has been linked to mitochondrial function via binding mitochondrial mRNAs and disrupting respiratory complex assembly, as well as to synaptic function and neurotransmitter regulation through effects on receptor recycling and synaptic protein expression. Some papers further implicate the protein in inflammatory and immune responses via activation of NF‑κB and inflammasome pathways, and in cell cycle regulation and DNA repair by modulating targets such as cyclin-dependent kinase expression and double-strand break repair.

Seven primary pathways emerge from these studies: 1. RNA metabolism and processing 2. Nucleocytoplasmic transport 3. Stress response and protein quality control 4. Mitochondrial function 5. Synaptic function and neurotransmitter regulation 6. Inflammatory and immune responses 7. Cell cycle regulation and DNA repair

Diverse experimental systems—including mouse models, cultured cells, and iPSC-derived neurons—support TDP-43’s multifaceted functions across these interrelated pathways.

Methods

We analyzed 40 sources from an initial pool of 493, using 6 screening criteria. Each paper was reviewed for 3 key aspects that mattered most to the research question. More on methods

Papers identified with Elicit search

n = 493

Papers screened using: Molecular Pathway Analysis, TDP-43 Function, Experimental Design, Study Type, Mechanistic Detail, Evidence Quality

n = 493

Papers screened out

n = 453

Papers included for extraction

n = 40

Press enter or space to select a node.You can then use the arrow keys to move the node around. Press delete to remove it and escape to cancel.

Press enter or space to select an edge. You can then press delete to remove it or escape to cancel.

Paper search

Using your research question “What biological pathways is TDP-43 known to be part of?”, we searched across over 126 million academic papers from the Semantic Scholar corpus. We retrieved the 493 papers most relevant to the query.

Screening

We screened in sources based on their abstracts that met these criteria:

We considered all screening questions together and made a holistic judgement about whether to screen in each paper.

Data extraction

We asked a large language model to extract each data column below from each paper. We gave the model the extraction instructions shown below for each column.

Extract all explicitly mentioned biological pathways or cellular processes that TDP-43 is involved in. Look in the methods, results, and discussion sections.

Specific areas to focus on:

If multiple pathways are mentioned, list them all. If a pathway is described but not explicitly named, summarize the description.

If no clear pathways are identified, write “No specific pathways explicitly described.”

Format examples:

Identify and list all molecular interactions and binding partners mentioned in the study.

Search in:

Extract:

If multiple interactions are found, list all. If no interactions are described, write “No molecular interactions reported.”

Format examples:

Extract specific functional outcomes or consequences resulting from TDP-43 interactions or modifications.

Look for:

Prioritize direct functional descriptions, not just descriptive observations.

If multiple consequences are found, list all. If no clear functional outcomes are described, write “No specific functional consequences reported.”

Format examples:

Results

Characteristics of Included Studies

Study

Study Design

Cellular/Model System

Pathways Investigated

Key Findings

Full text retrieved

“TDP‐43 Loss of Function,” 2016

In vitro study

Neuronal cell culture

Endosomal trafficking

The study reported that TDP-43 regulates endosomal trafficking and receptor recycling

No

Afroz et al., 2023

In vivo mouse model and in vitro studies

Mouse models, Amyotrophic Lateral Sclerosis (ALS) patient-derived microglia

Immune response, phagocytosis

The study reported that targeting TDP-43 C-terminal domain reduces pathology and enhances microglial function

No

Ayala et al., 2008

In vitro study

Human cells

Cell cycle regulation

The study reported that TDP-43 regulates Cyclin-dependent kinase 6 (Cdk6) expression and cell cycle progression

Yes

Chen, 2020

Review

Various

RNA metabolism, autophagy

The review reported that TDP-43 regulates cryptic exon splicing and autophagy pathways

No

Chou et al., 2017

In vitro and in vivo studies

Mouse primary neurons, human fibroblasts, induced pluripotent stem cell (iPSC)-derived neurons

Nucleocytoplasmic transport

The study reported that TDP-43 aggregates disrupt nuclear pore complexes and nucleocytoplasmic transport

Yes

Chou et al., 2018

In vitro and in vivo studies

Mouse primary neurons, human fibroblasts, iPSC-derived neurons

Nucleocytoplasmic transport

The study reported that TDP-43 pathology impairs nuclear protein import and RNA export

Yes

Deshaies et al., 2018

In vitro and in vivo studies

Human cells, mouse models

RNA splicing

The study reported that TDP-43 regulates alternative splicing of heterogeneous nuclear ribonucleoprotein A1 (hnRNP A1)

Yes

Fazal et al., 2021

iPSC-derived motor neuron study

Human iPSC-derived motor neurons

RNA metabolism, axonal transport

The study reported that TDP-43 mutations affect RNA processing and axonal transport

Yes

Feneberg et al., 2020

In vitro and in vivo studies

Mouse primary motor neurons

Protein-protein interactions, stress response

The study reported that TDP-43 interacts with proteins involved in stress granule formation and endosomal-extracellular transport

No

Feneberga et al., “Oxidative Stress in Motor Neurons”

In vitro and in vivo studies

Mouse primary motor neurons

Protein-protein interactions, stress response

The study reported that TDP-43 mutations affect stress granule formation and extracellular vesicle secretion

No

toof

Pageof

View 30 more rows

Our analysis of the 40 studies on TDP-43 pathways reveals:

The diversity of cellular models and pathways investigated highlights the multifaceted nature of TDP-43’s functions and its potential involvement in various aspects of neurodegeneration.

Based on our analysis, we identified seven primary pathways implicated in the cellular mechanisms of TDP-43:

  1. RNA metabolism and processing
  2. Nucleocytoplasmic transport
  3. Stress response and protein quality control
  4. Mitochondrial function
  5. Synaptic function and neurotransmitter regulation
  6. Inflammatory and immune responses
  7. Cell cycle regulation and DNA repair

These diverse pathways highlight the multifaceted role of TDP-43 in cellular function and its potential impact on various aspects of neurodegeneration.

Thematic Analysis

RNA Processing and Regulation Pathways

Study

RNA Processing Function

Specific Targets/Mechanisms

Cellular Impact

Chen, 2020

Cryptic exon splicing regulation

Repression of cryptic exons

Maintenance of normal transcriptome

Deshaies et al., 2018

Alternative splicing regulation

hnRNP A1 pre-mRNA

Production of aggregation-prone hnRNP A1B isoform

Fazal et al., 2021

RNA processing, splicing regulation

Various mRNA targets

Altered gene expression in motor neurons

Gu et al., 2019

mRNA processing

Tau mRNA

Regulation of tau mRNA stability and exon 10 inclusion

Igaz et al., 2009

RNA splicing

No mention found

The study reported abnormal RNA splicing caused by TDP-43 C-terminal fragments

Kim et al., 2010

mRNA regulation

HDAC6 mRNA

Reduced expression of HDAC6

Koehler et al., 2022

Autoregulation, RNA binding

TDP-43 mRNA

Maintenance of TDP-43 protein levels

Lauranzano et al., 2015

RNA metabolism

TARDBP RNA targets

Regulation of TDP-43 target gene expression

Ling et al., 2010

RNA processing

No mention found

The study reported enhanced interaction with FUS/TLS affecting RNA processing

Long et al., 2024

miRNA biogenesis

miRNAs

Disruption of miRNA processing through Dicer interaction

toof

Pageof

View 6 more rows

Our analysis of the included studies suggests that TDP-43 is involved in various RNA processing functions:

These findings suggest that TDP-43 plays a multifaceted role in RNA processing, potentially affecting various cellular processes through its interactions with different RNA targets.

Stress Response and Protein Quality Control Pathways

Study

Stress Response Mechanism

Protein Quality Control Function

Cellular Impact

Chen, 2020

Autophagy regulation

Regulation of ATG7 and other autophagy-related genes

Maintenance of cellular protein homeostasis

Chou et al., 2017

Stress granule dynamics

Interaction with nuclear pore complexes

Altered nucleocytoplasmic transport under stress

Feneberg et al., 2020

Stress granule formation

Interaction with stress granule proteins

Altered stress response under oxidative conditions

Feneberga et al., “Oxidative Stress in Motor Neurons”

Stress granule formation

Extracellular vesicle secretion

Impaired stress response and intercellular communication

Herzog et al., 2019

CREB signaling regulation

No mention found

Altered gene expression under stress conditions

Koehler et al., 2022

Phase separation

Autoregulation of TDP-43 levels

Maintenance of TDP-43 protein homeostasis

Luan et al., 2023

Integrated stress response activation

Regulation of stress-related gene expression

Enhanced TDP-43 aggregation and stress granule formation

Mann et al., 2019

Phase transitions

RNA binding-mediated regulation

Prevention of neurotoxic TDP-43 aggregation

McGurk et al., 2018

Stress granule localization

Interaction with poly(ADP-ribose)

Promotion of TDP-43 phase separation

Sidibé et al., 2020

Stress granule dynamics

G3BP1 mRNA stabilization

Regulation of stress granule assembly

toof

Pageof

View 3 more rows

Our analysis of the table reveals several key findings regarding TDP-43’s role in cellular stress responses:

These findings highlight the complex and multifaceted role of TDP-43 in cellular stress responses, with a particular emphasis on stress granule dynamics and phase separation processes.

Synaptic Function and Neurotransmitter Pathways

Study

Synaptic Function

Neurotransmitter Regulation

Cellular Impact

“TDP‐43 Loss of Function,” 2016

Regulation of receptor recycling

Affects ErbB4 surface delivery

Impaired trophic signaling

Heyburn and Moussa, 2016

Regulation of synaptic protein expression

Affects vesicular glutamate levels

Altered neurotransmitter release

Sephton et al., 2010

Binding to RNAs involved in synaptic function

No mention found

Regulation of synaptic gene expression

Herzog et al., 2019

Regulation of dendritic complexity

No mention found

Altered neuronal morphology and potential synaptic changes

Fazal et al., 2021

Affects axonal transport

No mention found

Potential impact on synaptic function and neurotransmitter release

toof

Pageof

Our analysis of the table reveals diverse effects of TDP-43 on synaptic function, neurotransmitter regulation, and cellular impact across the five studies:

The diversity of findings suggests that TDP-43 may have wide-ranging effects on neuronal function, affecting various aspects of synaptic activity, neurotransmitter regulation, and cellular processes. However, the lack of consistent findings across studies indicates a need for further research to establish more definitive patterns of TDP-43’s impact on neuronal function.

Pathway Interactions and Integration

Primary Pathway

Connected Pathways

Regulatory Points

Cellular Outcomes

RNA Processing

Stress Response, Synaptic Function

Splicing regulation, mRNA stability

Altered gene expression, protein homeostasis

Nucleocytoplasmic Transport

RNA Processing, Stress Response

Nuclear pore complex interactions

Disrupted RNA/protein localization, stress granule formation

Stress Response

RNA Processing, Protein Quality Control

Phase separation, stress granule dynamics

Altered stress granule assembly, protein aggregation

Mitochondrial Function

Stress Response, DNA Repair

Mitochondrial mRNA binding, unfolded protein response

Respiratory complex disruption, oxidative stress

Synaptic Function

RNA Processing, Protein Quality Control

Receptor recycling, synaptic protein expression

Altered neurotransmission, synaptic plasticity

Inflammatory Response

Stress Response, Mitochondrial Function

NF-κB and NLRP3 inflammasome activation

Microglial activation, neuroinflammation

DNA Repair

RNA Processing, Stress Response

Double-Strand Break (DSB) repair, R-loop metabolism

Genome stability, cellular survival

toof

Pageof

Our analysis of the pathway interactions and integration reveals:

The studies we analyzed suggest interconnections between multiple pathways involving TDP-43, potentially indicating its multifaceted cellular functions. These interconnections may help explain the widespread cellular dysfunction observed in TDP-43-associated neurodegenerative diseases, as disruptions in one pathway could potentially have cascading effects on others.

References

Liam Chen\ (2020).The important functional role of TDP-43 plays in amyotrophic lateral sclerosis-frontotemporal dementia. Neural Regeneration Research

L. Heyburn, C. Moussa\ (2016).TDP-43 overexpression impairs presynaptic integrity. Neural Regeneration Research

Ching-Chieh Chou, Y. Zhang, Mfon E Umoh, Spencer Vaughan, Ileana Lorenzini, and 20 more\ (2017).TDP-43 pathology disrupts nuclear pore complexes and nucleocytoplasmic transport in ALS/FTD. Alzheimer's & Dementia

Wenzhang Wang, Luwen Wang, Junjie Lu, S. Siedlak, H. Fujioka, and 10 more\ (2016).The Inhibition of TDP-43 Mitochondrial Localization Blocks Its Neuronal Toxicity. Nature Network Boston

R. Mutihac, R. Mutihac, J. Alegre-Abarrategui, D. Gordon, Lucy Farrimond, and 3 more\ (2015).\ TARDBP pathogenic mutations increase cytoplasmic translocation of TDP-43 and cause reduction of endoplasmic reticulum Ca2+ signaling in motor neurons. Neurobiology of Disease

Weihua Zhao, D. Beers, S. Bell, Jinghong Wang, Shixiang Wen, and 2 more\ (2015).TDP-43 activates microglia through NF-κB and NLRP3 inflammasome. Experimental Neurology

L. Igaz, L. Kwong, A. Chen-Plotkin, M. Winton, T. Unger, and 4 more\ (2009).Expression of TDP-43 C-terminal Fragments in Vitro Recapitulates Pathological Features of TDP-43 Proteinopathies*. Journal of Biological Chemistry

Chien-Hsiung Yu, Sophia Davidson, C. Harapas, J. B. Hilton, Michael J. Mlodzianoski, and 21 more\ (2020).TDP-43 Triggers Mitochondrial DNA Release via mPTP to Activate cGAS/STING in ALS. Cell

C. Sephton, Can Cenik, Alper Kucukural, E. Dammer, B. Cenik, and 7 more\ (2010).Identification of Neuronal RNA Targets of TDP-43-containing Ribonucleoprotein Complexes. Journal of Biological Chemistry

J. Mann, A. Gleixner, J. Mauna, Edward Gomes, M. DeChellis-Marks, and 14 more\ (2019).RNA Binding Antagonizes Neurotoxic Phase Transitions of TDP-43. Neuron

Shuo-Chien Ling, Claudio P. Albuquerque, Joo Seok Han, C. Lagier-Tourenne, Seiya Tokunaga, and 2 more\ (2010).ALS-associated mutations in TDP-43 increase its stability and promote TDP-43 complexes with FUS/TLS. Proceedings of the National Academy of Sciences of the United States of America

No authors found (2016).TDP‐43 loss of function inhibits endosomal trafficking and alters trophic signaling in neurons. EMBO Journal

Sang Hwa Kim, Naval P. Shanware, Michael J Bowler, R. Tibbetts\ (2010).Amyotrophic Lateral Sclerosis-associated Proteins TDP-43 and FUS/TLS Function in a Common Biochemical Complex to Co-regulate HDAC6 mRNA*. Journal of Biological Chemistry

Wei Luan, A. Wright, H. Brown-Wright, Sheng Le, Rebecca San Gil, and 5 more\ (2023).Early activation of cellular stress and death pathways caused by cytoplasmic TDP-43 in the rNLS8 mouse model of ALS and FTD. Molecular Psychiatry

Xing Sun, Yongjia Duan, Caixia Qin, Jian-Chiuan Li, Gang Duan, and 8 more\ (2018).Distinct multilevel misregulations of Parkin and PINK1 revealed in cell and animal models of TDP-43 proteinopathy. Cell Death and Disease

L. McGurk, Edward Gomes, Lin Guo, J. Mojsilovic-Petrovic, V. Tran, and 3 more\ (2018).Poly(ADP-Ribose) Prevents Pathological Phase Separation of TDP-43 by Promoting Liquid Demixing and Stress Granule Localization. Molecules and Cells

Josiah J. Herzog, Weijin Xu, Mugdha Deshpande, Reazur Rahman, H. Suib, and 3 more\ (2019).TDP-43 dysfunction restricts dendritic complexity by inhibiting CREB activation and altering gene expression. Proceedings of the National Academy of Sciences of the United States of America

Hadjara Sidibé, Yousra Khalfallah, S. Xiao, N. Gómez, Hana Fakim, and 15 more\ (2021).TDP-43 stabilizes G3BP1 mRNA: relevance to amyotrophic lateral sclerosis/frontotemporal dementia. Brain : a journal of neurology

E. Feneberg, D. Gordon, A. Thompson, M. Finelli, R. Dafinca, and 9 more\ (2020).An ALS-linked mutation in TDP-43 disrupts normal protein interactions in the motor neuron response to oxidative stress. Neurobiology of Disease

María M. Leal‐Lasarte, Jaime M Franco, A. Labrador-Garrido, D. Pozo, C. Roodveldt\ (2017).Extracellular TDP‐43 aggregates target MAPK/MAK/MRK overlapping kinase (MOK) and trigger caspase‐3/IL‐18 signaling in microglia. The FASEB Journal

Xiang Long, Mengni Jiang, Yongzhen Miao, Huanhuan Du, Ting Zhang, and 4 more\ (2024).TDP-43 mutations-induced defects in miRNA biogenesis and cytotoxicity by differentially obstructing Dicer activity in Drosophila and in vitro. bioRxiv

Y. Ayala, T. Misteli, F. Baralle\ (2008).TDP-43 regulates retinoblastoma protein phosphorylation through the repression of cyclin-dependent kinase 6 expression. Proceedings of the National Academy of Sciences of the United States of America

G. P. Ritson, Sara K Custer, Brian D. Freibaum, Jake B. Guinto, D. Geffel, and 8 more\ (2010).TDP-43 Mediates Degeneration in a Novel Drosophila Model of Disease Caused by Mutations in VCP/p97. Journal of Neuroscience

Zhipeng Yu, D. Fan, B. Gui, Lei Shi, C. Xuan, and 4 more\ (2012).Neurodegeneration-associated TDP-43 Interacts with Fragile X Mental Retardation Protein (FMRP)/Staufen (STAU1) and Regulates SIRT1 Expression in Neuronal Cells*. Journal of Biological Chemistry

Emily Feneberga, David Gordona, Alexander G. Thompsona, Mattéa J. Finellia, Ana Candalijaa, and 8 more\ (2020).Neurobiology of Disease An ALS-linked mutation in TDP-43 disrupts normal protein interactions in the motor neuron response to oxidative stress

Peng Wang, Jianwen Deng, Jie Dong, Jianghong Liu, E. Bigio, and 10 more\ (2019).TDP-43 induces mitochondrial damage and activates the mitochondrial unfolded protein response. PLoS Genetics

Kaitlin Weskamp, Elizabeth H M Tank, R. Miguez, Jonathon P. McBride, N. Gómez, and 6 more\ (2019).Shortened TDP43 isoforms upregulated by neuronal hyperactivity drive TDP43 pathology in ALS. bioRxiv

Jade-Emmanuelle Deshaies, L. Shkreta, A. Moszczynski, Hadjara Sidibé, Sabrina Semmler, and 15 more\ (2018).TDP-43 regulates the alternative splicing of hnRNP A1 to yield an aggregation-prone variant in amyotrophic lateral sclerosis. Brain : a journal of neurology

Nicole F. Liachko, P. McMillan, T. Strovas, Elaine Loomis, Lynne Greenup, and 7 more\ (2014).The Tau Tubulin Kinases TTBK1/2 Promote Accumulation of Pathological TDP-43. PLoS Genetics

Hadjara Sidibé, Yousra Khalfallah, S. Xiao, N. Gómez, Elizabeth H M Tank, and 14 more\ (2020).TDP-43 stabilizes transcripts encoding stress granule protein G3BP1: potential relevance to ALS/FTD. bioRxiv

J. Mitra, Erika N. Guerrero, Erika N. Guerrero, Pavana M. Hegde, Nicole F. Liachko, and 15 more\ (2019).Motor neuron disease-associated loss of nuclear TDP-43 is linked to DNA double-strand break repair defects. Proceedings of the National Academy of Sciences of the United States of America

Giulia Romano, Raffaella Klima, F. Feiguin\ (2020).TDP-43 prevents retrotransposon activation in the Drosophila motor system through regulation of Dicer-2 activity. BMC Biology

Y. Ke, Annika van Hummel, Carol Au, Gabriella Chan, W. S. Lee, and 16 more\ (2024).Targeting 14-3-3θ-mediated TDP-43 pathology in amyotrophic lateral sclerosis and frontotemporal dementia mice. Neuron

Jianlan Gu, Dandan Chu, Nana Jin, Feng Chen, Fei Liu\ (2019).Cyclic AMP-Dependent Protein Kinase Phosphorylates TDP-43 and Modulates Its Function in Tau mRNA Processing. Journal of Alzheimer's Disease

Seiji Watanabe, Hidekazu Inami, K. Oiwa, Yuri Murata, Shohei Sakai, and 5 more\ (2020).Aggresome formation and liquid–liquid phase separation independently induce cytoplasmic aggregation of TAR DNA-binding protein 43. Cell Death and Disease

Ching-Chieh Chou, Y. Zhang, Mfon E Umoh, Spencer Vaughan, Ileana Lorenzini, and 20 more\ (2018).TDP-43 pathology disrupts nuclear pore complexes and nucleocytoplasmic transport in ALS/FTD. Nature Neuroscience

L. Koehler, Z. Grese, Alliny C S Bastos, Lohany D. Mamede, T. Heyduk, and 1 more\ (2022).TDP-43 Oligomerization and Phase Separation Properties Are Necessary for Autoregulation. Frontiers in Neuroscience

T. Afroz, Elodie Chevalier, M. Audrain, C. Dumayne, T. Ziehm, and 21 more\ (2023).Immunotherapy targeting the C-terminal domain of TDP-43 decreases neuropathology and confers neuroprotection in mouse models of ALS/FTD. Neurobiology of Disease

Raheem Fazal, Steven Boeynaems, A. Swijsen, Mathias De Decker, L. Fumagalli, and 14 more\ (2021).HDAC6 inhibition restores TDP‐43 pathology and axonal transport defects in human motor neurons with TARDBP mutations. EMBO Journal

E. Lauranzano, S. Pozzi, L. Pasetto, Riccardo Stucchi, Tania Massignan, and 8 more\ (2015).Peptidylprolyl isomerase A governs TARDBP function and assembly in heterogeneous nuclear ribonucleoprotein complexes. Brain : a journal of neurology

Download BIBDownload RISDownload TXT

Report

Status

Gather sources

493 sources found

Details

Screen sources

40 sources included

Details

Extract data

120 data points extracted

Details

Generate report

Save PDF

BIBLaTeX, ZoteroRISZotero, MendeleyTXTAPA BibliographyPDFPDFDOCXMicrosoft Word

Chat

Got some follow-up questions?

Sign up or sign in to chat with this report.

Back

TDP-43 Oligomerization and Phase Separation Properties Are Necessary for Autoregulation

L. Koehler, Z. Grese, Alliny C S Bastos, Lohany D. Mamede, T. Heyduk, Yuna M. Ayala

Frontiers in Neuroscience·

2022·

27 citations

SourceDOI

Plain textPDFFigures

Searching for PDF

Unable to find PDF from source

Back

Specific Biological Pathways Involving TDP-43

RNA splicing regulation, R-loop metabolism regulation, Autoregulation through negative feedback loop involving mRNA binding

Molecular Interactions and Binding Partners of TDP-43

Binds to its own mRNA 3′ UTR, specifically the CLIP34 region

Functional Consequences of TDP-43 Interactions

Disrupts TDP-43 autoregulation, leading to increased aggregation and neurotoxicity; Decreases cellular clearance of TDP-43 and other RNA binding proteins; Alters phase separation properties, affecting gene expression regulation; Causes increased accumulation of endogenous TDP-43 protein; Leads to neuronal toxicity in ALS.

Loss of TDP-43 protein homeostasis and dysfunction, in particular TDP-43 aggregation, are tied to amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD). TDP-43 is an RNA binding protein tightly controlling its own expression levels through a negative feedback loop, involving TDP-43 recruitment to the 3′ untranslated region of its own transcript. Aberrant TDP-43 expression caused by autoregulation defects are linked to TDP-43 pathology. Therefore, interactions between TDP-43 and its own transcript are crucial to prevent TDP-43 aggregation and loss of function. However, the mechanisms that mediate this interaction remain ill-defined. We find that a central RNA sequence in the 3′ UTR, which mediates TDP-43 autoregulation, increases the liquid properties of TDP-43 phase separation. Furthermore, binding to this RNA sequence induces TDP-43 condensation in human cell lysates, suggesting that this interaction promotes TDP-43 self-assembly into dynamic ribonucleoprotein granules. In agreement with these findings, our experiments show that TDP-43 oligomerization and phase separation, mediated by the amino and carboxy-terminal domains, respectively, are essential for TDP-43 autoregulation. According to our additional observations, CLIP34-associated phase separation and autoregulation may be efficiently controlled by phosphorylation of the N-terminal domain. Importantly, we find that specific ALS-associated TDP-43 mutations, mainly M337V, and a shortened TDP-43 isoform recently tied to motor neuron toxicity in ALS, disrupt the liquid properties of TDP-43-RNA condensates as well as autoregulatory function. In addition, we find that M337V decreases the cellular clearance of TDP-43 and other RNA binding proteins associated with ALS/FTD. These observations suggest that loss of liquid properties in M337V condensates strongly affects protein homeostasis. Together, this work provides evidence for the central role of TDP-43 oligomerization and liquid-liquid phase separation linked to RNA binding in autoregulation. These mechanisms may be impaired by TDP-43 disease variants and controlled by specific cellular signaling.

Loss of TDP-43 protein homeostasis and dysfunction, in particular TDP-43 aggregation, are tied to amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD). TDP-43 is an RNA binding protein tightly controlling its own expression levels through a negative feedback loop, involving TDP-43 recruitment to the 3 untranslated region of its own transcript. Aberrant TDP-43 expression caused by autoregulation defects are linked to TDP-43 pathology. Therefore, interactions between TDP-43 and its own transcript are crucial to prevent TDP-43 aggregation and loss of function. However, the mechanisms that mediate this interaction remain ill-defined. We find that a central RNA sequence in the 3 UTR, which mediates TDP-43 autoregulation, increases the liquid properties of TDP-43 phase separation. Furthermore, binding to this RNA sequence induces TDP-43 condensation in human cell lysates, suggesting that this interaction promotes TDP-43 self-assembly into dynamic ribonucleoprotein granules. In agreement with these findings, our experiments show that TDP-43 oligomerization and phase separation, mediated by the amino and carboxy-terminal domains, respectively, are essential for TDP-43 autoregulation. According to our additional observations, CLIP34-associated phase separation and autoregulation may be efficiently controlled by phosphorylation of the N-terminal domain. Importantly, we find that specific ALS-associated TDP-43 mutations, mainly M337V, and a shortened TDP-43 isoform recently tied to motor neuron toxicity in ALS, disrupt the liquid properties of TDP-43-RNA condensates as well as autoregulatory function. In addition, we find that M337V decreases the cellular clearance of TDP-43 and other RNA binding proteins associated with ALS/FTD. These observations suggest that loss of liquid properties in M337V condensates strongly affects protein homeostasis. Together, this work provides evidence for the central role of TDP-43 oligomerization and liquid-liquid phase separation linked to RNA binding in autoregulation. These mechanisms may be impaired by TDP-43 disease variants and controlled by specific cellular signaling.

INTRODUCTION

Aggregation, cellular mislocalization and loss of nuclear TDP-43 (TAR DNA binding protein) are hallmarks of amyotrophic lateral sclerosis (ALS), frontotemporal dementia (FTD) (Arai et al., 2006;Neumann et al., 2006) and limbic-predominant age-related TDP-43 encephalopathy (LATE) (Nelson et al., 2019). In addition, TDP-43 inclusions are associated with multisystem proteinopathy (MSP) (Weihl et al., 2009) and other neurodegenerative disorders, including Alzheimer's disease (AD) and chronic traumatic encephalopathy (CTE) (McKee et al., 2010). TDP-43 is an RNA binding protein whose cellular levels are tightly controlled as its overexpression leads to increased cytoplasmic accumulation and aggregation. In addition, even moderate changes in TDP-43 protein expression disrupt the regulation of target genes (Arnold et al., 2013). Abnormal increase in TDP-43 levels results in neurotoxicity as observed in human neurons (Barmada et al., 2010) and in a wide range of animal models, including non-human primates (Tatom et al., 2009;Wils et al., 2010;Xu et al., 2010;Estes et al., 2011;Uchida et al., 2012;Diaper et al., 2013). So far, TDP-43 autoregulation through a negative feedback loop is the only known mechanism controlling TDP-43 expression levels. During this process, TDP-43 binds to its own mRNA 3 UTR and inhibits protein synthesis (Ayala et al., 2011;Polymenidou et al., 2011; Figure 1A ). This autoregulatory mechanism, reviewed in Tziortzouda et al. (2021), is conserved among vertebrates and its disruption in mouse models results in widespread splicing changes, neurotoxicity (Xu et al., 2010;Arnold et al., 2013;Sugai et al., 2019) and increased cytoplasmic TDP-43 aggregation (D'Alton et al., 2015;Koyama et al., 2016). The existence of mutations that are causative of ALS in noncoding regions involved in autoregulation (Gitcho et al., 2009;Pesiridis et al., 2009) and the finding that one of these mutations shows increased TDP-43 transcript expression (Gitcho et al., 2009), strongly suggest that defects in self-regulation may result in disease. This is also supported by a knock-in mouse model in which the Q331K ALS mutation exhibits cognitive impairment and disrupts autoregulation (White et al., 2018). Furthermore, accumulation of misfolded TDP-43 aggregates potentially increases protein production in affected cells by triggering a feed-forward mechanism, in which TDP-43 is sequestered and is no longer able to self-regulate, as suggested by recent studies (Sugai et al., 2019). In agreement with this model, ALS-affected motor neurons and FTD neurons that lack nuclear TDP-43 show impaired Tardbp autoregulation (Koyama et al., 2016;Liu et al., 2019). Collectively, these findings indicate that Tardbp may be one of the most critical TDP-43 targets to prevent the loss of protein homeostasis and neurotoxicity. Therefore, improved understanding of the factors controlling TDP-43 autoregulation are necessary to test this model and provide new avenues to maintain TDP-43 proteostasis. However, despite the central importance of this process in TDP-43 function and its link to disease, the mechanisms involved in TDP-43 autoregulation are ill-defined, and the factors controlling autoregulation are even less well understood.

The cellular organization of RNA binding proteins, which in turn impacts protein function, involves condensation into ribonucleoprotein (RNP) granules. This process is mediated by liquid-liquid phase separation (LLPS) into often dynamic, liquidlike complexes (Brangwynne et al., 2009;Wippich et al., 2013;Feric et al., 2016). The high concentration of the components while retaining dynamic properties within the droplets is fundamental for RNP granule function and organization in cells [reviewed in Alberti et al. (2019)]. Formation of RNP granules is mediated by multivalent interactions, including RNA binding and self-assembly through different protein domains, such as low complexity regions and oligomerization domains (Li et al., 2012;Lin et al., 2015;Zhang et al., 2015;Garcia-Jove Navarro et al., 2019). The liquid properties of these assemblies and their multivalent properties enable rapid regulation in response to cellular stimuli [reviewed in Shin and Brangwynne (2017); Wang et al. (2021)]. RNA is a principal component of RNP granules and recent studies, including our own, suggest that RNA binding modulates the LLPS properties of TDP-43 through specific interactions (Zhang et al., 2015;Garcia-Jove Navarro et al., 2019;Grese et al., 2021). However, how LLPS affects RNA processing following recruitment to the transcripts, and whether this is a gene-dependent mechanism is currently not known. Importantly, numerous studies also suggest that defects in LLPS are associated with disease, such as in the case of TDP-43, FUS (Fused in sarcoma) and other RNA binding proteins linked to neurodegeneration (Lin et al., 2015;Molliex et al., 2015;Patel et al., 2015). Specifically, conversion of the condensates into fibrils or complexes with solid-like properties is thought to cause accumulation of protein aggregates associated with pathology. Therefore, further insight on the regulation of TDP-43 LLPS properties and their involvement in controlling the expression of TDP-43 targets will help elucidate important pathogenic factors.

We sought to define the molecular mechanisms in TDP-43 autoregulation and determine the effect disease-associated conditions have in this process. Our results strongly suggest that autoregulation is mediated by TDP-43 self-assembly and LLPS upon RNA binding. In addition, specific TDP-43 binding to RNA sequences that mediate autoregulation promotes liquid properties of TDP-43 condensates or liquid droplets. In addition, we find that TDP-43 posttranslational modifications and diseaseassociated variants may alter autoregulation by modulating TDP-43 condensation. Together, these findings highlight the potential of developing therapies based on controlling pathways linked to TDP-43 autoregulation as well as specific TDP-43-RNA binding interactions.

MATERIALS AND METHODS

Reagents and Chemicals Were Purchased From Sigma-Aldrich Unless Otherwise Specified

Plasmid Construction

The HA tagged TDP-43 sequence was cloned into pcDNA5 TM /FRT/TO (Thermo Fisher) between BamHI and NotI restriction sites. Quikchange-Site directed mutagenesis protocol (Agilent) was performed to create single-site and multiple-site mutations. The CR deletion missing a.a. 316-346 was generated from gBlock sequence. pcDNA5 HA-mEGFP-TDP constructs were created by subcloning TDP-43 cDNA into the mEGFP-C1 vector (Addgene) between XhoI and HindIII. HA-mEGFP-TDP-43 was subcloned into pcDNA5 TM /FRT/TO using EcoRV and NotI. Oligonucleotides and gBlock sequence are listed in the Supplementary Table 1 .

Expression and Purification of Recombinant TDP-43

Recombinant TDP-43 expression and purification was carried out as previously described (Grese et al., 2021).

Liquid-Liquid Phase Separation, Turbidity Assays, and Fluorescence Recovery After Photobleaching

The protocols developed in our lab for these assays were recently described (Grese et al., 2021). Microscopy images were taken following equilibration of condensates onto glass coverslips, approximately 60 min after the starting point of the reaction. For turbidity assays, data points were taken as the absorbance at 600 nm λ at 25 min. Data points were then normalized using the equation Abs norm = (Abs -Abs blank )/(Abs max -Abs blank ), where

Abs max is the condition being normalized to 1. The RNA oligonucleotides used in these experiments are listed in Supplementary Table 3 . FRAP was performed as previously described (Grese et al., 2021).

Fluorescence Anisotropy

Purified TDP-43 WT and mutants previously purified were serially diluted in a 1:2 ratio in a final range of 0-2 µM protein into a 300 mM NaCl, 10 mM Tris (pH 8.0), 5% glycerol, 5% sucrose, 0.5 mM TCEP buffer solution.

Cycloheximide Post-Chase Experiment

WT and M337V HA-TDP stable cells were plated and induced with 1 µg/mL of tetracycline for 72 h to reach confluency of 85-90%. Media was then changed to DMEM, 10% FBS, 100 µg/mL cycloheximide. Samples were taken at 0, 4, 8, 16, 32, and 40 h. Each sample was washed once with PBS and collected by centrifugation. Cell pellets were immediately lysed in cell lysis buffer and sonicated before being prepared for immunoblotting.

Cell Lysate RIPA/Urea Fractionation

WT and M337V stable cell lines were plated and induced for 72 h before being harvested by centrifugation. Cell pellets were then resuspended in RIPA Buffer [50 mM Tris pH 8.0, 150 mM NaCl, 1% NP-40, 5 mM EDTA, 0.5% SOC, 0.1% SDS, 1x protease/phosphatase inhibitor (SigmaFAST EDTA-free protease inhibitor cocktail tablet)] and sonicated. Total fraction samples were taken after sonication and the samples were then centrifuged at 98,400 g for 30 min at 4 • C. The pellet was then rinsed with RIPA buffer and centrifuged again. The resulting pellet was resuspended in urea buffer (7 M urea, 2 M thiourea, 4% CHAPS, 30 mM Tris pH 8.5). Fractions were quantified by western blot as described above.

Immunoblotting

Immunoblotting was carried out using standard western blotting techniques and the antibodies for TDP-43 (Proteintech, 10782-2-AP), FUS/TLS (Santa Cruz Biotechnologies, sc47711), matrin-3 (Abcam, ab151739), and GAPDH (Abcam, Cat. No. ab181602). Membranes were imaged using Odyssey scanning (LI-COR) and quantified using ImageStudioLite software (LI-COR).

RNA Extraction and Quantitative RT-PCR

Trizol Reagent (Thermo Fisher) was used for RNA extraction and was carried out according to manufacturer instructions. M-MLV reverse Transcriptase (Thermo Fisher) was used for cDNA production according to the manufacturer instructions.

Each primer mentioned in Supplementary Table 2 was used at a final concentration of 0.4 µM for a CFX96 Real-Time PCR detection system.

RESULTS

Specific TDP-43 Binding to an RNA Sequence Linked to Autoregulation Increases the Liquid Properties of TDP-43 Condensates TDP-43 binds to an extended region in the 3 UTR of its own transcript (Tardbp), spanning approximately 500 nucleotides, during autoregulation (Figure 1A ; Ayala et al., 2011;Polymenidou et al., 2011;Tollervey et al., 2011;Bhardwaj et al., 2013). This TDP-43 binding region (TDPBR) is immediately upstream of the proximal alternative polyadenylation site pA 1 and is included in an alternatively spliced intron (in7).

The mechanisms proposed to follow TDP-43 recruitment to TDPBR are based on analyses of the endogenous Tardbp transcript and derived gene reporters. These studies support the engagement of different processes during autoregulation: alternative polyadenylation, alternative splicing producing isoforms targeted for degradation via nonsense mediated decay (NMD), and nuclear retention of specific transcripts. Detailed studies by Koyama et al. (2016) suggest that autoregulation results from a combination of these mechanisms, but the extent to which each of these pathways contributes to the control of TDP-43 expression remains unknown (Polymenidou et al., 2011;Bembich et al., 2014;Koyama et al., 2016). A 34-nt. region within TDPBR (CLIP34) is commonly found in crosslinking immunoprecipitation (CLIP) analyses of TDP-43-bound transcripts in human cells and mouse brain (Polymenidou et al., 2011;Tollervey et al., 2011; Figures 1A, B ). This isolated sequence is capable of decreasing TDP-43 toxicity in neurons, suggesting that CLIP34 binding prevents TDP-43 aggregation and aberrant phase transitions (Mann et al., 2019). We recently reported that CLIP34 was among the GU-rich RNA sequences capable of modulating TDP-43 phase separation properties (Grese et al., 2021). In the present work, we sought to gain further insight of the mechanisms that mediate TDP-43 LLPS linked to CLIP34 and explore whether LLPS plays a role in TDP-43 autoregulatory function. For this, TDP-43 LLPS was analyzed using purified full-length TDP-43 and we observed a dramatic change in the behavior of TDP-43 condensates in the presence of CLIP34. At 150 mM NaCl in the absence of RNA, TDP-43 forms small condensates that fail to coalesce and cluster into chain-like structures (Figure 1C ). As previously shown, these rapidly obtain fibril-like properties that are more similar to gel or solid complexes instead of behaving as liquid droplets (French et al., 2019). Remarkably, addition of CLIP34 promoted TDP-43 condensate fusion and coalescence increasing droplet size. This was not observed in the presence of an RNA oligonucleotide control sequence of similar length, A(CA) 18 .

We measured the size of the condensates in the different conditions to quantify the changes. The condensates mediated by CLIP34 are more than twofold significantly larger than control conditions (Supplementary Figure 1 ). Addition of A(CA) 18 showed no significant difference relative to control. These observations are in agreement with recently published data from our lab, showing that CLIP34 promotes TDP-43 condensation at higher NaCl concentration (250 mM). Our findings suggest that at physiological salt conditions (150 mM NaCl) TDP-43 binding to CLIP34 specifically increases the liquid and dynamic properties of TDP-43 condensates. Fluorescence recovery after photobleaching (FRAP) of the TDP-43-CLIP34 condensates showed rapid recovery of Oregon green-labeled TDP-43 indicating a dynamic exchange between the light and dense phases. We observed approximately 40% recovery in TDP-43 condensates formed with specific-binding RNA (CLIP34), while condensates without RNA or in the presence of non-binding RNA showed virtually no recovery (Figures 1D, E ). This suggests that while a substantial immobile fraction is present in the TDP-43-CLIP34 sample, there is a difference between the dynamics of the specific vs. non-specific RNA conditions. The immobile fraction present in the CLIP34 condition could be explained by the tight binding between TDP-43 and CLIP34 RNA, as the recovery could be partially dependent on exchange of RNA-bound and free labeled TDP-43, and thereby the off-rate of the TDP-43-RNA interaction. TDP-43-CLIP34 binding affinity is in the nanomolar range suggesting that the off-rate may be slow and this could explain the slow fluorescence recovery. In addition, the amount of CLIP34 in our reaction may also affect the degree of recovery. High RNA concentrations in similar experiments causing FRAP values that do not recover to baseline levels have been previously observed (Zhang et al., 2015). Our results are further supported by videos of condensates that show the dynamic properties of these complexes. Here, we show the marked effect of CLIP34, which increased condensate fusion and coalescence, generating larger complexes compared to control and A(CA) 18 (Supplementary Movies 1-3). Collectively, our findings indicate that TDP-43 phase separation may be modulated by binding to CLIP34 specifically, preventing fibrillization of TDP-43 condensates.

To test whether condensation of CLIP34-bound TDP-43 may be observed in a cellular environment, we adapted a recently developed protocol to observe biomolecular condensates in human cellular lysate (Freibaum et al., 2021). We prepared cellular extract from human embryonic kidney cells (HEK293) cells stably expressing a single copy of monomeric mEGFP-TDP-43 wild-type (WT) (Figure 2A ). The lysate was mixed with purified TDP-43 after addition of RNA. At 250 mM NaCl, in the absence of RNA, the mixture appeared homogeneous and lacked visible foci (Figure 2B ). However, addition of CLIP34 readily promoted formation of mEGFP-TDP-43-positive granules. Control conditions in which no CLIP34 was added did not promote mEGFP-TDP-43 condensation. As a control, we tested the effect of A(CA) 18 RNA and observed a marked decrease in phase separation compared to CLIP34. The condensates observed with A(CA) 18 , which is not expected to bind TDP-43 specifically (Buratti and Baralle, 2001), may be due to interactions of cell lysate components with the oligonucleotide and TDP-43 through non-specific binding. Additionally, the higher concentration of TDP-43 and RNA that was required to generate condensate formation under these conditions could promote liquid-liquid phase separation through less specific interactions. We also found that, at lower salt concentrations (150 mM NaCl), purified TDP-43 and cell lysate resulted in the formation of the small condensates clustering together into fibrillike structures (Supplementary Figure 2 ). However, addition of CLIP34 generated larger round-shaped condensates, suggesting increased liquid properties consistent with our experiments using purified TDP-43 alone. This was not observed upon addition of A(CA) 18 RNA (Supplementary Figure 2B ). To further test the requirement of specific TDP-43 binding to CLIP34, we analyzed RNP granule formation of cellular lysate from HEK293 cells expressing an RNA binding-deficient mutant mEGFP-TDP-43 F4L (Phe147/149/229/231Leu) (Buratti and Baralle, 2001). Addition of CLIP34 to the F4L cellular lysate showed a decrease in condensate formation relative to WT (Figure 2B ). We also observed a dramatic decrease in the in the number and size of granules formed in F4L cell lysates mixed with A(CA) 18 , relative to WT. To further compare and quantify TDP-43 condensation in the various conditions, we measured lysate turbidity, as this often correlates with phase separation (Babinchak and Surewicz, 2020;Grese et al., 2021). Addition of CLIP34 increased the turbidity more than threefold compared to lysate or buffer only controls (Figure 2C ). Lysate turbidity upon mixing with CLIP34 was also significantly higher than A(CA) 18mixed lysate, consistent with our imaging data. Addition of CLIP34 to F4L cellular lysate showed 50% reduction in turbidity compared to WT. As the imaging studies showed, addition of A(CA) 18 to F4L lysate did not significantly increase turbidity compared to control. This data suggests that in the context of cellular lysate, the remaining TDP-43-RNA contacts of F4L are sufficient for condensate formation, albeit with lower efficiency relative to WT. Collectively, these results indicate that CLIP34 binding specifically increases the liquid properties of TDP-43 condensates and promotes phase separation of TDP-43 in celllike conditions.

TDP-43 Autoregulation Is Mediated by Oligomerization and Phase Separation

Based on our results showing modulation of TDP-43 phase separation by CLIP34, we next asked whether LLPS plays a role during TDP-43 autoregulation in human cells. For this, we introduced site directed mutations in different TDP-43 domains that disrupt self-assembly and phase separation (Figure 3A ). We targeted the N-terminal domain (NTD) with specific substitutions that inhibit oligomerization. Residue E17 of each NTD monomer makes multiple contacts upon dimerization, whereby the N-terminal domains are arranged in head-to-tail configuration (Afroz et al., 2017;Wang et al., 2018). Consequently, the E17R mutation greatly impairs NTD oligomerization. We also evaluated the role of S48 in the NTD, which undergoes phosphorylation in human cells (Rigbolt et al., 2011;Hornbeck et al., 2015;Wang et al., 2018). The phosphomimetic substitution S48E disrupts oligomerization and LLPS of full-length TDP-43 (Wang et al., 2018), suggesting that S48 phosphorylation modulates TDP-43 self-assembly. In agreement with these previous results, we observed fewer S48E condensates in the presence or absence of CLIP34 compared to WT or to the phospho-null substitution S48A (Figure 3A ). At the same time, S48A showed moderately decreased condensation compared to WT under control and CLIP34 conditions. We found that in stark contrast with WT TDP-43, E17R failed to phase separate and addition of CLIP34 was unable to promote condensation of this mutant (Figure 3B ). As further validation of the role of LLPS in CLIP34-mediated condensates, we deleted an evolutionarily conserved region in the C-terminal domain (a.a. 316-343) ( CR) that forms a partially α-helical structure, which is important for self-assembly and phase separation (Conicella et al., 2016(Conicella et al., , 2020;;Schmidt and Rohatgi, 2016). As seen with the NTD oligomerization mutants, CR inhibited LLPS in the presence or absence of RNA (Figure 3B ). To better compare the differences in LLPS among the variants, we quantified phase separation in the presence of CLIP34 by turbidity (Figure 3C ). Compared to WT, disruption of NTDmediated oligomerization by E17R and S48E showed greater than 80 and 70% significant reduction in turbidity, respectively. Consistent with the microscopy data, S48A reduced turbidity compared to WT, but not as much as S48E. The CR mutant showed greater than 80% significant reduction in turbidity. We examined whether the defects in CLIP34-driven LLPS were caused by decreased mutant binding to the RNA, quantifying the TDP-43-CLIP34 interactions by fluorescence anisotropy. We recently estimated that more than one TDP-43 molecule may associate with the CLIP34 sequence (Grese et al., 2021). However, further studies are necessary to determine the actual number of TDP-43 molecules binding CLIP34. Due to this limitation, we analyzed the apparent IC 50 (IC 50,app ) for the interaction between WT and mutant TDP-43 and CLIP34, as an estimate of the binding affinity. IC 50,app is the protein concentration in which 50% of the maximal anisotropy change is observed. Our results indicate that NTD mutations and CR do not significantly affect CLIP34 binding (Table 1 and Supplementary Figure 3 ). As expected, we could not determine the binding of F4L for CLIP34 as these mutations greatly disrupt sequence-specific contacts (Buratti and Baralle, 2001;Lukavsky et al., 2013). Based on our results, we propose that the mutants bind CLIP34 as WT TDP-43 but fail to form condensates due to impaired self-assembly.

To test the role of TDP-43 oligomerization and LLPS on autoregulation, we compared autoregulatory function in human cells expressing each of the LLPS-defective mutants. Stable HEK293 cells were generated expressing a single copy of the tagged TDP-43 transgene upon tetracycline induction, as previously described (Ayala et al., 2011). To determine autoregulatory activity of the mutants we measured protein and mRNA levels of endogenous TDP-43 in these cells at 72 h post transgene induction, which typically results in 70% reduction of endogenous TDP-43 levels in the case of WT (Figures 3D-H ).

In Figures 3F, H we report autoregulatory activity of the different mutants compared to WT at the protein and RNA levels 3F, H ), suggesting that LLPS is required for this function.

We found a dramatic disruption of autoregulation in E17R cells at the protein and mRNA levels, resulting in >90% loss in activity compared to WT (Figures 3F, H ). Similarly, S48E showed reduced autoregulatory function of approximately 50 and 70% compared to WT at protein and RNA levels, respectively. S48A function was not significantly different from WT, indicating that introduction of the negative charge at this site specifically disrupts autoregulation. In addition, these results suggest that the decrease in phase separation observed with S48A (Figures 3B, C ) is not sufficient to significantly disrupt autoregulation in the cellular assay. Instead, our findings suggest that TDP-43 phosphorylation at Ser48 may be able to modulate autoregulatory activity and control TDP-43 levels in cells. These results indicate a crucial role of NTD-directed self-assembly in autoregulation, which is particularly highlighted by the degree of functional loss in E17R. This mutation was even more disruptive than the RNA-binding deficient mutant F4L (Ayala et al., 2011), here used as control (Figures 3E-H ). Based on these observations, we propose that TDP-43 autoregulation requires TDP-43 selfassembly/oligomerization and condensate formation mediated by both N-and C-terminal domains. It is possible that these domains may also mediate heterotypic protein interactions that are important during autoregulation and this should be further investigated.

TDP-43 Disease-Associated Variants Show Defects in CLIP34 LLPS and Autoregulatory Function

We and others have shown that specific ALS-associated TDP-43 mutations alter the dynamic properties of TDP-43 condensates (Grese et al., 2021). Here, we investigated whether these TDP-43 mutations or variants linked to disease alter CLIP34-driven LLPS and consequently diminish autoregulatory function. We selected mutations positioned in the conserved region of the CTD (Figure 4A ; Conicella et al., 2020). Phase separation of purified TDP-43 mutants A321G, Q331K and M337V was analyzed upon mixing with CLIP34 and compared to A(CA) 18 or no RNA control (Figure 4B ). We found that as in the case of WT TDP-43, CLIP34 promoted Q331K condensate liquidity, shown by increased fusion and coalescence into larger droplets. In contrast, A321G and M337V remained in clusters or chainlike structures in the presence of CLIP34, suggesting that these mutations retain their gel/solid properties and do not acquire liquid properties upon binding to this RNA molecule. Addition of A(CA) 18 RNA did not significantly change the LLPS behavior of the mutants compared to control, as seen with WT TDP-43. Previous studies of the isolated CTD showed that these mutations disrupt LLPS through alterations in α-helix structure and helix-helix interactions (Conicella et al., 2016). Q331K and M337V similarly disrupt helix-helix contacts, while A321G shows helix-breaking behavior. However, we find that in the context of the full-length protein and RNA-driven condensation Q331K is distinct from M337V and A321G. The differences in the behavior of the mutants between these studies and the previous report (Conicella et al., 2016) may due to the use of the isolated CTD in previous work compared to the full-length protein in our studies, which specifically analyzed RNA binding-driven LLPS. The specific properties acquired by M337V and A321G, but not Q331K, should be further investigated as these may highlight important disease mechanisms. Next, we sought to determine whether a disease-associated short isoform of , recently characterized by the Barmada group (Weskamp et al., 2020), alters protein condensation associated to autoregulation. Alternative splicing of Tardbp produces sTDP-43, which lacks the CTD and is replaced by a unique 18 amino acid long sequence (Figure 4A ; Wang et al., 2004;D'Alton et al., 2015;Weskamp et al., 2020). This variant is toxic upon overexpression in neurons and is more highly expressed in ALS neurons and glia (Weskamp et al., 2020). We analyzed LLPS properties of sTDP-43 in the presence of CLIP34 and anticipated that loss of the entire CTD would strongly impair LLPS of sTDP-43, as in the case of C (a.a.1-269) which lacks the entire CTD (Figure 4A ) and prevented LLPS regardless of RNA presence, under the conditions tested (Figure 4B ).

In contrast, we found that the sTDP-43 variant spontaneously formed fibril-like structures (Figure 4B ). Addition of CLIP34 did not promote liquid behavior, but increased abundance of the fibril complexes. The presence of A(CA) 18 RNA did not significantly alter properties compared to control. The stark difference between sTDP-43, WT and C suggests that the short peptide at the C terminus dramatically alters the conformation of the rest of the protein by conferring aggregation-prone properties. Alternatively, the short tail, unique to sTDP-43, may drive intermolecular interactions that promote fibrilization. We then investigated whether changes in condensate behavior seen with the ALS mutations affect autoregulation. We created stable HEK293 cell lines expressing a single copy of each mutant transgene upon tetracycline induction as for Figures 3D-H . To compare autoregulatory function between WT and mutants we quantified endogenous TDP-43 protein and mRNA levels in uninduced cells and after 72 h of induction (Figures 5A-F ). We observed no significant differences in autoregulation in A321G and Q331K expressing cells relative to WT at protein and transcript levels (Figures 5B-F ), suggesting that the mutants were similarly capable of autoregulation as WT TDP-43. Our results regarding the Q331K mutation are in contrast with observations using a knock-in mouse model expressing this mutant which results in altered Tardbp mRNA processing and increased TDP-43 protein and transcript levels (White et al., 2018). However, our observations are supported by similar behavior of Q331K dynamics in cells and autoregulatory activity compared to wild-type seen by others (Hallegger et al., 2021). The observed discrepancies between our cell-based model and the animal model may be caused by species or cell type-specific factors that influence TDP-43 condensation and/or autoregulation. As previously reported by Weskamp et al. (2020), we observed strong defects in sTDP-43 autoregulation using our stable cell line upon quantifying protein and mRNA levels of endogenous TDP-43 (Figures 5B-F ). These results are consistent with the important role of the CTD in this function (Ayala et al., 2011).

Because of the dramatic decrease in autoregulation associated with sTDP-43, we measured its binding affinity for CLIP34 RNA, comparing it to WT and C (Table 1 and Supplementary Figure 3 ). While C did not significantly change binding relative to WT, sTDP-43 caused a significant twofold increase in IC 50 , suggesting a moderate decrease in binding affinity of sTDP-43. The spontaneous fibrilization of sTDP-43 may account for this reduction in binding and should be further investigated. Nevertheless, based on previous studies, we predict that a decrease of binding affinity in this range is unlikely to alter cellular RNA processing function (Lukavsky et al., 2013). We also found that sTDP-43 was consistently expressed at lower protein levels compared to WT and the other mutants in the stable cell lines (Figure 5B ). This may be caused by increased clearance of this isoform because of its propensity to aggregate/misfold, as seen in our assays (Figure 4B ). Importantly, the decreased sTDP-43 protein levels may partly account for the reduction in its autoregulatory activity. Therefore, we propose that the extreme defects in autoregulation in the case of sTDP-43 arise from impaired LLPS, decreased protein expression, and impaired protein-protein interactions in the absence of the CTD. The behavior observed for sTDP-43 may partly explain the association of this isoform with neurotoxicity and TDP-43 pathology in neurons and in patients (Weskamp et al., 2020).

We found that M337V did not affect autoregulatory function when analyzing mRNA transcript levels (Figures 5E, F ). These findings suggest that although M337V decreased the liquid properties of the condensates, these defects are not sufficient to significantly alter RNA processing in the case of autoregulatory function. Intriguingly, we found a 40% reduction in autoregulatory function when quantifying endogenous TDP-43 protein levels in M337V cells (Figures 5B-D ). This difference derives from significantly greater accumulation of endogenous TDP-43 protein in M337V expressing cells, relative to WT (Figure 5C ). To investigate the mechanism associated with this phenotype, we asked whether M337V increases protein aggregation in cells, as we previously observed using the purified protein (French et al., 2019). The M337V protein was found to be less soluble than the WT counterpart upon cell lysate fractionation into soluble and insoluble fractions (Supplementary Figures 4A-C ). These results are in agreement with increased levels of soluble and insoluble TDP-43 in human induced pluripotent stem cells (iPSCs)-derived neurons expressing M337V (Bilican et al., 2012). Next, investigated whether M337V prevents the normal turnover and clearance of endogenous TDP-43 (Figure 6A ). M337V-expressing cells were treated with cycloheximide (CHX), a translation inhibitor, and we measured the levels of induced, exogenous M337V and endogenous TDP-43 protein at different time points posttreatment. Cells expressing WT TDP-43 were similarly treated as control. Quantification of the protein levels for up to 40 h post-CHX treatment showed a stark increase in the half-life of induced M337V protein relative to induced WT protein (Figure 6B ). Furthermore, the half-life of endogenous TDP-43 in M337Vexpressing cells was also drastically increased relative to that measured in WT-expressing cells (Figure 6B ). These results strongly suggest that M337V protein homeostasis is impaired and that these defects lead to aberrant accumulation of M337V that affects non-mutated TDP-43 in cells as well. We explored whether these defects may be reflected as differences in TDP-43 cellular localization. However, we did not observe significant changes in the nuclear-cytoplasmic localization of M337V compared to WT, or in the diffuse vs. focal organization of the transgene and endogenous proteins in the nucleus, based on immunofluorescence (Supplementary Figure 4D ). Interestingly, we found that the levels of two other RNA binding proteins associated with ALS, FUS and matrin-3, also showed abnormal accumulation post-CHX treatment in M337V, compared to WT-expressing cells (Figure 6B ). The stark differences in the proteostasis of M337V may be caused by defects in the loss of liquid properties unique to this mutation as seen by our in vitro condensation assays. These intriguing findings also suggest that the defects caused by M337V alter general pathways of cellular proteostasis, affecting aggregation-prone proteins that are relevant to disease.

DISCUSSION

The autoregulatory activity of TDP-43 is central to its function in RNA processing as even moderate changes in TDP-43 levels may significantly alter gene expression. The importance of autoregulation for TDP-43 metabolism is further highlighted by two lines of evidence. First, in vivo models of TDP-43 pathology and patient-derived tissue show aberrantly increased levels of TDP-43 expression (Gitcho et al., 2009;White et al., 2018;Liu et al., 2019). Second, defects in autoregulation cause TDP-43associated neurotoxicity (Tatom et al., 2009;Barmada et al., 2010;Wils et al., 2010;Xu et al., 2010;Estes et al., 2011;Uchida et al., 2012;Diaper et al., 2013;White et al., 2018). These observations strongly indicate that impaired autoregulation may result in elevated protein levels, increasing the aggregation propensity of TDP-43. Here, we show that this essential TDP-43 function is mediated by TDP-43 condensation into ribonucleoprotein complexes through a mechanism implicating specific RNA binding and protein domain interactions. We show that TDP-43 binding to CLIP34 RNA specifically, whose sequence forms part of the TDP-43 binding region mediating autoregulation, increases the liquid properties of TDP-43 condensates. Importantly, we find that TDP-43 ALS-associated mutations and the short variant isoform sTDP-43 affect different interactions involved in this process. In particular, our studies provide insight into possible pathogenic mechanisms associated with the M337V mutant and the short TDP-43 isoform associated with neurotoxicity in ALS. These findings provide mechanistic insight of an important process regulating TDP-43 homeostasis and may help in the design of new therapeutic strategies to control TDP-43 function and solubility.

Based on our results, TDP-43 autoregulation requires TDP-43-self-assembly and condensate formation. We find that disruption of TDP-43 LLPS activity through targeted mutations strongly impairs autoregulatory activity (Figure 3 ). These results are supported by studies from the Ule lab published while this manuscript was in preparation (Hallegger et al., 2021). Hallegger et al. (2021) showed that condensation activity is correlated with autoregulatory function by disrupting the conserved α-helical region in the LCD. Here, we show that in addition to interactions mediated by the LCD, self-assembly through the N-terminal domain of TDP-43 is essential for autoregulation (Figure 3 ). This mechanism of oligomerization is necessary for TDP-43-mediated splicing regulation of numerous targets and regulation of R-loop metabolism, according to our previous data and others (Afroz et al., 2017;Wang et al., 2018;Wood et al., 2020), suggesting that these processes and autoregulation share common mechanisms. Our results showing that autoregulation-linked condensation requires RNA binding as well as NTD and interactions is consistent with the pivotal role of multivalent interactions in mediating condensate networks. We also note that, at this time, we may not rule out that the N and C-terminal regions are required for oligomerization or interactions with protein partners, in the absence of phase separation. While this possibility needs to be investigated, we provide strong evidence for an important role of TDP-43 condensation during autoregulation. We showed that CLIP34 specifically and potently induces liquid-liquid phase separation of purified TDP-43 as well as in the presence of cellular lysate and promotes liquid properties of TDP-43 droplets. In addition, our work shows inhibition of autoregulation following the introduction of mutations that inhibit LLPS. Interestingly, our findings suggest that TDP-43 autoregulation may be controlled by phosphorylation at residue S48 in the N-terminal domain, which strongly inhibits self-assembly and phase separation (Figure 3 ; Wang et al., 2018). The cellular pathway controlling S48 phosphorylation is unknown, yet our data suggests that it may be a powerful modulator of TDP-43 function and proteostasis.

To explore possible links between the condensationautoregulation process and TDP-43-associated disease, we investigated the function of TDP-43 mutations causative of ALS. Among the disease mutations within the CTD conserved region, we found that A321G and M337V were able to form condensates on their own, similar to WT. However, CLIP34 RNA did not confer liquid properties to the condensates formed by these mutants, in contrast to WT and the other mutations tested (Figure 4 ). These defects were consistent with our recent observations that GU-repeat RNA binding is not capable of increasing the liquid properties to A321G and M337V condensates, in contrast to its effect on WT TDP-43 (Grese et al., 2021). The altered behavior of A321G and M337V suggests less dynamic assemblies compared to WT, which may result in accelerated conversion into fibrils or aggregates. This model is consistent with our previous findings that M337V accelerates aggregate formation and increases aggregate seeding capacity in cells (French et al., 2019). In addition, aberrant M337V condensate behavior is supported by defects in stress granule function observed in a previous M337V mouse model (Gordon et al., 2019). However, despite the observed loss in liquid properties, A321G and M337V mutants did not significantly decrease autoregulatory function at the transcript level, compared to WT (Figure 5 ). This is in contrast with mutations that greatly decreased or even blocked LLPS entirely, such as CR and NTD mutations (Figure 3 ). We propose that the ability of A321G and M337V to form condensates, albeit of decreased liquidity, is sufficient for autoregulation in cells where additional protein interactions (e.g., chaperones) may counteract defects in A321G and M337V complexes. It is also possible that additional regulatory factors in cells, such as 3 UTR RNA sequences flanking CLIP34 in the entire TDPBR are able to rescue the defects observed with the purified A321G and M337V proteins.

Our present studies led to intriguing observations of diseaseassociated factors disrupting normal TDP-43 homeostasis. We found significantly increased accumulation of endogenous TDP-43 protein in M337V expressing cells, compared to wild-type (Figures 5B-D ). These results are consistent with previous reports of neurons derived from M337V carrier patients, which show decreased survival and higher levels of total TDP-43 compared to control (Bilican et al., 2012). We also find that elevated endogenous TDP-43 protein levels were caused by a dramatic decrease in its clearance upon coexpression of M337V (Figure 6 ). The M337V protein also shows greatly reduced turnover compared to WT, as was previously reported in a different human cell line (Yin et al., 2021). Of note, the WT and M337V proteins were expressed in isogenic cell lines in our studies, suggesting that M337V homeostasis is intrinsically different from WT. Based on these results, we speculate that structural and functional defects in M337V disrupt the clearance of non-mutated TDP-43 in the affected cells, having the overall effect of increasing TDP-43 levels. Moreover, our observations that FUS and matrin-3 proteins also showed abnormal accumulation in M337V cells suggest strong defects in cellular protein homeostasis. Based on the accelerated maturation of M337V condensates, we speculate that M337V sequesters TDP-43, and perhaps other related proteins (e.g., FUS, matrin-3), into insoluble aggregates. It is also possible that the decreased solubility of M337V (Supplementary Figure 4 ) reduces the capacity of cellular protein clearance through a yet uninvestigated mechanism. These findings are relevant to patients carrying the M337V mutation as heterozygous, whereby expression of the M337V protein also affects the homeostasis of the WT allele as well as other proteins.

Our findings show that TDP-43 binding to CLIP34, specifically modulates liquid-liquid phase separation properties of TDP-43 (Figure 1 ). We previously showed evidence that TDP-43 LLPS is upregulated with increasing number of available RNA binding sites. According to band shift analyses, we estimated that 3-4 TDP-43 molecules bind to CLIP34 (Grese et al., 2021). The increased valency of the RNA-bound complex is likely to increase TDP-43 phase separation. Importantly, this sequencespecific RNA binding activity is also demonstrated in a cellular environment using human cell lysate (Figure 2 ). These results are consistent with our recent findings that interactions with specific RNA sequences promote TDP-43 condensation, and at the same time, increase the liquidity of phase separation (Grese et al., 2021). Additionally, Mann et al. (2019) previously showed that CLIP34 RNA decreases TDP-43-mediated neurotoxicity. Collectively, this evidence indicates that specific TDP-43-RNA interactions are important to prevent TDP-43 aggregation in the condensed state, during the processing of specific targets.

DATA AVAILABILITY STATEMENT

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

AUTHOR CONTRIBUTIONS

LK, ZG, AB, and YA designed the study. ZG and LM generated the recombinant protein and performed the phase separation analysis. LK carried out the cell biology experiments and analyses. AB measured the RNA binding affinity. TH participated in the design, analysis and interpretation of the RNA binding assays. LM generated the DNA constructs and mutagenesis. YA wrote the manuscript. LK, ZG, AB, LM, and YA reviewed and edited the manuscript. All authors contributed to the article and approved the submitted version.

SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fnins. 2022.818655/full#supplementary-material Supplementary Figure 1 | CLIP34 RNA specifically increases the liquid properties of TDP-43 condensates. The size of TDP-43 condensates in the presence and absence of RNA was quantified as an estimation of differences in their liquid properties. The area of condensates composed of purified TDP-43 control, in the presence of CLIP34 or A(CA) 18 RNA shown in

Conflict of Interest:

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher's Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Copyright © 2022 Koehler, Grese, Bastos, Mamede, Heyduk and Ayala. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

Funding

FUNDING This work was supported in part by funding from the National Institutes of Health (NIH), National Institute for Neurological Disorders and Stroke (NINDS), and National Institute on Aging (NIA) R01 NS114289 and NINDS R56 NS105806 to YA; and the Department of Defense CDMRP/ALSRP W81XWH-20-1-0241 to YA.

Failed to load PDF:

StripeM-Inner

Paper sources

Abstract screening pilot

Abstract screening results

Extraction pilot

Extraction results

Research report

Modify setup