Amyotrophic lateral sclerosis (ALS) is among the most prevalent and fatal neurodegenerative diseases in which motor neurons in spinal cord and motor cortex are progressively lost. Patients diagnosed with ALS suffer from the gradual respiratory dysfunction, having the mean survival time of 3–5 years due to no effective therapeutic treatment (Mejzini et al., 2019). Approximately 90–95% of patients show sporadic ALS, while only less than 10% belong to familial cases (Ling et al., 2013). Frontotemporal lobar degeneration (FTLD), the second-highest incidence rates of early-onset dementia after Alzheimer’s disease, is characterized by the cumulative neuronal loss in the frontal and temporal lobes of the brain, manifesting behavioral abnormalities, personality changes, and gradual language inability (Van Mossevelde et al., 2018; Ferrari et al., 2019). It was reported that about 15% cases of FTLD show ALS symptoms, meanwhile as much as 15% of ALS patients also develop classic features of FTLD such as cognitive impairment (Ringholz et al., 2005; Wheaton et al., 2007). Such shared modality can be explained by the fact that there is a distinctively broad but also overlapping spectrum of ALS and FTLD in terms of clinical aspects, neuropathological mechanisms, and genetic mutations (Figure 1A) (Ling et al., 2013; Ferrari et al., 2019). In fact, TAR DNA-binding protein-43 (TDP-43) was found to account for the connecting pathology of more than 90% of ALS and about 50% of FTD (i.e., FTLD-TDP) cases (Ling et al., 2013). Strikingly, ubiquitinated and hyper-phosphorylated TDP-43 was identified as a primary constituent of the mislocalized and insoluble cytoplasmic inclusions in the patient brain samples (Arai et al., 2006; Neumann et al., 2006). This observation may support the idea that the proteolytic control of TDP-43 is intimately connected to the pathomechanisms in ALS and FTD (Gao et al., 2018). However, it still remains as questions of how the TDP-43 pathogenesis upends this quality controlling protein clearance system. In this review, we briefly summarize the recent progress of ubiquitination and deubiquitination mechanisms that can modulate TDP-43 protein stability, localization and stress response, which may provide the mechanistic insights into developing proteolysis-based treatment for TDP-43 proteinopathies. Detailed physiology and pathology of TDP-43 have been extensively discussed recently (Prasad et al., 2019; Keating et al., 2022).

A 414-amino acid TDP-43 protein contains an N-terminal domain with a nuclear localization signal (NLS), two RNA recognition motifs (RRM1 and RRM2) with a nuclear export signal (NES), mitochondria localization signals (M1-M5), and a glycine-rich C-terminal domain where multiple ALS-associated mutations were reported (Figure 1B) (Guerrero et al., 2016; Prasad et al., 2019). This feature defines the predominant localization of TDP-43 in the nucleus where it regulates RNA metabolism, including transcription, splicing, translation as well as stability (Pinarbasi et al., 2018; Weskamp and Barmada, 2018; Bjork et al., 2022), while being also capable of translocation to the cytoplasm to mediate mRNA transport (Chu et al., 2019) or stress granule formation (Khalfallah et al., 2018) (Figure 1C). Indeed, deletion or A90V mutation in the NLS promotes the insoluble cytoplasmic TDP-43 mislocalization and aggregation (Winton et al., 2008a; Winton et al., 2008b; Barmada et al., 2010). Two RRM domains and possibly their dimerization are necessary for the proper binding of TDP-43 to DNA/RNA molecules with high specificity towards TG/UG-rich sequences. Several studies indicated that mutations in these regions impair TDP-43’s RNA binding and splicing activities (Lukavsky et al., 2013; Kuo et al., 2014; Furukawa et al., 2016), confirming its critical role in RNA metabolism. Notably, most of ubiquitination sites on TDP-43 so far have been identified within or near these RRM domains (Figure 1B) (Kim et al., 2011; Dammer et al., 2012; Kametani et al., 2016; Hans et al., 2018). The C-terminal domain (CTD) represents a highly disordered and low-complexity domain (LCD) which consists of a glycine-rich region separated by a glutamine and asparagine (Q/N)-enriched segment (Figure 1B). The CTD has been of great interest due to its intrinsic aggregation-prone property like the prion-like domain (King et al., 2012) which apparently contributes to TDP-43-induced pathological inclusions and neurotoxicity in ALS or FTLD-TDP (Afroz et al., 2017; Santamaria et al., 2017). Furthermore, most of ALS-causing mutations, cytotoxic truncated forms, and phosphorylation sites have been associated with the CTD of TDP-43 (Figure 1B) (Hasegawa et al., 2008; Pesiridis et al., 2009; Zhang et al., 2009; Xiao et al., 2015; Berning and Walker, 2019). The N-terminal domain (NTD) also exhibits its own oligomerization or aggregation propensity (Tsoi et al., 2017), and intriguingly, the part of NTD adopts a novel ubiquitin-like fold that can bind to ssDNA (Qin et al., 2014). It should be further investigated whether this structural element can also provide extra avidity for forming aggregates or association with ubiquitin binding proteins under pathological conditions. Overall, these structural features endow multi-dimensional regulation of TDP-43 protein, in which ubiquitin signaling may also serve as a critical surveillance system for its proper functioning through proteostasis networks.

Failure of TDP-43 regulation culminates in the fatal outcome of neuronal defect obviously because its pleiotropic nature may represent the sum from any of proteostasis, RNA homeostasis, liquid condensate homeostasis (as in stress granule), and organelle homeostasis (as in mitochondria) (Keating et al., 2022). Under physiological conditions, TDP-43 predominantly exerts nuclear functions in RNA-related processes but also shuttles between nucleus and cytoplasm to participate in stress granule formation and mRNA translation (Figure 1C). By contrast, under pathological mutations or stressors, nuclear depletion of TDP-43 seems to precede, and then entails its cytoplasmic mislocalization, aggregation, and inclusion formation (Tomé et al., 2020). Also, the pathological hallmark of TDP-43 proteinopathies highlights the aberrant deposition of ubiquitinated and hyper-phosphorylated TDP-43 in the cytoplasmic inclusions, which indicates that the proteolytic control of TDP-43 must have been severely compromised (Klaips et al., 2018). Therefore, for TDP-43-induced neurotoxicity, TDP-43 loss-of-function likely occurs due to its nuclear depletion accompanying cytoplasmic accumulation, which in turn progressively drives the pathogenic aggregation and deposition of insoluble TDP-43 inclusions (Lee et al., 2012). Probably, such dramatic changes of the molecular signature inevitably drive the progression of TDP-43 pathogenesis by altering or disturbing its bona fide protein (and also RNA) interaction networks. Above certain threshold that can be held by protein quality control mechanisms–such as ubiquitin-proteasome system (UPS), heat shock response, or autophagy-lysosomal degradation pathway, TDP-43 proteopathies can become runaway or even aggravated by faulty proteostasis pathways. Moreover, aberrant accumulation of pathological TDP-43 species sequesters the critical UPS components, directly impairs the proteasome activity, induces the accumulation of insoluble polyubiquitinated proteins, and also depletes free ubiquitin pool, thereby leading to perturbed protein and ubiquitin homeostasis (Cicardi et al., 2018; Lee et al., 2020b; Farrawell et al., 2020; Riemenschneider et al., 2022). In consistent with this view, dysregulation of TDP-43 in ALS and FTLD-TDP disrupts multiple physiological events and manifests a complex set of pathomechanisms–for example, 1) the loss-of-TDP-43 function impairs a range of its nuclear functions in RNA metabolism such as alternative or exon splicing, RNA biogenesis or stability, and polyadenylation (Yang et al., 2014; Mitra et al., 2019; Ni et al., 2021); 2) the gain-of-toxicity of TDP-43 may further exacerbate the pathogenesis by sequestering other important biomolecules and organelles through undesirable interactions (Igaz et al., 2011; Zuo et al., 2021). Accumulation of cytoplasmic TDP-43 also represses the global protein synthesis in neuroblastoma models and FTD brain samples (Russo et al., 2017; Charif et al., 2020). In addition, a growing list of evidences indicate that TDP-43-induced neuroinflammation and innate immune responses are critically associated with the TDP-43 pathogenesis via NF-κβ/p65, cGAS-STING, NLRP3 inflammasome and PTP1B pathways (Zhao et al., 2015; Lee et al., 2020a; Dutta et al., 2020; Yu et al., 2020; Bright et al., 2021).

So far, over 50 missense mutations in TARDBP gene (encoding TDP-43) have been associated with ALS and FTLD-TDP (Buratti, 2015; Harrison and Shorter, 2017), most of which being clustered within the aggregation-prone CTD region (Figure 1B). Remarkably, although TARDBP mutations account for only 5–10% of familial ALS and the remaining over 90% are attributable to other genes such as C9ORF72, SOD1, FUS, and UBQLN2 (Kim et al., 2020), the majority of ALS cases (about 97%) also exhibit TDP-43-induced pathology (Ling et al., 2013). TDP-43 mutations were proposed to alter the protein stability or increase the propensity of mislocalization or aggregation. The reported half-lives of TDP-43 wild-type and mutants have been somewhat inconsistent among the literatures: while some studies observed the more prolonged turnover rate of TDP-43 mutants (Ling et al., 2010; Watanabe et al., 2013; Austin et al., 2014), others reported the faster degradation of the mutants and the C-terminal truncated forms (Araki et al., 2014; Scotter et al., 2014). Disease-associated TDP-43 mutants (e.g., G376D, G335D, G343R, A315T, and M337V) were also reported to increase the cytoplasmic TDP-43 mislocalization and stress granule or inclusion formation, which may become deleterious to neuronal cells (Jiang et al., 2016; Mitsuzawa et al., 2018; Ding et al., 2021).

While ubiquitin-enriched cytoplasmic inclusion of TDP-43 is considered to be the proteopathic signature of ALS and FTD, the pathophysiological meaning of the ubiquitination has been a long-standing question. Dramatic alteration of TDP-43 localization, stability, and post-translational modifications under the pathological conditions may disturb its genuine functional interaction networks by losing nuclear functions or gaining cytoplasmic toxicity. On top of TDP-43’s cytoplasmic mislocalization and aggregation-prone property, the initial pathogenic cue may trigger ‘butterfly effect’ and eventually the TDP-43 proteinopathy would prevail. Reduced proteasomal or autophagic activity and free ubiquitin pool depletion will badly influence the quality control of TDP-43 protein, or rather aberrant or undesirable ubiquitination of TDP-43 from proteostasis failure may only further exacerbate the disease phenotype. In fact, ubiquitination-rich pathogenic inclusions or liquid condensates, if the modification is not cleared properly, may act as avid and irreversible absorbent chambers to sequester many important proteolysis components such as ubiquitin binding proteins and proteasome. In this sense, pharmacological modulation of the pathogenic TDP-43-specific ubiquitination or the protein turnover itself may provide promising therapeutic opportunities to cope with TDP-43 proteinopathies. Remodeling or editing the ubiquitin chains or fine control of TDP-43 half-life may favor its proteolysis flux or antagonize the disease progression. It should be also noted that the total depletion of TDP-43 is not a viable option because of its essentiality. Indeed, pharmacological activation of the UPS or lysosomal pathways have been shown to bear the promise for TDP-43 clearance. For example, as noted above, a small-molecule inhibitor targeting proteasome-bound USP14 was reported to accelerate TDP-43 turnover by enhancing the proteasome activity (Lee et al., 2010). Similarly, forskolin-activated cAMP-PKA pathway increased proteasome function and remarkably reduced the levels of TDP-43 WT and its pathological mutants (Lokireddy et al., 2015). In addition, rapamycin, which works as an autophagic activator, strongly decreased the pathogenic TDP-43 species and attenuated TDP-43-induced neurotoxicity in ALS and FTD models (Cheng et al., 2015; Lattante et al., 2015). Of note, recently emerging targeted protein degradation or proteolysis targeting chimera (PROTAC) approaches have been showing great success in induced proteolysis of the undruggable targets including neurotoxic proteins (Moon and Lee, 2018). In fact, TDP-43 PROTACs were also recently reported, and await further improvement and validation (Gao et al., 2019; Hyun and Shin, 2021). In any cases, deep understanding of ubiquitinating and deubiquitinating mechanisms in TDP-43 pathology should provide the fundamental basis to develop the proteolysis-controlling therapeutic strategies for TDP-43-associated neurodegenerative diseases.