TDP-43 mislocalization and aggregation are key features observed in a majority of ALS cases and approximately 40% of FTD cases [1, 50, 51]. The neuropathological and clinical findings indicate that overlapping pathogenic mechanisms involving TDP-43 proteinopathies contribute to neurodegeneration [52–54]. Despite this, replicating human ALS/FTD symptoms in rodent models has been challenging due to TDP-43's complex regulation and sensitivity to dosage changes [39, 55].
In our initial cell model experiments, we observed that even partial nuclear clearance of TDP-43 was sufficient to disrupt the balance of endogenous DNA damage and repair, leading to an accumulation of unrepaired DNA breaks in the nuclear genome (Fig. 1b-e). While Winton et al. 2008 [56] previously reported the association between the deletion of NLS sequence in TDP-43 and its enhanced tendency for mislocalization and aggregation, our study is the first to establish the connection between TDP-43 mislocalization and spontaneous genome instability in neuronal cells. This phenomenon was more pronounced under conditions that induce DNA damage stress such as etoposide (a DNA topoisomerase-II poison), due to a delayed repair of induced DNA breaks (Supplementary Fig. S1b-c) [57]. Consistently, neuronal cells expressing TDP-43-mNLS mutant were more vulnerable under oxidative stress conditions (Fig. 1h).
To investigate ALS pathogenesis further, we generated a novel endogenous CRISPR KI Tdp-43∆NLS mouse model conditionally expressing a murine Tdp-43 variant with deleted NLS (82-98aa) sequence. This model uniquely mimics the early stages of ALS, showing TDP-43 mislocalization and progressive aggregation, key pathological features of the disease progression. Our approach eliminates the limitations of previous models, such as rapid onset of unrelated motor symptoms and technical artifacts from constitutive transgene insertion.
We utilized the FLEX-based activation to target the Tardbp allele, employing Cre driver-mediated recombination of loxP-loxP or mutant loxM-loxM sites. This design ensures control over the recombinase reaction in differentiated or mature cell types, preventing Cre-loxP-mediated aberrant chromosomal rearrangements and loss of the target allele in embryonic stem cells [58]. Our endogenous KI model is thus distinct from previous ALS-TDP-43 overexpression and downregulation mice models [27, 43, 44, 59–61], and expresses an NLS-deleted Tdp-43 variant from the murine Tardbp gene’s locus, specifically in motor neurons. This approach eliminates the potential complications of transgene insertion and the development of aggressive motor phenotypes, which do not align with human ALS pathophysiology.
To closely mimic the disease progression from pre-symptomatic to symptomatic stages, we incorporated a hemizygous bigenic MN-specific Tdp-43∆NLS line in most of the analyses in this study. The bigenic Cre::Tdp-43∆NLS mice presented progressive motor dysfunctions, gait asymmetry, early-stage myogenic ALS pathology, and MN degeneration, along with pTDP-43- and ubiquitin‐positive pathology in the dorso-lateral and dorso-ventral spinal cord, reflecting early ALS symptoms [62]. This progression is more nuanced compared to milder phenotypes of neuromuscular abnormalities seen in mice with mutations in Fus, VCP, and Sod1 [63–65]. By 12 months, these mice progressively developed initial signs of gait disorders in their hind limbs, without paralysis or premature death. Notably, some males in this group experienced excessive weight gain potentially linked to Tdp-43 proteinopathy-induced abnormal fat metabolism [66], which may contribute to the observed gait disturbances.
Furthermore, our model is the first to demonstrate the link between Tdp-43 pathology accumulation of DNA break accumulation, as well as an enhanced neuroinflammatory response in both the brain and spinal cord. These aspects, while shown in various in vitro models by us [11, 21] and others [18, 20], have not been comprehensively explored in previous ALS models. Using a combination of cellular, molecular, and histopathological readouts, along with in vivo motor function tests, we demonstrate that aberrant mislocalization of Tdp-43∆NLS and its subsequent aggregation can recapitulate the key pathologic features of ALS-TDP-43. Our analysis also revealed a potential crosstalk between proteinopathy, genome damage, neuroinflammation, and neuronal senescence in this early symptomatic ALS model.
While increasing evidence points to a critical connection between genome damage and neuron loss in ALS/FTD-TDP-43 and related diseases, to date, only rNLS8 (hTDP-43∆NLS transgenic) line has shown approximately 2–3 fold of overexpression of DNA-damage inducible transcript 3 (Chop), growth arrest, and DNA-damage-inducible 45 gamma (Gadd45γ), as the early signs of cellular stress and death mechanisms [67–69]. However, the perturbed DNA repair and DNA damage response pathways were not addressed in ALS-affected neurons. In this context, our study, using an endogenous ALS-Tdp-43∆NLS mouse model, is the first to demonstrate in the brain (cortex and hippocampus) and spinal cord that nuclear clearance and subsequent aggregation of Tdp-43 pathology in a MN-specific manner can cause accumulation of DNA break foci, hyperactivation of neuroinflammatory factors such as Iba-1, Il-6, and Tnf-α, culminating in neuronal death. We also recapitulated our initial finding that aggregated mutant TDP-43 can trap DNA repair factors in the cytosol [21], thereby preventing their nuclear translocation in response to genome damage and inhibiting DNA repair processes such DNA DSB repair. Notably, although GFAP-positive astrocytes were activated or accumulated in the vicinity of damaged neurons (with DNA DSBs and protein aggregation), they did not exhibit any stress phenotypes, indicating that observed motor phenotypes were specifically caused by MN-specific Tdp-43 proteinopathy in these Mnx1-Cre::Tdp-43∆NLS mice. Detailed investigations with isolated cortical MNs from adult ALS mouse brains are critical to explore mechanisms of impaired DNA repair pathways and to develop DNA repair-targeted therapies for early-stage ALS patients. Thus, our unique ALS-Tdp-43∆NLS mouse model is poised to serve as an ideal drug screening model.
NeuN, encoded by FOX-3 gene, is an important neuron-specific transcription factor as well as RNA splicing regulator [70, 71]. NeuN also interacts with synapsin-1 that plays a pivotal role in synaptic plasticity and regulation of levels of inhibitory neurotransmitters at the synapse through synaptic vesicles [72, 73]. Interestingly, a study on traumatic brain injury using a non-transgenic mouse model has revealed that brain trauma can induce TDP-43 mislocalization, along with altered localization of NeuN from the nucleus to the cytosol in the ipsilateral region of the cortex compared to the contralateral region [74]. Such altered sub-cellular localization of NeuN has also been reported in HIV-associated neurocognitive disorders [75]. Notably, we also found that Tdp-43 mislocalization markedly altered the localization and sequestration of murine NeuN to the cytosol of stressed and degenerative neurons (clustered) [76] in the inner layers (III-V) of the motor cortex and ventral horn of the thoracic spinal cord of mutant mice through a yet unknown mechanism. Future studies should further investigate the possibility that that NeuN, being an RNA splicing factor, may interact with TDP-43 directly within a spliceosome.
Moreover, our findings highlight the potential role of neuronal senescence in ALS/FTD pathogenesis. We found that senescent neurons were positive for DNA DSB marker γH2ax and senescence probe (Fig. 9a-c), suggesting that loss of Tdp-43 function-linked DNA DSB accumulation might be a critical driver of neuronal senescence and cell death in ALS-TDP-43. This observation opens new avenues for understanding the relationship between cellular senescence, neuroinflammation, and neurodegenerative diseases. Although early senescence may confer protection to cells against lethal damage [77–79], emerging evidence suggest that neuronal senescence is one of the pivotal mechanisms contributing to neuron loss and resulting in the motor and cognitive dysfunctions in ALS/FTD [80–82]. Furthermore, there is an important crosstalk between cellular senescence and neuroinflammation. C-X-C motif chemokine receptor 2 (CXCR2) is found to increase significantly triggering neuronal apoptosis in sporadic ALS [83], and on the other hand, senescent cells activate CXCR2-mediated a self-amplifying secretory network reinforcing growth arrest [84]. Consistently, we demonstrate that senescent cells were Nissl-positive and γH2ax-positive, indicating damage-associated senescence in neuronal cells in MN-specific Tdp-43∆NLS mice brain and spinal cord. Future research should focus on dissecting the molecular characteristics of these senescent motor neurons and exploring DNA repair-targeted therapies for early-stage ALS. Our unique ALS-Tdp-43∆NLS mouse model provides an ideal platform for such investigations, potentially advancing our understanding of ALS and FTD and informing the development of novel therapeutic strategies.
Our ALS-Tdp-43 mouse model demonstrates the clear manifestation of key pathological hallmarks of ALS/FTD at the molecular level, while maintaining a non-paralytic motor deficit condition. As such, it can offer investigators unique opportunities to decipher the disease-causing early-stage pathomechanisms in MNs that might be reverted by therapeutic drugs, even in long-term treatments – a condition that is difficult sustain in other aggressive disease models.
In conclusion, establishing multiple animal models is fundamental for a comprehensive understanding of the complex and progressing disease pathogenesis of neurodegenerative disorders like ALS/FTD. Each model, including those based on overexpression or knockdown techniques, plays a crucial role in unraveling specific aspects of the functions and/or toxicity of disease-related proteins. Our model, uniquely replicating both the nuclear loss of Tdp-43 and its cytosolic aggregation – two hallmark features of ALS/FTD – adds a significant dimension to the existing array of animal models. It not only mirrors key disease mechanisms, including protein mislocalization, aggregation, and markers of TDP-43's pathological forms, but also encapsulates genome instability, inflammation, senescence, and neuronal dysfunction, along with a motor phenotype. This comprehensive representation makes our model an invaluable tool for testing new therapeutic concepts and deepening our understanding of these debilitating neurodegenerative diseases.