What is the key player in TDP-43 pathology in ALS: Disappearance from the nucleus or inclusion formation in the cytoplasm?



Osamu Onodera, Department of Molecular Neuroscience, Resource Branch for Brain Disease Research, Center for Bioresource-based Research, 1-757 Asahimachi-dori, Chuo-ku, Niigata-City, Niigata 951-8585, Japan. Email: onodera@bri.niigata-u.ac.jp


C9ORF72 and the 43 kDa TAR DNA-binding protein (TDP-43) are key molecules in the development of TDP-43 pathology in amyotrophic lateral sclerosis (ALS). The hexanucleotide repeat expansion in C9ORF72 also leads to frontotemporal lobar degeneration, whereas mutation of TARDBP mainly causes ALS, indicating that TDP-43 plays a more fundamental role in the development of ALS. In tissues affected with ALS, TDP-43 is dislocated from the nucleus, forms cytoplasmic inclusions, and is phosphorylated and truncated. Accumulating evidence suggests that the disappearance of TDP-43 from the nucleus precedes inclusion formation, indicating that its disappearance from the nucleus is crucial in the development of TDP-43 pathology. Alterations in the quality and quantity of TDP-43 might result in the disappearance of TDP-43 from the nucleus. Regarding quality, phosphorylation and truncation of TDP-43 is not necessary for its disappearance from the nucleus or for inclusion formation. Although it has been speculated that studies of TDP-43 harboring ALS-associated mutations are useful for understanding the molecular pathogenesis of sporadic ALS, the functional and biochemical differences between mutated and wild-type TDP-43 remain unclear. Regarding quantity, an increased amount of TDP-43 is an attractive hypothesis as it has been shown that increased amounts of TDP-43 are toxic. Moreover, several reports have suggested that increased levels of TDP-43 are found in sporadic ALS as well as in ALS with TDP-43 mutations. However, these findings remain controversial. Increased understanding of the mechanisms responsible for regulating TDP-43 will provide a basis for determining the molecular pathogenesis of ALS.


Identification of the 43 kDa TAR DNA-binding protein (TDP-43) has been the most exciting breakthrough in amyotrophic lateral sclerosis (ALS) research in the past two decades.[1, 2] TDP-43 has two pathological hallmarks in tissues affected by sporadic ALS (SALS): cytoplasmic inclusions and disappearance from the nucleus.[2-5] These features of TDP-43 may play a key role in the molecular pathogenesis of ALS. Any theory of the molecular pathogenesis of ALS should explain why and how TDP-43 disappears from the nucleus and forms inclusions in the cytoplasm. In this review, we summarize findings that highlight the importance of TDP-43 in ALS pathogenesis and discuss the molecular processes underlying the development of TDP-43 pathology in ALS.

Is alteration of TDP-43 a primary event in the pathogenesis of SALS?

TDP-43 inclusions have also been described in other neurodegenerative diseases, including frontotemporal lobar degeneration (FTLD), Alzheimer's disease, temporal sclerosis, diffuse Lewy body disease, and aging.[6] In addition, PGRN and DCTN mutations result in TDP-43 inclusions in the affected neurons.[6] Although these findings suggest that TDP-43 inclusions might result from the dysfunction of affected cells in these disorders, TDP-43 pathology is not observed in motor neuron diseases caused by mutations in SOD1, AGN, or by CAG repeat expansion in the androgen receptor gene.[6] Therefore, general dysfunction of spinal motor neurons does not result in TDP-43 pathology, indicating that TDP-43 plays a primary role in the pathogenesis of SALS.[3, 5, 6]

The detection of mutations in TARDBP, the gene that encodes TDP-43, in familial ALS (FALS) as well as SALS patients confirmed the importance of TDP-43 in ALS.[6-12] We observed a TARDBP mutation in a patient with FALS who had pathological hallmarks of ALS including TDP-43 pathology and Bunina bodies.[5, 12] Moreover, the biochemical features of TDP-43 are identical to those of SALS.[12] These findings suggest that alterations of TARDBP are a direct cause of the ALS phenotype. However, the penetration of most TARDBP mutations observed in ALS patients appears to be low,[6-12] raising the possibility that some of the TARDBP mutations reported in ALS patients are risk factors for the disease onset but not definitive causes of the disease. In addition, because the molecular mechanism by which mutated TDP-43 causes ALS remains obscure, we cannot estimate the pathogenesis of a TARDBP mutation by a biochemical assay. Therefore, we have to be very careful when analyzing individual TARDBP genes for diagnostic purposes.

Mutations of PGRN and DCTN also result in TDP-43 pathology in neurons.[6] Although these genes are expressed ubiquitously in the central nervous system, the inclusions are restricted to specific regions of the central nervous system and are not observed in motor neurons. Thus, these molecules are less likely to play an important role in the pathogenesis of SALS with TDP-43 pathology. VCP mutations cause FTLD as well as motor neuron diseases.[13, 14] Thus, the function of VCP could be associated with the pathogenesis of ALS. Further studies are necessary to clarify whether the clinico-pathological features of motor neurons in patients with VCP mutations are identical to those of SALS.

The impact of the discovery of a causative gene for FTD/ALS linked to chromosome 9

A recent exciting finding in ALS was the identification of a hexanucleotide repeat expansion in the C9ORF72 gene in individuals with frontotemporal dementia (FTD) and/or ALS linked to chromosome 9.[15, 16] The pathological findings of TDP-43 in individuals with a C9ORF72 mutation presenting with ALS are identical to those observed in SALS patients.[17] Thus, a study of molecular pathogenesis related to the C9ORF72 mutation would be useful in revealing TDP-43 pathology in SALS. The expansion of a repeat sequence of a non-coding region suggests that an alteration in messenger RNA (mRNA) splicing causes TDP-43 pathology.[15, 16]

The most characteristic feature of the C9ORF72 mutation is that it results in ALS as well as FTD phenotypes even in the same families, and sometimes both phenotypes are observed in one individual.[15, 16, 18, 19] Clinical evaluation of patients with the C9ORF72 mutation revealed that the tendency for the clinical phenotypes to converge with disease progression was more prominent in the FTD phenotype than the ALS phenotype.[20] Therefore, we propose a hypothesis for the development of the diversity of clinical phenotypes in patients with the C9ORF72 mutation (Fig. 1). Expansion might increase the risk for pathological alteration of TDP-43 in cortical neurons or spinal motor neurons. The region where the initial alteration of TDP-43 occurs might determine the clinical phenotype.[20] If the alteration of TDP-43 starts in the spinal cord, the disease expansion might be relatively limited to motor neurons and show an ALS phenotype. If it starts in the cerebral cortex, pathologically changed TDP-43 might propagate into the remainder of the cerebral cortex and show an FTD or FTD/ALS phenotype.

Figure 1.

The basis of clinical diversity caused by repeat expansion in C9ORF72. Expansion of GGGGCC repeats in C9ORF72 increases the risk for pathological alteration of TDP-43 in the cerebral cortex (frontal, temporal or parietal) or spinal motor neurons. The pathological change occurs by chance and may spread to other regions. Occurrence in the spinal motor neurons restricts the disease to the motor neuron system, and occurrence in the cerebral cortex results in the disease possibly spreading to other cortical areas. FTD, frontotemporal dementia; ALS, amyotrophic lateral sclerosis.

TDP-43 pathology in FTLD has been classified into four types based on the intracellular distribution of TDP-43 inclusions.[21] Of these, the type A and D pathologies are closely associated with specific gene mutations.[21] Although the pathology of FTLD linked to chromosome 9 has been proposed as type B,[21] the type of TDP-43 pathology in FTLD with the C9ORF72 mutation varies.[17] The diversity of the pathological findings of TDP-43 in FTLD with the C9ORF72 mutation suggests that these pathological differences do not play a fundamental role in the molecular pathogenesis of FTLD with the C9ORF72 mutation.

Recently, Tsuji et al. reported that the pattern of the protease-resistant fragment of TDP-43 differs among ALS and the FTLD types.[22] Therefore, investigating whether the pattern of protease-resistant fragments of TDP-43 in ALS with C9ORF72 mutations is identical to that of SALS is of interest to further our understanding of the similarities between these disorders. Moreover, if the biochemical signature of TDP-43 in FTLD patients with a C9ORF72 mutation is diverse, we might suggest that these biochemical signatures are not determined by genetic factors in FTLD with C9ORF72 mutations.

The process of the development of TDP-43 pathology in ALS

Next, we discuss the process of the development of TDP-43 pathology in SALS based on pathological findings and biochemical analyses of TDP-43. Two possible processes for the development of TDP-43 pathology have been proposed: cellular dysfunction resulting in a TDP-43 alteration, and a TDP-43 alteration resulting in cellular dysfunction. We have already discussed the former possibility. Thus, we would like to discuss the latter possibility by focusing on three aspects of TDP-43 alterations in this section: character (quality and quantity); intracellular localization; and inclusion formation (Fig. 2). To understand the most fundamental process for the development of TDP-43 pathology, we would like to review our current understanding of the relationships among these factors.

Figure 2.

Paradigm for the development of TDP-43 pathology in amyotrophic lateral sclerosis (ALS). VCP, valocin-containing protein.

Dislocation of TDP-43 from the nucleus may precede inclusion formation

First, we consider the relationship between the localization and inclusion formation of TDP-43. Many neurodegenerative disorders also involve inclusion bodies in the affected tissues or cells. Among these, expanded polyglutamine proteins form aggregates and recruit other proteins into the inclusion, which is then followed by the disappearance of soluble polyglutamine proteins from the cytoplasm. Thus, the finding that TDP-43 forms inclusions along with its disappearance from the nucleus raises the possibility that nuclear TDP-43 is recruited into the inclusions. In addition, TDP-43 forms a dimer by interacting with the middle portion of the protein.[23] Indeed, exogenous expression of the C-terminal fragment of TDP-43 or nuclear localization of signal-mutated TDP-43 leads to inclusion formation in the cytoplasm and recruitment of intrinsic TDP-43 into the inclusions, whereas substantial amounts of TDP-43 remain present in the nucleus.[24-27] In addition, if the inclusion formation precedes the cytoplasmic location of TDP-43, then why does TDP-43 form inclusions in the cytoplasm and not in the nucleus where TDP-43 is normally found? Moreover, several studies have demonstrated that some neurons show diffuse cytoplasmic TDP-43 without nuclear TDP-43.[4, 14, 28] Thus, clearance of nuclear TDP-43 may precede TDP-43 inclusion formation in the cytoplasm.

Is modification of TDP-43 a result or a cause of TDP-43 dislocation and inclusion formation?

Next, we review the role of TDP-43 post-transcriptional modification in the disappearance of TDP-43 from the nucleus and in inclusion formation. Under pathological conditions, TDP-43 is modified by phosphorylation, ubiquitination and truncation.[1, 2, 29] Ubiquitination may result from an increase in insoluble materials, and so we will focus on the truncation and phosphorylation of TDP-43 in terms of TDP-43 pathology.

The C terminus of TDP-43 is truncated, phosphorylated and forms inclusions in tissues affected by ALS.[1, 2, 29-31] The C terminus of TDP-43 has been predicted to demonstrate a disordered structure that folds properly in complexity with target structures and tends to form aggregates without target structures (Fig. 3). Indeed, overexpression of the C terminus of TDP-43 forms insoluble inclusions containing phosphorylated TDP-43 in the cytoplasm.[24, 25, 32] Several reports have indicated that TDP-43 with a mutation associated with ALS is more truncated than the wild-type form.[8, 9, 33, 34] Therefore, the truncation of TDP-43 is an attractive hypothesis for explaining aggregate formation by TDP-43. However, the insoluble protein fraction from tissues affected by ALS and cytoplasmic inclusions of TDP-43 contain substantial amounts of full-length TDP-43,[1, 2, 29, 31] indicating that TDP-43 truncation is not an essential event for inclusion formation.

Figure 3.

Disordered profile plot for TDP-43 Disordered profile plot for TDP-43 was calculated using DISOPRED (http://bioinf.cs.ucl.ac.uk/disopred). The bold line shows the mutation cluster region in individuals with amyotrophic lateral sclerosis (ALS).

C-terminal phosphorylated full-length and truncated TDP-43 is another characteristic feature in tissues affected by ALS.[4, 14, 28] Phosphorylated TDP-43 is observed as small punctate aggregates in cytoplasm but is rarely observed in the nucleus.[4, 14, 28] The existence of small punctate aggregates containing phosphorylated TDP-43 suggests that these small aggregates may act as seeds for round or skein-like inclusions and contribute to TDP-43 cytoplasmic localization. However, artificially mutated TDP-43 mimicking the phosphorylated or dephosphorylated states did not result in altered intracellular localization or decreased solubility.[32, 35] In addition, the mutation of 409/410 to phosphomimetic aspartic acid residues reduced aggregation.[32, 35] These findings indicate that the phosphorylation of TDP-43 may not be a key process in TDP-43 pathology.

Taken together, these findings suggest that the disappearance of nuclear TDP-43 may precede TDP-43 inclusion formation, truncation or phosphorylation (Fig. 2). The nuclear localization of the protein is a dynamic process regulated by the balance between the amount of proteins that enter and exit the nucleus. Although several factors are associated with the intracellular localization of TDP-43,[36, 37] the precise molecular mechanism underlying the intracellular localization of TDP-43 has not been elucidated. We speculate that three factors affect the disappearance of TDP-43 from the nucleus: decreased inflow of TDP-43 into the nucleus; increased outflow of TDP-43 from the nucleus; and suppression of TDP-43 synthesis. To resolve the mystery of the disappearance of TDP-43 from the nucleus, these factors should be evaluated in affected cells in the future.

Qualities of ALS10-associated mutated TDP-43

Studying the characteristics of ALS-associated mutant TDP-43 may contribute to our understanding of the mechanism of TDP-43 pathology in SALS. Regarding the intracellular localization of ALS-associated mutant TDP-43, we and others have observed that these mutants are diffusely distributed in the nucleus in cultured cells,[12, 38] whereas ALS-associated mutant FUS (fused in sarcoma) dislocates from the nucleus.[39] TDP-43 binds to the intron or 3′-untranslated region (UTR) of pre-mRNAs and heterogeneous ribonucleoproteins (hnRNPs)[40, 41] and affects pre-mRNA splicing and levels of mRNA. The C terminus of TDP-43 binds to hnRNPs.[42, 43] Interestingly, mutations of TDP-43 identified in ALS patients cluster at the C terminus (Fig. 3).[6] Although mutations in this portion of the protein have been speculated to affect protein–protein interactions and to alter pre-mRNA splicing function of TDP-43, to date there have been no reports of differences in the splicing function of pre-mRNAs in conjunction with these mutations.[43] In addition, a comprehensive liquid chromatography coupled to tandem mass spectrometry (LC–MS/MS) analysis revealed that the binding protein profiles of wild-type and mutated TDP-43 were identical.[44]

Concerning the possible gain of toxic properties by mutated TDP-43, although toxicity of mutated TDP-43 has been reported in cultured cells,[45] Drosophila,[46] Caenorhabditis elegans,[47, 48] chicken embryo,[10] zebrafish[38] and rats,[34] most research involving model mice has also found toxicity in wild-type TDP-43. In contrast, many studies have failed to demonstrate a clear difference in toxicity between mutated and wild-type TDP-43.[27, 33, 49-57] Thus, despite extensive research, the toxicity of mutated TDP-43 has not yet been shown conclusively, and it is clear that overexpressed TDP-43 is toxic regardless of its mutation status.

Regulation of the quantity of TDP-43

An abnormal quantity of TDP-43 is an attractive hypothesis, as increasing the amount of TDP-43 has been reported to be toxic and also increases the levels of insoluble TDP-43.[27, 33, 49-57] Before discussing the quantity of TDP-43 harboring ALS-associated mutations, we will introduce the mechanism responsible for regulating the quantity of TDP-43. Levels of TDP-43 are regulated by TDP-43 itself.[40, 58, 59] Such an autoregulatory mechanism is frequently observed among splicing factors,[60] of which TDP-43 is a member. In addition, the last exon of TARDBP gene has characteristic features for autoregulated genes: it is very large, conserved with other species, and is disordered.[61, 62] To date, no alteration in the copy number of TARDBP or nonsense mutation that fulfills the criteria for nonsense-mediated mRNA decay has been reported in an individual with ALS. Thus, the autoregulatory mechanism of TDP-43 is likely to maintain normal levels of TDP-43, even in individuals with these mutations.

TDP-43 binds to TDP-43 mRNA at the 3′-UTR and excludes this portion, resulting in alterations in TDP-43 mRNA.[40, 58, 59] First, this event creates a new last exon after the stop codon, fulfilling the criteria for nonsense-mediated mRNA decay.[40] Nonsense-mediated mRNA decay is a mechanism for reducing the amount of unfavorable mRNA possessing a termination codon at least approximately 50 bp upstream from the last exon–exon junction.[63] Second, this event eliminates the traditional poly-A site and thus, the newly synthesized mRNA uses other poly-A sites with longer 3′-UTRs, which may preferentially localize to the nucleus and result in a reduced amount of TDP-43 protein.[58] To reveal how these mechanisms regulate the levels of TDP-43 mRNA, the metabolism of TDP-43 mRNA in each tissue should be evaluated.

Quantity of ALS10-associated mutated TDP-43

Next, we review alterations in the amount of TDP-43 in cases of ALS-associated mutation. The region in which mutations of TARDBP are clustered is predicted to be disordered in structure. Because disordered proteins maintain their structure by interacting with other proteins, mutations in this portion are speculated to affect protein–protein interactions and alter solubility. Indeed, ALS-associated TDP-43 has been reported to demonstrate more stability than wild-type TDP-43 as well as an increased tendency to form aggregates in vitro.[25, 33, 45, 64] In contrast, no convincing report has demonstrated increased amounts of TDP-43 inclusions in tissues of model animals with TDP-43 mutations.[27, 51-54, 57] In addition, we found no cytoplasmic inclusions or increased insoluble fraction of TDP-43 harboring the Q343R mutation in fibroblasts. These findings raise the question of whether the increased stability of mutated TDP-43 in vitro contributes to the molecular pathogenesis of SALS. Moreover, the increased stability hypothesis does not explain why inclusions are formed in the cytoplasm instead of the nucleus where TDP-43 is normally found.

An increasing amount of TDP-43 mRNA is another interesting hypothesis for partly explaining TDP-43 pathogenesis in ALS. However, the increase in TDP-43 mRNA and/or protein in SALS patients remains controversial.[65-68] If the autoregulatory mechanism remains intact in cells without nuclear TDP-43,[40, 58, 59] TDP-43 mRNA might continuously increase. In addition, TDP-43 protein and/or mRNA were found to be increased in motor neurons derived from induced pluripotent stem cells (iPS) from patients with TDP-43 mutations.[69, 70] However, the results varied among iPS lines, and too few iPS lines were evaluated.[69, 70] In addition, we found no change in the amounts of TDP-43 mRNA harboring the Q343R mutation in fibroblasts. Thus, to evaluate these results, we have to wait for further studies to increase our fundamental knowledge regarding iPS.


Accumulating evidence suggests that the disappearance of TDP-43 from the nucleus plays a fundamental role in the development of TDP-43 pathology (Fig. 2). Elucidating the mechanism responsible for the disappearance of TDP-43 from the nucleus may increase our understanding of the molecular mechanism of ALS. Investigation of how the ALS-associated mutation in TDP-43 develops the TDP-43 pathology may be useful for resolving this issue. In addition, the fact that the TDP-43 pathology is limited in the central nervous system in individuals with TDP-43 mutations suggests that host cell factors also have an important role in the development of TDP-43 pathology. To elucidate the molecular mechanism of ALS, we must further our understanding of the mechanisms responsible for regulating the quantity and quality of TDP-43 and the fundamental characteristics of the motor neuron system.


This research was supported through a Grant-in-Aid for Scientific Research (A) from the Japan Society for the Promotion of Science, a Grant-in-Aid for the Research Committee of CNS Degenerative Diseases and Comprehensive Research on Disability Health and Welfare from the Ministry of Health, Labor and Welfare, Japan, a Grant-in-Aid from the Uehara Memorial Foundation.

Conflicts of interest

The authors declare no conflicts of interest.