TDP43 ribonucleoprotein granules: physiologic function to pathologic aggregates

ABSTRACT

Ribonucleoprotein (RNP) assemblies are ubiquitous in eukaryotic cells and have functions throughout RNA transcription, splicing, and stability. Of the RNA-binding proteins that form RNPs, TAR DNA-binding protein of 43 kD (TDP43) is of particular interest due to its essential nature and its association with disease. TDP43 plays critical roles in RNA metabolism, many of which require its recruitment to RNP granules such as stress granules, myo-granules, and neuronal transport granules. Moreover, the presence of cytoplasmic TDP43-positive inclusions is a pathological hallmark of several neurodegenerative diseases. Despite the pervasiveness of TDP43 aggregates, TDP43 mutations are exceedingly rare, suggesting that aggregation may be linked to dysregulation of TDP43 function. Oligomerization is a part of normal TDP43 function; thus, it is of interest to understand what triggers the irreversible aggregation that is seen in disease. Herein, we examine TDP43 functions, particularly in RNP granules, and the mechanisms which may explain pathological TDP43 aggregation.

Introduction

RNA-binding proteins (RBPs) are highly abundant in eukaryotic cells and control essential aspects of cellular function through their interactions with target messenger RNAs (mRNAs). These ribonucleoprotein complexes (mRNPs) may further assemble into reversible, higher order structures, referred to as mRNP granules, to regulate the localization and utilization of their associated transcripts. The dysregulation of RBPs recruited to these granules can lead to the formation of pathologic aggregates associated with numerous progressive degenerative disorders. A striking example of these disease-associated RBPs is TAR DNA-binding protein of 43 kD (TDP43), a major component of cytoplasmic aggregates found in a diverse set of neurodegenerative and progressive muscle diseases [1,2]. TDP43 is essential to cell health and plays critical roles in RNA metabolism, many of which require its recruitment into mRNP granules. Here, we examine the role of physiologic TDP43 assemblies in normal cellular function and how they may be converted to pathologic aggregates in a disease context.

Physiologic TDP43 function

TDP43 is an essential, ubiquitously expressed RNA-binding protein that is integrally involved in RNA transcription, translation, splicing, and stability [3,4]. Genome-wide RNA immunoprecipitation (CLIP-seq) studies have demonstrated that TDP43 recognizes UG-rich sequences present within approximately one third of all transcribed genes [5–7], which suggests that it broadly influences the processing of hundreds to thousands of transcripts.

TDP43 is a transcription factor, and it both localizes to sites of transcription [8] and regulates the transcription of numerous target genes [9,10]. Additionally, TDP43 strongly binds and represses the expression of transposable elements, or highly abundant mobile genetic elements that are a significant source of genomic instability [11]. TDP43 also regulates mRNA splicing [5,12], and nuclear depletion of TDP43 results in global splicing aberrations [13,14]. Moreover, TDP43-mediated splicing may regulate the stability of target mRNAs, including its own transcript, via nonsense mediated decay or alternative polyadenylation [6,15,16]. Finally, proteomics studies demonstrate that TDP43 interacts with several proteins involved in translation [17], and mislocalization of TDP43 from the nucleus to the cytoplasm results in a global reduction in protein synthesis [18].

In addition to its roles in transcription, splicing, and translation, TDP43 may be involved in the biogenesis and processing of microRNA (miRNAs). TDP43 interacts with the miRNA processing enzymes Drosha and Dicer [19], and its downregulation leads to dysregulated miRNA levels in immortalized cell lines, primary rodent neurons, and induced pluripotent stem cell (iPSC)-derived neurons [20,21].

TDP43 domains

A greater understanding of TDP43’s roles in RNA metabolism is obtained by characterizing its domains and their contributions to function. TDP43 is a 414 amino acid protein composed of four primary domains: an N-terminal domain (NTD), two RNA-recognition motifs (RRM1 and RRM2), and a C-terminal low-complexity domain (LCD) (Fig. 1).

Figure 1.

TDP43 structure and function

TDP43 consists of an N-terminus that contains a nuclear export sequence (yellow), two RNA-recognition motifs (green) and a C-terminus (blue) that encompasses both a prion-like domain and a glycine rich domain. In healthy cells (top left) TDP43 is primarily nuclear and plays critical roles in RNA processing and metabolism. However, in numerous neurodegenerative and neuromuscular diseases, TDP43 is mislocalized from the nucleus to the cytoplasm where it forms cytoplasmic aggregates. These aggregates contain several TDP43 species, and may contribute to ALS pathogenesis through a gain of toxic TDP43 function, a loss of normal TDP43 function, or a combination of the two.

The N-terminus (amino acids (aa) 1–102) is involved in TDP43 self-association [22,23]. TDP43 exists in a variety of oligomeric states in both the nucleus and cytoplasm that are likely required for normal protein function [24], although the relationship between oligomerization and specific TDP43 functions is not well explored. However, there is some evidence that N-terminal dimerization is required for splicing activity, wherein dimer disruption results in impaired alternative splicing of known target transcripts [25]. The NTD also contains a nuclear localization signal (NLS, aa 82–98); although TDP43 is primarily nuclear, it does shuttle between the nucleus and cytoplasm in healthy cells [26].

TDP43’s RNA-binding activity is mediated by its two highly-conserved RRMs, RRM1 (aa 104–176) and RRM2 (aa 192–262), that widely bind both DNA and RNA with higher specificity towards UG/TG-rich sequences [27]. These RRMs regulate several RNA metabolic processes including mRNA splicing, export, and stability [3,12]. Additionally, some interactions between TDP43 and other RBPs appear to be dependent on RNA binding [17,28], suggesting that TDP43 function is closely linked with its ability to interact with RNA.

The C-terminal LCD (aa 277–414) encompasses a prion-like domain (aa 345–366) and a glycine-rich region (aa 366–414). Several deletion studies have established the LCD as a critical regulator of protein-protein interactions [29,30], alternative splicing [3,29,30], and localization [26]. LCDs are regions that exhibit low variety in amino acid composition and are often predicted to have little secondary structure, and these disordered regions are implicated in both reversible binding and pathological aggregation in numerous RBPs [31]. The TDP43 LCD is no exception [32], and although it is canonically aggregation-prone [33], emerging evidence suggests that it also serves as a key regulator of protein solubility and folding in healthy cells [34]. Moreover, most disease-associated TDP43 mutations are found within the LCD, and we will explore how mutations in this highly unstructured region contribute to TDP43 aggregation later in this review.

Thus, the primary structural domains of TDP43 give rise to a highly versatile RBP capable of fine tuning its function to the underlying cell state and the availability of protein or nucleic acid interaction partners.

TDP43 liquid-liquid phase separation

In recent years, great interest has emerged in the principle that RNP assemblies may often exist as liquid droplets within the cell, in a phenomenon termed liquid-liquid phase separation (LLPS) [35]. Like oil in water, these RNP droplets de-mix from the cytoplasm into their own phase, which serves to concentrate proteins and RNAs into a separate cellular compartment.

Numerous reports have now demonstrated that TDP43 undergoes LLPS both in vitro and in cells, although the function of TDP43 droplets remains unknown [36,37]. TDP43 droplets in cells have been observed under conditions of physiological levels of TDP43 expression [37,38], suggesting that this phenomenon is not an artefact of TDP43 overexpression. TDP43 LLPS is mediated by LCD, which, as previously stated, harbours the majority of disease-associated TDP43 mutations [39–41]. Indeed, one hypothesis for the mechanism of pathogenicity of these mutations is that they shift TDP43 assemblies from liquid-like droplets to a more solid state, which may seed the formation of disease aggregates.

Several factors affect the phase separation of TDP43 (reviewed here [42]), including protein and RNA binding partners. For example, TDP43 binding to poly(ADP-ribose) (PAR) promotes LLPS of TDP43 in vitro [43], and the addition of RNA to TDP43 in vitro enhances TDP43 phase separation [39]. In contrast, a separate study demonstrated that RNA-binding antagonizes LLPS in cells [44], and further work showed that TDP43 acetylation, which inhibits RNA-binding, induces aggregation rather than droplet formation [45]. These diverse observations indicate that more research is needed to understand the role of RNA in TDP43 LLPS, and how different biochemical conditions for LLPS in vitro relate to the intracellular milieu. Inhibition of the proteasome and HSP70 activities, which mimic ageing, also appear to be important for regulating TDP43 LLPS in cells, and perturbations in these activities led to the formation of TDP43 aggregates [37].

Conditions which are conducive for phase separation of TDP43 may also be conducive to amyloid fibril formation [46], and furthermore, cellular perturbations which alter LLPS of TDP43 may lead to pathological aggregation of TDP43. Further research is necessary to understand the connection, if any, between physiological LLPS of TDP43 and disease aggregates of TDP43, and this may give insight into the mechanisms of TDP43 toxicity in neurological disease.

Physiologic functions of TDP43 in mRNP granules

TDP43 is essential to cell health and plays several critical roles in RNA metabolism. In some cases, TDP43 function is reliant on its assembly with RNA and other RBPs into mRNP granules. Here, we discuss the physiologic role of TDP43 in mRNP granules in the context of stress, injury, and cellular transport (Fig. 2).

Figure 2.

Physiologic functions of TDP43 mRNPs

Stress granules. Stress granules allow cells to dynamically sequester RNA transcripts (blue) in times of cellular stress. Stress granules are composed of numerous RBPs (purple) including TDP43 (green), which is essential for later stages of SG assembly. Myo-granules. Following skeletal muscle injury (pink), TDP43 forms transient, cytoplasmic assemblies. These granules contain other RBPs (orange), as well as sarcomeric transcripts that encode structural proteins essential for proper muscle formation (red). Neuronal transport granules. These granules enable the transport of mRNAs along axons and dendrites to distal compartments where they disassemble to allow the local translation of associated transcripts (orange).

TDP43 regulates stress granule dynamics

Stress granules (SGs) are transient, cytoplasmic mRNP granules that form following translational inhibition and contribute to the regulation of gene expression during cellular stress [47]. SGs contain numerous RBPs including TDP43, which interacts with both RNAs and proteins within these structures [48]. Although TDP43 is not essential for SG nucleation, it can influence SG dynamics via regulation of other SG components. McDonald et al. demonstrated that TDP43 mediates SG formation by regulating expression of the key stress granule nucleating protein G3BP, wherein G3BP is significantly downregulated following TDP43 knockdown [49]. Furthermore, G3BP overexpression in TDP43-depleted cells fully rescues SG formation [50], suggesting that TDP43 mediates SG dynamics via the differential regulation of key SG components [49].

TDP43 involvement in SG formation is also dependent on the type of stress. In response to most stressors, TDP43 remains in the nucleus and only a small amount is included in cytosolic SGs. However, certain osmotic, oxidative, and endoplasmic reticulum stressors induce cytosolic redistribution and robust inclusion of TDP43 into these granules [51–53]. Similarly, ALS-associated mutations that lead to increased cytoplasmic TDP43 localization strongly induce TDP43 inclusion in SGs [54]. Mutant TDP43 also increases the number and size of the SGs themselves [51,54], suggesting that disease-associated mutations may stabilize these structures by altering TDP43-mediated regulation of SG function and dynamics [55].

TDP43 forms transient, cytoplasmic mRNPs in response to injury

In addition to its roles in RNA processing and SG dynamics, TDP43 may be involved in the physiologic response to cellular injury. In injured skeletal muscle, TDP43 is upregulated and localized to the cytoplasm where it forms oligomeric assemblies that resolve following skeletal muscle repair [56]. These assemblies, referred to as myo-granules, contain other RBPs, as well as sarcomeric transcripts that encode structural proteins essential for proper muscle formation [56].

TDP43 is also transiently localized to the cytoplasm following neuronal injury. Both TDP43 protein and mRNA levels are significantly upregulated in axotomized sciatic motor neurons, wherein cytoplasmic protein levels peak at day 7 post-injury and return to baseline by day 28 [57]. Similarly, axonal ligation of hypoglossal neurons results in both cytoplasmic accumulation and nuclear exclusion for 7–14 days following ligation and restoration of normal TDP43 distribution by day 28 [58]. Furthermore, a stab-injury to the murine motor cortex induces the formation of cytoplasmic TDP43 granules that peak 3 days after injury and resolve within 40 days [59]. Intriguingly, ALS-related G298S TDP43 mutations result in an elevated injury response, wherein TDP43 granules are larger and more persistent [59]. Taken together, these findings suggest that reversible TDP43 mRNPs may be important for orchestrating the response to acute injury and eventual recovery, potentially through regulation of mRNAs essential for repair processes or transport of nascent transcripts to their sites of function.

TDP43 trafficks mRNAs in neuronal transport granules

Neurons are among the largest and most complex cells in the body, characterized by one or more axons and branching dendrites that extend far beyond the soma. Although most axonal and dendritic proteins are synthesized in the nucleus [60], a subset of transcripts are shuttled to distal compartments and translated locally [61]. This process is facilitated by neuronal transport granules, which form when regulatory proteins condense around specific mRNAs and suppress their translation during transport until the transcript is released at its destination. Neuronal transport granules are composed of numerous RBPs, one of which is TDP43 [62].

TDP43 is actively trafficked in neurons [63,64] and is observed in both axonal [63] and dendritic granules [64,65]. Moreover, Alami et al. demonstrated that these TDP43-positive granules undergo bidirectional, microtubule-dependent transport and facilitate delivery of target mRNAs to distal neuronal compartments [66]. Mutant TDP43 exhibits defective axonal transport via decreased anterograde movement and a depletion of TDP43-containing granules from the distal axons [66,67], as well as defective trafficking of the TDP43 target mRNA Neurofilament-L [66]. As such, TDP43 mRNP granules play a role in the transport of target mRNAs in healthy cells.

Prevalence of TDP43 aggregates in disease

In contrast to reversible, physiologic granules, TDP43 aggregates are a hallmark of numerous degenerative diseases affecting both neurons and skeletal muscle [1,2]. The presence of hyperphosphorylated, ubiquitin-positive, insoluble, cytoplasmic TDP43 inclusions in affected patient tissue is broadly termed ‘TDP43 pathology’.

Neuronal TDP43 pathology is a hallmark feature of the neurodegenerative disorders amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) [68,69]. Although mutations in the gene encoding TDP43 (TARDBP) account for only a small proportion of the disease burden in both disorders (2–5%), cytoplasmic TDP43 inclusions are observed in >90% of individuals with ALS and >45% of individuals with FTD [70]. This suggests that the formation of TDP43 aggregates is not primarily caused by changes to the protein via disease-associated mutations, but rather by the dysregulation of TDP43 homoeostasis [71–73]. Moreover, mutations in several other ALS/FTD-associated genes – including C9orf72 [74], ANG [75], TBK1 [76], PFN1 [77], UBQLN2 [78], VCP [79], and hnRNPA2/B1 [80] – result in TDP43 pathology, suggesting that TDP43 dysregulation is a convergent downstream consequence of multiple aberrant pathways. In support of this conclusion, TDP43 proteinopathy is described as a secondary pathology of several other neurodegenerative diseases including Alzheimer’s disease, Parkinson’s disease, and Huntington’s disease [81–84]. Moreover, TDP43 pathology is not limited to neurons. TDP43 aggregates have been detected in the skeletal muscle of patients with several muscle disorders including inclusion body myopathy (IBM) [85,86], oculopharyngeal muscular dystrophy (OPMD) [87], distal myopathies with rimmed vacuoles [87], and myofibrillar myopathies [86] despite a lack of disease-associated TDP43 mutations. Intriguingly, post-mortem immunohistochemistry studies have detected TDP43 aggregates in the brains of up to 40% of cognitively normal subjects over the age of 65 [88–90], suggesting that accumulation of cytosolic TDP43 may be related to the impairment of protein clearance pathways with age that are exacerbated in a variety of disease contexts.

Although TDP43 pathology is observed in numerous disorders, how and if TDP43 aggregates contribute to disease progression remains unclear. Numerous studies indicate that the mislocalization of TDP43 into cytoplasmic aggregates leads to a loss of nuclear TDP43 function, and the resultant dysregulation of mRNA processing has widespread deleterious effects on cell health [5,6]. Further studies indicate that the TDP43 aggregates themselves are inherently toxic and sequester functional TDP43, transcripts, and other RBPs [91–93]. Overall, TDP43 pathology likely contributes to cellular toxicity through a combined gain and loss of toxic function (reviewed here [2,94,95]). As such, further investigation is required to identify early stages of TDP43 pathology, and to identify conditions under which aggregation occurs. Given that TDP43 is a key component of both physiologic mRNP granules and pathogenic aggregates, here we explore how functional TDP43 granules may serve as precursors to TDP43 pathology.

Transformation of functional TDP43 granules to pathogenic aggregates

An appealing model is that the formation of RNP granules creates a high local concentration of TDP43, which may increase the probability of TDP43 aggregation [96,97]. Here, we explore potential mechanisms by which physiologic granules may be converted to the pathologic aggregates observed in numerous neurodegenerative and progressive muscle disorders (Fig. 3).

Figure 3.

Conversion of physiologic mRNPs to pathologic aggregates

Reversible mRNPs assemble from mRNAs (blue), TDP43 (green), and other RBPs (purple) in response to stress, injury, and during neuronal transport. These granules may be stabilized by numerous factors including TARDBP mutations (orange), C-terminal fragments, or chronic inflammation, resulting in the formation of irreversible aggregates (dark green). These aggregates may inappropriately sequester WT TDP43, other RBPs, and transcripts, further disrupting TDP43 cellular function.

Do TARDBP mutations influence TDP43 mRNP disassembly?

While TDP43 readily forms reversible assemblies in vitro and in vivo, disease-associated mutations increase its propensity to form irreversible aggregates [98]. Of the >50 identified ALS/FTD-linked TDP43 mutations, the vast majority are located within the LCD [99,100]. This unstructured domain is a critical regulator of protein-protein interactions [29,30], splicing [3,30], and localization [26], and it is heavily implicated in TDP43 aggregation due to its propensity to adopt β-sheet-rich conformations that pack tightly into higher order structures [33]. In support of this, Hughes et al. recently demonstrated that several short segments of the TDP43 LCD form stable kinked structures termed low-complexity aromatic-rich kinked segments (LARKS) in vitro [101]. LARKS are formed by pairs of kinked β-sheets which interact closely with aromatic residues, forming interactions with the backbones of the mating sheet. While WT TDP43 forms labile LARKS structures in vitro, the mutants ³¹²NFGEFS³¹⁷ and ³¹²NFGpTFS³¹⁷ form irreversible aggregates [32]. Additionally, WT TDP43’s LCD is capable of forming hydrogels in complex with single-stranded DNA (ssDNA), while the mutant LCDs form irreversible precipitates at same protein and ssDNA concentrations [102]. These data suggest that some ALS-linked mutations in the LCD may disrupt TDP43’s normal ability to reversibly oligomerize in complex with nucleic acids [102]. Consistent with this prediction, ALS-linked LCD mutations alter the biophysical properties of TDP43 neuronal granules leading to altered neuronal transport granule function [67].

Although the TDP43 LCD is canonically aggregation-prone, emerging evidence suggests that it serves chaperone-like functions to regulate protein solubility and folding in healthy cells. The LCD protects against misfolding and self-aggregation by enabling reversible phase transitions during conditions of supersaturation [34,103], permitting higher local protein concentrations than would otherwise be possible without misfolding and/or aggregation. In support of this, several RBPs aggregate upon removal of their LCDs, including PUB1, PAB1, SUP35, and TDP43 itself [34,103–105]. Taken together, this suggests that LCDs play a role in solubilizing proteins either through their hydrophilic properties or interactions in cis, and ALS-associated mutations in the TDP43 LCD may facilitate aggregate formation through both the combined effect of forming stabilizing interactions and reducing protein solubility.

In addition to modulating TDP43’s aggregation propensity and solubility, disease-associated mutations may stabilize physiologic mRNP granules by increasing cytoplasmic levels of TDP43 through either impaired nuclear import or increased protein half-life [106,107]. The A90V mutation in the NLS increases cytoplasmic localization of TDP43, which is accompanied by increased formation of inclusions that also sequester the WT protein [54,108]. Additionally, the G290A, A315T/E, and N390S mutations increase the levels of cytoplasmic TDP43 and lead to neuronal death [109,110]. Moreover, other ALS-linked TDP43 mutations, including Q331K, G298S, and M337V, appear to increase TDP43 protein half-life [107]. These results demonstrate that several disease-associated TDP43 mutations can increase TDP43 aggregation, decrease its solubility, and increase its abundance in the cytoplasm, potentially disrupting the kinetics of physiologic granules to form pathologic aggregates.

Do TDP43 cleavage products aberrantly stabilize TDP43 granules?

A well-studied, yet confusing property of TDP43 is its cleavage into smaller, C-terminal fragments. Cleaved TDP43 fragments ranging from approximately 18 to 35 kDa in size are often detected in ALS and FTD patient aggregates by immunoblotting [69,111], although the potential accumulation of such C-terminal fragments in muscular TDP43 proteinopathies has yet to be explored.

These C-terminal fragments may be more prone to aggregation due to the loss of the NTD. As previously discussed, TDP43 exists as a variety of oligomeric species in the cytoplasm of healthy cells and homodimerizes via its NTD. NTD-mediated TDP43 oligomerization is hypothesized to spatially separate the aggregation-prone LCD of each monomer [25], limiting TDP43 aggregation. This suggests that free-floating C-terminal fragments may disrupt and stabilize these physiologic oligomeric assemblies. In support of this, overexpressed TDP43 fragments are more aggregation-prone than the full-length protein and form highly phosphorylated, insoluble inclusions [112]. Moreover, these fragments are shown to sequester full-length TDP43 into stable, cytosolic aggregates [112–115]. Together, these results suggest that cleaved TDP43 fragments may confer toxicity by stabilizing pre-existing TDP43 assemblies, thereby leading to the disruption of various cellular processes including RNA processing [113], splicing [116], and neurite outgrowth [115].

C-terminal fragments may also disrupt SG dynamics. In cell culture, these fragments are recruited to SGs when exogenously expressed [54,117,118] or when fragmentation is induced by oxidative stress [52]. However, the 25 kD species is less enriched in SGs than the 35 kD fragments or full-length TDP43⁵⁴[119]. This may be attributable to the fact that the 25 kD fragment lacks RRM1 and RRM2 [120], further supporting the claim that that both RRM1 and the LCD are required for TDP43 recruitment to SGs [48]. In addition, fragments with intact RRMs are less aggregation-prone than those without, indicating that RNA-binding activity may mitigate aggregation [121]. However, further studies are required to determine if C-terminal fragments alter SG disassembly kinetics. Taken together, these studies suggest that aggregation-prone C-terminal fragments can bind to full-length TDP43, thereby altering normal TDP43 binding dynamics and disrupting critical cellular processes.

Does chronic inflammation alter TDP43 disassembly kinetics?

Although TDP43 assemblies likely play a role in orchestrating the physiologic injury response, it is unclear if repeated injury can shift these transient assemblies into persistent, pathologic aggregates. In many cases, injury elicits inflammation, a complex and coordinated immune reaction that aids in tissue regeneration [122]. However, in the context of chronic injury immune cell modulation can go awry, leading to fibrosis [123,124] and impaired tissue repair [122]. Moreover, inflammatory mediators can lead to protein misfolding [125], suggesting that chronic inflammation resulting from either injury or disease may perturb normal TDP43 function and exacerbate TDP43 pathology.

As discussed above, myo-granule formation and dissolution is a physiologic response to muscle injury. However, these granules share structural characteristics with disease-associated amyloid oligomers and are capable of seeding TDP43 aggregation in vitro [56]. This suggests that the dysregulation of myo-granule formation and/or clearance may lead to TDP43 pathology, where repeated injury may increase myo-granule persistance [126]. Moreover, increased inflammation associated with chronic injury may lead to protein misfolding that stabilizes these structures [125].

Iterative injury and inflammation may also stabilize injury-induced TDP43 granules in neurons. Although characterization of these granules is limited, injury does result in TDP43 pathology. Work by Johnson et al. shows that TDP43 pathology is present in 87% of post-mortem brain tissue samples collected within 14 days of severe TBI [127]. Moreover, TDP43 mislocalization was less extensive in post-mortem tissue from patients that survived a year or more following TBI [127], suggesting that cytoplasmic TDP43 that persists after injury is cleared slowly over time. However, patients that experience repetitive brain trauma, such as those with chronic traumatic encephalopathy (CTE), show persistent TDP43 pathology in widespread regions of the brain [128]. Moreover, general inflammation increases with age, and chronic low-grade inflammation is observed in numerous neurodegenerative disorders [129]. Even in the absence of TBI, inflammation is a potential driver of neurodegenerative changes and cognitive decline by triggering amyloid beta accumulation and tau phosphorylation [130], and similar mechanisms may result in TDP43 misfolding or post-translational modifications. Taken together, inflammation resulting from injury or disease may lead to conformational protein changes that shift injury-induced TDP43 granules to pathologic aggregates.

Are faulty protein clearance pathways to blame?

TDP43 pathology may also be attributed to a failure to clear misfolded protein. Normally, proteostasis is maintained by a network of cellular mechanisms that monitors protein folding, concentration, and localization. If a protein is misfolded, it is identified and degraded by protein clearance systems. However, when these pathways fail to maintain proteome stability, misfolded TDP43 may accumulate and drive the conversion of physiologic granules to pathologic aggregates.

The ubiquitin–proteasome system (UPS) is the primary proteolytic pathway in eukaryotic cells [131], in which targeted proteins are tagged with sequential ubiquitins that mark them for degradation by the proteasome [132]. While ubiquitinated TDP43 is not frequently detected in healthy brains [133], ubiquitinated TDP43-positive inclusions are a hallmark feature of both ALS and FTD [68,69]. These aggregates also include other components implicated in protein clearance, such as ubiquilin2⁷⁸ and p62 [134,135]. Moreover, mutations in ubiquilin2⁷⁸ and the degradation pathway adaptor valosin containing protein (VCP) [136] are causative for ALS and FTD, suggesting that a failure of the UPS contributes to TDP43 pathology.

A second major protein degradation pathway is macroautophagy or ‘autophagy’, a catabolic mechanism where intracellular material is engulfed in vesicular structures called autophagosomes which then fuse to lysosomes for degradation. Autophagy degrades insoluble and aggregated protein substrates that are too large to be degraded by the proteosome [137], and it has been suggested to be the dominant pathway responsible for degrading aggregated TDP43 [138,139]. Although autophagy was initially considered to be a non-selective system, it has been shown that ubiquitin can direct the autophagic degradation of specific proteins [140,141] suggesting that ubiquitinated TDP43 may be selectively targeted by both the UPS and autophagy.

Finally, TDP43 turnover may also be regulated by the endocytic pathway, in which cells absorb extracellular materials by invagination of the plasma membrane. The resultant vacuoles are then transported to lysosomes where their contents are degraded. In a recent study, Liu et al. demonstrated that TDP43 colocalizes with endocytic proteins and that impairment of endocytosis increases TDP43 toxicity, aggregation, and protein levels [142]. Moreover, disease-associated mutations in a key endocytic pathway components including FIG4 [143], CHMP2B [144], and Rab5 [145] have been associated with rare forms of ALS and ALS syndrome, and TDP43 colocalizes with Rab5 in ALS patient tissue [142]. Although further studies are required to determine how TDP43 is trafficked by the endocytic pathway, the dysregulation of endocytosis is clearly implicated in TDP43 dysregulation.

Taken together, these data suggest that failed TDP43 clearance is a primary driver of aggregate formation. Several studies have shown that inhibition of either the UPS or autophagy in cell culture results in the accumulation of aggregated, polyubiqutinated TDP43 [138,146], and when both the UPS and autophagy are functional within a cell, large TDP43 aggregates can be cleared over time [138]. Additionally, stimulation of the endocytic pathway is also sufficient to clear TDP43 aggregates and rescue toxicity [142]. Thus, prolonged failure of these protein degradation pathways to clear misfolded TDP43 may contribute to the formation of pathogenic aggregates.

Impairment of these protein clearance pathways may be partially attributable to age. Disrupted proteostasis is a hallmark of normal ageing, and this phenomenon is observed in numerous mammalian cell and tissue types [147,148]. Age-related impairment of the UPS has been attributed to deterioration of the proteosome due to changes in proteosome maintenance [149] and assembly [150], as well as its inactivation by protein aggregates [151,152]. Autophagy is also impaired with age due to both a decrease in autophagic inductors [153,154] and a diminished cellular response to autophagy inductors [155]. Moreover, decreased autophagy may be a consequence of inefficient clearance of the autophagosome cargo by the lysosome, leading to accumulation of undigested products [156]. This progressive deterioration of the intracellular quality control systems with age has been proposed as the main reason for the late onset protein accumulation in both neurodegenerative disorders and progressive myopathies [157]. As such, faulty clearance pathways may lead to increased cytoplasmic levels of misfolded TDP43 that sequester functional protein or alter the dynamics and function of mRNP granules.

Concluding remarks

TDP43 is a critical regulator of several mRNA processing steps, including transcription, translation, splicing, and stability. Some of these functions are dependent on the recruitment of TDP43 to mRNP granules, which further allow TDP43 to contribute to the regulation of cell stress, injury response, and mRNA transport. However, in specific contexts these high local concentrations of TDP43 may seed the protein aggregates found in numerous neurodegenerative and progressive muscle disorders. There are numerous pathways through which this may occur, some of which are inherent to TDP43, wherein disease-associated mutations increase TDP43 aggregation, cytoplasmic mislocalization, or half-life and stabilize physiologic granules. Alternatively, these granules may be stabilized by upstream processes, wherein inflammation or faulty protein clearance increase protein misfolding and cytosolic accumulation which, in turn, disrupts normal mRNP dynamics. Regardless, TDP43 involvement in various granules is essential for cell health, and therapies aimed at TDP43 pathology should not seek to interfere with TDP43’s abilities to form complexes. Rather, therapeutics aimed at preventing the conversion of physiologic mRNPs to aggregates may be a viable therapeutic strategy to retain TDP43 functionality and avoid pathologic aggregate formation.

Funding Statement

This work was supported by funds from the National Institutes of Health (grant R01 GM45443 to R.P.). R.P. is an investigator of the Howard Hughes Medical Institute.

Disclosure statement

We have no relevant financial or non-financial competing interests to report.