A paper published March 5 in Neuron takes a subcellular look at the biology of TDP-43 aggregation, a feature of amyotrophic lateral sclerosis (ALS), frontotemporal dementia (FTD), and Alzheimer’s disease. The authors, led by Sandrine Da Cruz and Don Cleveland, both at the University of California at San Diego, found that the protein forms droplets in the nucleus as part of its normal function. However, as described in another recent Neuron paper, Da Cruz and colleagues found that TDP43 condenses independently of stress granules in the cytoplasm of stressed cells. This steals away other proteins needed for effective transport in and out of the nucleus. The results suggest how wayward TDP-43 may disrupt nucleocytoplasmic transport and kill neurons.

  • In the nucleus, TDP-43 normally forms liquid droplets.
  • When cells are stressed, TDP-43 forms cytoplasmic droplets independently of stress granules.
  • These cytoplasmic droplets deplete proteins needed for nucleocytoplasmic transport.

First author Fatima Gasset-Rosa examined endogenous TDP-43 using high-resolution microscopy in various kinds of somatic cells, including mouse and human neurons. In each, she found that normal TDP-43 formed droplets in the nucleus that dynamically fused and broke apart, rapidly exchanging with TDP-43 in the nucleoplasm. This suggested that dynamic liquid-liquid phase separation (LLPS) is a physiological part of TDP-43 function in the nucleus.

However, if the researchers treated human SH-SY5Y neuroblastoma cells with fibrils of either TDP-43 or FUS, TDP-43 drained out of the nucleus and accumulated in liquid droplets in the cytoplasm. The same droplets formed if the researchers overexpressed TDP-43 in the cytoplasm. These membraneless organelles were also dynamic, readily exchanging their TDP-43 with that in their surroundings. Like the TDP-43 droplets identified by Christopher Donnelly and colleagues at the University of Pittsburgh, they were distinct from stress granules (Mar 2019 news). These are liquid droplets of RNA and RNA-binding proteins, including TDP43, that temporarily form when cells are under duress. They are believed to protect cells by putting a brake on translation. One theory suggests these droplets provide an incubator for the aggregation of TDP43 and other proteins to eventually form solid aggregates (Jul 2010 conference news; April 2018 news). These new papers indicated that can happen sans stress granules.

Death Cascade. Neurons normally form droplets of TDP-43 in the nucleus. As cellular stress builds, toxic droplets form in the cytoplasm, sequestering other proteins, disrupting nucleocytoplasmic transport, and ultimately killing the cells. [Courtesy of Gasset-Rosa et al., Neuron.]

Gasset-Rosa found that the cytoplasmic TDP-43 droplets appeared to be toxic. Initially, they didn’t kill cells, but after about two weeks, 60 percent of neurons had died, and by six weeks, all had. In dying cells, nuclear proteins RanGAP1 and Ran-GTP, both required for nucleocytoplasmic transport, had mislocalized and accumulated in the cytoplasm. Similarly, Nup62, a component of nuclear pores, hnRNPA1, which shuttles between nucleus and cytoplasm, and importin-α, which reads the TDP-43 nuclear localization signal, all aggregated with TDP43.

When the cells were exposed to additional stress, such as sodium arsenite, the cytoplasmic droplets congealed into gels or solids containing phosphorylated TDP-43. Most of it assumed amyloid conformations, as detected by binding to the A11 antibody, which recognizes prefibrillar oligomers of Aβ and other proteins.

Taken together, the results demonstrate that independently of stress granules, TDP-43 phase-separates in the cytoplasm in response to various stressors, and that this can both interfere with nucleocytoplasmic transport and siphon away nuclear TDP-43, resulting in cell death.

“It suggests that cytoplasmic liquid TDP-43 droplets can be toxic, without the need to mature into solid aggregates, as was commonly believed,” wrote Steven Boeynaems, Stanford University, to Alzforum (full comment below). That could mean the TDP-43 pathology observed in postmortem patient material may not necessarily represent aggregates, but could constitute an aberrant liquid/gel phase, he added.

The results bolster the idea that liquid TDP-43 droplets form in early stages of disease, then yield gels or solids as pathology worsens, suggested Gasset-Rosa. “This process may be central to understanding the emergence of TDP-43 pathology,” wrote Yuna Ayala, St. Louis University, who was not involved in the research (see full comment below). The result jibes with findings from Ayala’s own lab published last week, indicating that soluble, detergent-resistant oligomers of TDP-43 precede formation of large aggregates or fibrils (French et al., 2019).—Gwyneth Dickey Zakaib


  1. This work is important because it provides evidence for TDP-43 liquid-liquid phase separation (LLPS), stress granule interactions, and fibrilization/solidification of TDP-43 in cells.

    The authors use stable cell lines for exogenous protein expression and not transient transfection, which often leads to protein overexpression. TDP-43 overexpression alone alters protein function and solubility, preventing an accurate interpretation of the phenotypes observed while studying protein localization and solubility.

    Briefly, the findings are of paradigm-shifting potential, challenging the idea that stress granules are the crucibles for cytoplasmic TDP-43 pathology in cultured cell models and in disease. The authors elegantly show that solid-like cytoplasmic accumulation of TDP-43, recognized by markers of pathology, occurs independently of stress granule recruitment. These structures are observed upon treatment with preformed fibrils, oxidative stress (arsenite treatment), and increasing cytoplasmic TDP-43 levels.

    The authors find that internalization of aggregates derived from recombinant TDP-43 or FUS induces liquid-liquid phase separation of endogenous TDP-43 in the cytoplasm, but not stress granule formation. The cytoplasmic complexes increase with time and some treated cells show depletion of nuclear TDP-43 at later time points, which is a feature that accompanies TDP-43 pathology in ALS and FTD. Even after exposing these cells to oxidative stress, TDP-43 structures fail to significantly co-localize with stress granule markers. These observations contrast numerous previous findings postulating that it is the co-localization of TDP-43 with stress granules that initiates cytoplasmic aggregation.

    The experiments studying the ΔNLS TDP-43 mutant add another layer of complexity in understanding TDP-43-stress granule interactions. The authors find only initial co-localization of this mutant with stress granules. At later time points, TDP-43 separates and redistributes into separate cytoplasmic foci in U2OS cells (see Fig4B).

    The separation of ΔNLS TDP-43 from stress granules is in agreement with McGurk et al., 2018, which showed that TDP-43 accumulation in stress granules is mediated by poly(ADP-ribose) (PAR) binding through the nuclear localization sequence (NLS). This interaction, however, is not the only required factor for TDP-43-stress granule accumulation, as the Cleveland group also shows that ΔNLS TDP-43 is initially recruited to stress granules during early arsenite treatment.

    In future studies, it would be important to increase TDP-43 cytoplasmic localization without disrupting PAR binding to identify the specific mechanisms that induce cytoplasmic TDP-43 granules independent of stress granule accumulation. This is important because, as Gasset-Rosa et al. and McGurk et al. suggest, these stress-granule-independent TDP-43 foci are particularly prone to conversion into aggregates. Therefore, stress-granule recruitment may actually be a protective mechanism to reduce toxic aggregates. This has also been suggested by Alberti and Hyman (Mateju et al., 2017), who found that chaperones are recruited to stress granules and may help in maintaining proper folding of proteins targeted to these complexes. In addition, aberrant stress granules are targeted for degradation by the aggresome to prevent accumulation of misfolded proteins.

    Another interesting observation reported here is the conversion of TDP-43 complexes with liquid properties to structures with gel or solid-like properties in cells. This process may be central to understanding the emergence of TDP-43 pathology. We are particularly interested in this process based on our own evidence of a TDP-43 aggregate intermediate, which we suggest acts as a scaffold for large aggregates. Using purified protein, we show that TDP-43 aggregation is initiated by soluble, detergent-resistant oligomers and is followed by fibril or large aggregate formation (French et al., 2019). 


    . Poly(ADP-Ribose) Prevents Pathological Phase Separation of TDP-43 by Promoting Liquid Demixing and Stress Granule Localization. Mol Cell. 2018 Sep 6;71(5):703-717.e9. Epub 2018 Aug 9 PubMed.

    . An aberrant phase transition of stress granules triggered by misfolded protein and prevented by chaperone function. EMBO J. 2017 Jun 14;36(12):1669-1687. Epub 2017 Apr 4 PubMed.

    . Detection of TAR DNA-binding protein 43 (TDP-43) oligomers as initial intermediate species during aggregate formation. J Biol Chem. 2019 Apr 26;294(17):6696-6709. Epub 2019 Mar 1 PubMed.

  2. The authors show that upon arsenic stress TDP-43 goes initially to stress granules, but when stress persists, TDP-43 demixes from the stress granule and forms a separate type of condensate. While TDP-43 initially remained dynamic in these new condensates, further prolonging arsenic exposure induced a hardening transition. These data are in line with recent reports by McGurk et al. and Mann et al. (McGurk et al., 2018; Mann et al., 2019). 

    What really sets this paper apart from the two previous reports is that treatment of cells with fibrils from TDP-43 results in the spontaneous phase separation of TDP-43 in the cytoplasm. Compellingly, these condensates persisted for up to a month without maturing into solid-like assemblies, contrary to what happens upon arsenic stress conditions. Moreover, these assemblies were positive for phospho-TDP-43 (an established pathological hallmark) and recruited nuclear transport factors, leading to nuclear transport defects, further depletion of TDP-43 from the nucleus, and eventual cell death. This suggests that cytoplasmic liquid TDP-43 droplets can nonetheless be toxic when they persist, without the need to mature into solid aggregates, as is commonly believed. Additionally it shows that the details of TDP-43 phase separation may be very dependent on the specific stress (arsenic stress vs fibrils in this case). The latter observation also emphasizes that we always need to be careful with extrapolating results made in conditions of arsenic stress to the situation in patients, where likely other stress pathways will be at play.

    The finding that TDP-43 can remain in dynamic liquid droplets for extended periods of time that are nonetheless toxic also suggests a provocative possibility: The TDP-43 pathology we have observed for more than a decade in fixed patient material may not necessarily represent aggregates, but could constitute an aberrant liquid/gel phase. This would be in line with the apparent absence of classic amyloid fibrillary structure in TDP-43 pathology. Whether true or not, there is a lot of exciting work ahead to get to the bottom of this!


    . Poly(ADP-ribose) Engages the TDP-43 Nuclear-Localization Sequence to Regulate Granulo-Filamentous Aggregation. Biochemistry. 2018 Dec 26;57(51):6923-6926. Epub 2018 Dec 17 PubMed.

    . RNA Binding Antagonizes Neurotoxic Phase Transitions of TDP-43. Neuron. 2019 Apr 17;102(2):321-338.e8. Epub 2019 Feb 27 PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Stress Granules: No Incubator for Inclusions, After All?
  2. Honolulu: TDP-43 Gets a Place in the Sun
  3. Liquid Phase Transition: A Deluge of Data Points to Multiple Regulators

Antibody Citations

  1. Soluble Amyloid Oligomers

Paper Citations

  1. . Detection of TAR DNA-binding protein 43 (TDP-43) oligomers as initial intermediate species during aggregate formation. J Biol Chem. 2019 Apr 26;294(17):6696-6709. Epub 2019 Mar 1 PubMed.

Further Reading


  1. . Molecular Mechanisms of TDP-43 Misfolding and Pathology in Amyotrophic Lateral Sclerosis. Front Mol Neurosci. 2019;12:25. Epub 2019 Feb 14 PubMed.
  2. . The role of liquid-liquid phase separation in aggregation of the TDP-43 low-complexity domain. J Biol Chem. 2019 Apr 19;294(16):6306-6317. Epub 2019 Feb 27 PubMed.

Primary Papers

  1. . Cytoplasmic TDP-43 De-mixing Independent of Stress Granules Drives Inhibition of Nuclear Import, Loss of Nuclear TDP-43, and Cell Death. Neuron. 2019 Apr 17;102(2):339-357.e7. Epub 2019 Mar 7 PubMed.