Looking for answers far away from the soma—the (un)known axonal functions of TDP-43, and their contribution to early NMJ disruption in ALS

Abstract

Background

For many years, the molecular basis of neuromuscular junction (NMJ) disruption and motor neuron death in the fatal neurodegenerative disease Amyotrophic Lateral Sclerosis (ALS) has remained elusive. Recent fundamental discoveries in axonal biology in health and disease have advanced our understanding of the local and independent subcellular mechanisms that sustain motor neuron axonal function and integrity. Several of those discoveries also highlighted the pivotal contribution of axonal degeneration to the development and progression of ALS pathology. These may mark a new era for the development of novel therapeutic strategies, as well as diagnostic tools for this lethal disease.

TAR DNA Binding Protein (TDP-43) is a DNA and RNA binding protein that belongs to a group of heterogenous nuclear ribonuclear proteins (hnRNPs). TDP-43 was first discovered in 1995 by Ou and colleagues as a protein that binds the pyrimidine-rich motifs within the TAR DNA of human immunodeficiency virus-1 (HIV-1) [1]. It was demonstrated that TDP-43 represses the transcription of HIV-1 by altering or blocking the assembly of tat-responsive transcription complexes. A few years later, Buratti and colleagues were the first to describe an RNA-binding property for TDP-43, which mediates the exon 9 skipping of the CFTR mRNA during alternative splicing and has a fundamental role in the pathogenesis of several monosymptomatic and full forms of Cystic Fibrosis [2]. Ever since, numerous publications have reported various nuclear roles for TDP-43 in transcriptional and post-transcriptional regulation, including splicing of primary transcripts and miRNA biogenesis [3, 4]. The canonical TDP-43 protein includes several domains: Nuclear Localization Sequence (NLS), two RNA Recognition Motifs (RRM1 and RRM2) two Intrinsically Disordered Domains (IDRs), and a Low Complexity Domain (LCD; glycine-rich C’ terminal). Additionally, TDP-43 was shown to undergo multiple posttranslational modifications (PTMs) such as phosphorylation, acetylation, SUMOylation, and ubiquitination [5]. TDP-43 was also found to interact with other RNA-binding proteins, such as ubiquitin-2 (UBQLN2), FMRP, G3BP1, and TIA1 [6–12]. TDP-43 can also undergo proteolytic cleavage, leading to the formation of truncated, possibly pathological forms known as C-terminal fragments [13]. Mutations in the gene encoding for TDP-43, TARDBP, were reported in about 0.5% of ALS patients (familial and sporadic forms) [14, 15].

A primary interest in the non-nuclear roles of TDP-43 began shortly after discovering the presence of phosphorylated and ubiquitinated cytoplasmic inclusions of TDP-43 within motor neurons in the brains of patients with ALS and Frontotemporal Dementia (FTD) [16, 17]. Some of the inclusions seemed to include C-terminal fragments [16], which were also ubiquitinated and phosphorylated, although those were apparent only in brain tissue but not in spinal motor neurons [18]. In postmortem spinal cord samples from deceased ALS patients there was a clear indication of both nuclear clearance and cytoplasmic accumulation of TDP-43 in motor neurons [16]. This TDP-43 pathology was shown to be common for over ~ 97% of all ALS cases, regardless of their genetic background, and thus has almost completely rerouted ALS research towards gaining knowledge on the biology and pathobiology of TDP-43. This movement led to a long debate between researchers in the field who either advocate for the contribution of TDP-43 loss of nuclear function as a driver of ALS pathology, and those who favor the gain of toxic function driven by the aberrant mislocalization and accumulation of TDP-43 in the cytoplasm [19]. The high load of novel and comprehensive evidence on both ends indicates that the two pathologies co-exist, yet whether one is more detrimental than the other remains to be further discussed and studied. Deciphering the temporal relationship and the co-dependence between loss of TDP-43 from the nucleus and the appearance of cytoplasmic inclusions is key in understanding the disease development, and whether the formation of axonal TDP-43 inclusions is necessary and sufficient for motor neuron death and synaptic pathologies in ALS. Nevertheless, how those paradigms could explain the early pathological disruption of NMJs in ALS is still puzzling, although both loss and gain of TDP-43 function seem to play a role in axonal pathology [20, 21].

Motor neuron axons are the longest and most arborized ones in the nervous system. The volume of motor neuron axonal and synaptic cytoplasm (also known as axoplasm) exceeds that of the cell body by several orders of magnitude, and thus also its protein content. It has now been established that local protein synthesis and axonal transport are two essential mechanisms by which motor neurons use to regulate their axonal and synaptic proteome [22]. Alteration in both processes has been reported repeatedly in various animal models of ALS and in patient iPSC-derived motor neurons [20, 23–25]. Recently, we and others have reported that the accumulation and condensation of TDP-43 in ALS models and ALS patients are not limited to the cell body and can be detected early during the disease in peripheral motor neuron axons, concomitantly to the early NMJ disruption and axon degeneration [20, 26, 27]. Hence, the relation between these processes as well as the opportunity to intervene with the axonal TDP-43 accumulation should be extensively researched. In this review, we aim to discuss possible pathways by which TDP-43 can accumulate within motor neuron axons and synapses, dissect how the formation of TDP-43 axonal condensates disrupts typical axonal-specific roles of TDP-43, and how these can lead to ALS pathology. We will focus mainly on the disruption of RNA localization and local translation and the process by which their alteration may lead to the disease’s early degeneration of axons and NMJs.

Part 1 – TDP-43 phosphorylation and its toxic role in axons of ALS patients

ALS is a neurodegenerative disease that targets lower and upper motor neurons, with vast heterogeneity of proposed mechanisms that lead to toxicity in the disease. However, the existence of one common pathology has wide approval among experts, the presence of cytoplasmic inclusions of TDP-43 protein. Still, the notion that TDP-43 cytoplasmic inclusions can also accumulate in remote processes and facilitate local pathology in axons and synapses was until recently, a topic of controversy. This issue directly relates to the “dying back” theory, which depicts the NMJ and axons as the primary location for degeneration in the disease [28, 29]. Following this theory, which is supported mostly by animal models but also by some human ALS patient data [26–28], distal stress at the motor neuron axons and NMJ precedes the occurrence of cell death in the spinal cord and the brain. Recently, several research groups have successfully showed the presence of phosphorylated TDP-43 inclusions in peripheral motor axons [20, 26, 27] (Fig. 1). Here, we will present the collective evidence for axonal pTDP-43 presence in ALS, suggest possible mechanisms for how distal TDP-43 accumulation and phosphorylation occur and discuss if this pathology is solely an early marker of ALS or is a direct mediator of axonal toxicity.

Open in a separate window

Fig. 1

TDP-43 axonal pathology. An illustrative map of a motor neuron with annotations for previous observations of axonal TDP-43 pathologies in human patients. Brettschneider et al., reported TDP-43 pathology in spinal motor neurons and their neurites in postmortem spinal cord sections from ALS patients. Altman et al., Riva et al., and Kurashige et al., reported TDP-43 and phosphorylated-TDP-43 pathologies in distal nerve biopsies, and in intramuscular nerves within muscle biopsies of pre-diagnosed ALS patients. Arrowhead in image of Kurashige et al., indicates accumulation of pTDP-43 in intramuscular nerve bundles of sALS patient (Rabbit polyclonal pTDP-43 antibody; 22309–1-AP, pSer409/410, Proteintech). Kushirage et al., have also validated their findings with an additional pTDP-43 antibody (not shown; mouse monoclonal pTDP-43 antibody (TIP-PTD-M01, pSer409/410, CosmoBio). Arrows in image of Riva et al., indicate accumulation of pTDP-43 in large axons within motor nerve biopsies. Arrow in inset indicates accumulation of pTDP-43 in Schwann cells as well (Rabbit polyclonal pTDP-43 antibody; 22309–1-AP, pSer409/410, Proteintech). Authors have validated their observations with another antibody (not shown; anti-TDP-43 rabbit polyclonal (1:700, 10782–2-AP, Proteintech). Brettschnider et al., used two types of antibodies for detecting pTDP-43 in spinal cord samples: pTDP-43 pSer409/410 rabbit polyclonal antibody (CAC-TIP-PTD-P07; CosmoBio), and pTDP-43 pSer409/410 rat monoclonal antibody developed by Dr. M Neumann [30]. Altman et al., used two types of antibodies to detect TDP-43 and pTDP-43 accumulation in intramuscular nerves: Rabbit polyclonal pTDP-43 antibody; 22309–1-AP, pSer409/410, Proteintech, and Rabbit polyclonal TDP-43 antibody; 10782–2-AP, Proteintech. Images were adapted from original publications [20, 26, 27, 31]

Possible causes for TDP-43 axonal accumulation

Several cellular mechanisms may explain the increase in local concentrations of proteins within axons: 1. Enhancement in cytoplasmic localization and axonal transport towards the nerve ending. 2. Decrease in axonal degradation and clearance. 3. Cell-to-cell protein transmission. 4. Local protein synthesis in axons. Although there is no concrete evidence as to which possibility is most likely in the case of TDP-43, several recent discoveries revealed few relevant pathways for this phenomenon in ALS (Fig. 2).

Open in a separate window

Fig. 2

Proposed mechanisms of TDP-43 accumulation in axons. Illustration of motor units with the various possibilities for formation and accumulation of pathological TDP-43 in axons. A Alterations in axonal transport may result in a shift towards more anterograde or less retrograde transport of TDP-43 which will subsequently modify its local concentrations in axons and synapses. B TDP-43 spreading via extracellular vesicles. One way by which TDP-43 is disposed from cells is via extracellular vesicles. These TDP-43-containing vesicles could be removed from Schwann cells or skeletal muscles and be taken-up into neighboring axons, which cannot sustain protein overloads, and are less equipped with the appropriate machinery to cope with protein accumulation. C Alterations in proteostasis equilibrium driven by either non-regulated synthesis of TDP-43 in axons, or as been shown, by dysfunction in the process of TDP-43 proteolysis and clearance from axons due to its phosphorylated state

The first and most established mechanism is the mislocalization of TDP-43 to the cytoplasm and axons via defective nucleocytoplasmic transport. Several groups have now established proof that nucleocytoplasmic defects are involved in ALS pathology in both sporadic and familial ALS cases [35–37]. Due to those defects, the nuclear localization of TDP-43 is disrupted and it accumulates in the cytoplasm [38]. However, the sequence of events is not entirely clear. Can the formation of cytoplasmic or axonal TDP-43 condensates be the seed event that disrupts nucleocytoplasmic transport and facilitate the clearance of nuclear TDP-43? Or is it the other way around – nucleocytoplasmic defects are the initiating event? One hypothesis is that key proteins of the nuclear pore complex, Importin-α and Importin-β, bind the TDP-43 nuclear-localization-signal and prevent most of its cytoplasmic localization [39]. Interestingly, recent discoveries indicate that phosphorylation of TDP-43 in proximity to its nuclear-localization-signal prevents binding of Importins [39] and therefore can promote the cytoplasmic accumulation of the protein, thus suggesting that TDP-43 phosphorylation might be a key process that facilitates redistribution of the protein from the nucleus to the cytoplasm [40]. Additionally, a recent article by Khalil et al., indicates a new NLS-independent mechanism for TDP-43 cytoplasmic mislocalization, where the activity of karyopherin-β1 (KPNB1/Importin-β), increase the solubility of TDP-43 upon their recruitment to nucleoporin rich pTDP-43 condensates. Importantly, this activity was sufficient to prevent the formation of those condensates, acting as a molecular chaperone [41].

Additional mechanisms for TDP-43 accumulation in the cytoplasm are based on TDP-43 ability to go through anterograde axonal transport into axons as part of RNP granules for delivering mRNA to distal locations [42–44]. Therefore, if more TDP-43 is present in the cytoplasm, it is possible that increased anterograde or decreased retrograde axonal transport of TDP-43 to and from motor neurons axons and synapses will follow (Fig. 2A). Another pathway to be considered is the spreading of TDP-43 into axons by cell-to-cell protein propagation, which stems from several reports which discovered a distinct spreading pattern for TDP-43 aggregates [45, 46] (Fig. 2B). This possibility is based on the prion-like properties of TDP-43 [47, 48] and its ability to be secreted via exosomes [49] and through axon terminals [50]. One such possibility would be the transmission of TDP-43 or pTDP-43 from post-synaptic muscles into pre-synapses. Although a previous report has indicated that TDP-43 pathology is absent in ALS muscles [51], other reports have suggested muscular pTDP-43 pathology is apparent in a subset of ALS patients [52]. Additionally, the subsynaptic transcriptome and proteome at NMJs are unique compared to extrasynaptic regions [53] and thus, the possibility that TDP-43 is produced and transmitted at the postsynapse in muscles cannot be eliminated. Recently, pTDP-43 pathology was reported in Schwann cells as well as in axons within patient biopsies [26], yet whether these two coexist independently, or are the result of intercellular transmission from Schwann cells to axons (or vice versa) remains to be determined.

Lastly, and perhaps the most basic of all, the accumulation of TDP43 in axons could occur due to altered proteostasis in axons – either via inhibited proteolysis, or via local synthesis of TDP-43 (Fig. 2C). TDP-43, and specifically pathological truncated forms of it, known as C-terminal fragments, have been shown to create insoluble, phosphorylated and ubiquitinated cytoplasmic inclusions that can sequester cellular components critical for proteostasis [54]. It is currently unknown if truncated forms of TDP-43 are present in axons and can initiate similar events. Still, it is plausible that axonal accumulation of TDP-43 inclusions can result in altered protein degradation and clearance, even of TDP-43 itself. For example, this can lead to the prevention of proper protein folding which is normally carried out by specific heat shock proteins. The heat-shock chaperone HSPB1, was shown to prevent TDP-43 phase transition and therefore limit its ability to form cytoplasmic condensates. Upon their formation, those condensates are difficult to remove and are probable to be phosphorylated and generate toxicity [55]. Additional findings revealed that TDP-43 protein degradation is directly affected by chaperone mediated autophagy, mainly through Hsc-70 [56]. The axonal localization and function of those heat shock proteins is yet to be discovered, but it is possible that similar utilities apply there as well.

Alternatively, a recent report by Patel et al., has identified that the mRNAs of several RNA-binding proteins, such as Khsrp, are localized to peripheral nerve axons and are locally translated in response to injury [57]. TDP-43 is known to localize into axons and to carry mRNAs, including its own [39, 40, 53]. Additionally, detailed analysis of axonal transcriptome data from motor neurons revealed that TDP-43 mRNA is abundant in axons [58]. Therefore, the possibility that also TDP-43 can be locally synthesized in axons under certain conditions, such as in response to axonal or synaptic stress, should be further investigated.

Altogether, there are several conceivable pathways by which TDP-43 could aberrantly accumulate in axons. These pathways are distinct from one another but may explain the common axonal TDP-43 pathology among different ALS disease subtypes. Still, as the pathological consequence typically associates with TDP-43 phosphorylated form, it is highly possible that phosphorylation by itself plays a key role in TDP-43 localization and entrapment in axons.

Part 2—Possible mechanisms by which TDP-43 implicates axonal toxicity

Two pivotal questions which remain open are how TDP-43 is capable of mediating excessive damage to motor neuron axons and NMJs, and if there is higher vulnerability of those distal segments to TDP-43 condensation compared to cell bodies. To address those issues, it is first necessary to understand the unique properties of motor neurons and the roles TDP-43 has in supporting their growth and maintenance. Motor neuron axons are of the longest and most arborized axons in the human body. To coordinate complex spatiotemporal processes within their highly compartmentalized cytoplasm, neurons have adapted key mechanisms such as axonal transport of essential proteins [81, 82], organelles [83, 84], mRNAs [85], and an ability to locally translate mRNAs into proteins [22]. These two processes are co-dependent on each other, as axonal transport shuttles mRNA into axons, while the local synthesis of several motor protein subunits and adaptors [86, 87], as well as proteins maintaining the microtubule infrastructure [88, 89] are required for rapid and efficient response to both intra and extracellular changes. mRNA shuttling into axons, their local translation, and their stabilization, are tightly regulated by RNA-binding proteins (RBP), which recognize and bind to specific nucleic-acid motifs [85] within RNAs to assemble RNP granules [90, 91]. RNP granules have been shown to halt translation of the mRNA within them during their shuttling and localization, supporting mRNA polarization and local translation [91]. TDP-43 contains two RRMs (RRM1, RRM2) and thus can function as an RBP, mainly through its high affinity to GU rich sequences within RRM1 [92]. Additionally, TDP-43 contains LCD at its C-terminus which drives the formation of these RNP granules [66]. Despite the fact that it is primarily nuclear, when localized in the cell cytoplasm, TDP-43 is actively transported into axons [42, 43]. TDP-43 interactome analysis revealed its binding to additional proteins associated with RNA trafficking and axon localization, including several RBPs such as Staufen-1, FMRP, G3BP1 and SMN [8, 9, 93–95].

Owing to the tremendous efforts invested in identifying TDP-43 interactors and TDP-43-bound mRNA, our knowledge of the cytoplasmic role of TDP-43 and more specifically in axons advanced in recent years. As TDP-43 is strongly associated with the formation of cytoplasmic and axonal inclusions in ALS, data from studies on these pathological forms assisted to reveal the normal context of TDP-43 at those subcellular environments. Various approaches were used to characterize the TDP-43-bound, and TDP-43-associated RNA populations in the nucleus, as well as in the cytoplasm [96, 97]. Sephton and colleagues performed TDP-43 RNA-Immunoprecipitation (RIP) from primary rat cortical neurons, followed by deep sequencing, and identified that TDP-43 binds several protein-coding RNAs involved in neurodegeneration including its own mRNA, FUS/TLS mRNA, progranulin, Ataxin1/2, and TAU. Using gene-ontology, the authors also revealed that TDP-43 binds groups of mRNAs related to synaptic function, mRNA metabolism, and neurodevelopment. Interestingly, they report that the transcripts which co-precipitate with TDP-43, contain binding motifs for the RBP PTBP2, suggesting TDP-43 and PTBP2 may collaborate in regulating TDP-43 mRNA targets.

Due to the spatial complexity of motor neurons, and considering the axonal pathology in ALS, a key unresolved question is identifying the RNAs associated with TDP-43 also in axons and NMJs. Alami and colleagues have identified that TDP-43 is a component of axonal mRNA granules where it binds and shuttles RNA [42]. TDP-43 was shown to interact with its own mRNA in an autoregulatory fashion [98], as well as with mRNAs involved in axonal cytoskeleton (Nefl, Map1b) and mitostasis process (Cox4i1, ATP5A1) [20]. Moreover, a recent report also demonstrated that the knock-down of TDP-43 disrupts axonal transcriptome and impairs local translation [21]. Recently, Liao and colleagues have suggested that RNP granules interact with Annexin A11 to hitchhike lysosomes and traffic long distances in axons [44], while others have indicated that nuclear-encoded mitochondrial mRNA transports into axons on top of mitochondria [99, 100]. Cioni et al., have reported that RNP granules and ribosomes associate with late endosomes that anchor axonal mitochondria and serve as a platform for local synthesis of essential mitochondrial proteins [101]. TDP-43 also associates with mitochondria [102] and was shown on several occasions to bind nuclear encoded mitochondrial mRNAs [20, 102]. Hence, through direct and indirect interactions, TDP-43 serves a critical function in regulating the localization and translation of mitochondria related mRNAs in axons, thus influencing distal mitochondria maintenance. These discoveries may serve to explain the many studies demonstrating dysfunctions in mitochondria activity and transport which are considered to be key pathologies in ALS and were also described in TDP-43 mutant models [20, 102–104].

Part 3 – Clinical implications of TDP-43 mislocalization in axons on ALS pathology

As years go by, scientific discoveries have promoted our understanding of ALS disease etiology, genetics, pathology and progression. Still, those achievements did not yet translate into the clinic. Therefore, it is necessary to ask, are we “looking under the lamp”?. Meaning, are current therapeutic interventions developed for ALS the most likely to succeed or the most available for testing. One key point regards the relative overlook of axonal projections and NMJs when trying to analyze ALS pathology. Until recently, the knowledge about ALS pathology was based mostly on histological sections of post-mortem CNS tissues, such as the brain and spinal cord [16, 30, 33]. Isolation of tissues from early symptomatic patients was considered complicated ethically and practically, and as a result, most patient data was not obtained from peripheral nerves and at early stages. Thus, the crucial majority of studies regarding TDP-43 based pathology and mechanistic implications were gained from cell bodies without assessing axonal projections. Furthermore, even when observing primary motor neurons or IPSC-derived motor neurons, some popular hypotheses were validated only on cell bodies without confirming their influence on axons and NMJs. However, the new discoveries of TDP-43 pathology in distal axonal projections which appear already in early stages of the disease [26, 27], suggest that this approach did not provide a complete picture. This novel data directs ALS researchers to acknowledge also the motor neurons' periphery, which remained understudied mostly due to technical reasons. Thus, the implications of TDP-43 axonal localization on motor neuron toxicity in ALS require much more attention. Many mechanisms were previously suggested to explain the effect of TDP-43 aggregation, spanning from alterations in mitochondria activity [20, 104, 136], axonal transport [137], protein translation [20, 123], stress granule formation [111, 138], aberrant splicing and genetic instability [139, 140], nucleocytoplasmic defects [37] and more. However, the knowledge about the relevance of those toxic effects on axonal and NMJ viability is less studied. Thus, a deeper look is required at how existing known pathways of TDP-43 toxicity converge to form the disease [141], especially in distal neuronal compartments.

Conclusions

The novel finding regarding the presence of pTDP-43 pathology in peripheral motor neuron axons of ALS patients provides an exciting new perspective for understanding ALS pathology and etiology, as well as an avenue for developing new therapeutic approaches. Together with recent discoveries regarding the importance of TDP-43 in axonal health and the implications of its phosphorylation on RNA biology, local protein synthesis, and mitochondrial viability, an urgent need for timely research is at hand. One possible hypothesis which will need further proof is that the initial step in the disease is enhanced axonal localization of TDP-43 at distal motor nerves. This process can lead to a buildup of pTDP-43 condensates in the NMJ, which in turn facilitates repression in local protein synthesis, mainly of mitochondrial and ribosomal proteins. The following result is the failure of the NMJ to maintain enough functioning mitochondria and ATP, resulting in the accumulation of reactive oxygen species and NMJ degeneration, delivering a retrograde signal causing motor neuron cell death (Fig. 3). It is early to determine the exact order of events, but to do that, future research must determine if the effects studied for TDP-43 in the motor neuron cell body and peri-nuclear cytoplasm occur in distal neuronal compartments and when. The next stage will be to try and utilize this new knowledge both as an anchor for therapeutic development and as a call for ALS researchers to enhance their studies of axonal and synaptic biology. Hopefully, those joined forces will be able to look for the convergence of ALS mechanisms and their effect on the peripheral nervous system. This will lead to mechanism-based therapy that will aid ALS patients in their struggles.

Open in a separate window

Fig. 3

Proposed pathological mechanism of TDP-43 in axons/NMJs. Recent evidence on both the biological and pathological roles of axonal TDP-43 imply that TDP-43 accumulates in remote axons and presynaptic compartments of motor neurons and forms pTDP-43 condensates. Those structures sequester mRNAs critical for local synthesis of mitochondrial and ribosomal proteins. The disruption in local protein synthesis of these initiates a vicious cycle between mitochondria dysfunction, generation of Reactive Oxygen Species (ROS), and further impairment of local protein synthesis. This specifically impacts the NMJ due to the strict dependency of the NMJ on those processes, thus facilitating NMJ and axon degeneration

Acknowledgements

Abbreviations

Authors’ contributions

AI, TA and EP, conceptualized and authored this review assay. Figures were prepared by AI. All authors read and approved the final manuscript.

Funding

This work was done in the support of the following funding agencies: Israel Science Foundation (ISF), Radala Foundation, Human Frontiers (HFSP).

Availability of data and materials

Not applicable.

Declarations

Footnotes

Contributor Information

Ariel Ionescu, Email: moc.liamg@reppiksdlo.

Topaz Altman, Email: moc.liamg@40namtla.

Eran Perlson, Email: li.ca.uat.xeuat@epnare.

References