Unmasking the skiptic task of TDP‐43

Abstract

The mechanism by which mutations in TAR DNA‐binding protein 43 (TDP‐43) cause neurodegeneration remains incompletely understood. In this issue of The EMBO Journal, Fratta et al (2018) describe how a point mutation in the C‐terminal low complexity domain of TDP‐43 leads to the skipping of otherwise constitutively conserved exons. In vivo, this mutation triggers late‐onset progressive neuromuscular disturbances, as seen in amyotrophic lateral sclerosis (ALS), suggesting that TDP‐43 splicing gain‐of‐function contributes to ALS pathogenesis.

TDP‐43 is a multi‐faceted protein mostly found in the nucleus where it regulates the splicing of a number of pre‐mRNAs (Polymenidou et al, 2011). A decade ago, TDP‐43 was identified as the constituent of neuronal cytoplasmic inclusions characteristic of amyotrophic lateral sclerosis (ALS) and fronto‐temporal dementia (FTD) (Neumann et al, 2006). These are neurodegenerative diseases with a relatively short‐term, fatal prognosis that, intriguingly, co‐occur in many patients (Gao et al, 2017). Furthermore, a subset of familial cases of ALS and FTD harbours mutations in TARDBP, the gene encoding TDP‐43, which reinforces the notion of a pathological continuum between the two conditions. Whether mutated or not, the aggregation of TDP‐43 in the cytoplasm is accompanied by a complete nuclear clearance of the protein—and thus loss of its nuclear function—and this could participate in neurodegeneration (Fig 1A). Consistently, reduced expression of TDP‐43 has been shown to generate an ALS‐like motor phenotype (Kraemer et al, 2010; Yang et al, 2014). However, recent studies found that neurodegeneration can be induced by mutant TDP‐43 overexpression in the absence of cytoplasmic aggregation and nuclear clearance, suggesting that these events are not the sole contributors to toxicity (Arnold et al, 2013). At present, how mutations in TARDBP trigger ALS or FTD remains a mystery. The task is made even more complex by the lack of appropriate animal models. In fact, transgenic overexpression of either wild‐type or mutant TDP‐43 affects motor neurons in both cases, even when the exogenous expression rates approach physiological levels (Arnold et al, 2013).

Figure 1. Endogenous point mutation in TDP‐43 LCD domain leads to motor neuron death, by promoting skipping of constitutive exons and ubiquitin and p62 pathology.

(A) Under physiological conditions, TDP‐43 (green) is primarily located in the nucleus. In most ALS and FTD patients, TDP‐43 is aggregated in the cytoplasm, leading to complete absence in the nucleus. Whether ALS/FTD is caused by loss of TDP‐43 nuclear function (LOF) or newly gained cytoplasmic function (GOF) remains unknown. In LCDmut mice, TDP‐43 appears normally localized in the nucleus, yet motor neuron degeneration is observed and typical ubiquitin and p62 pathology is present. These findings suggest that LCDmut mice recapitulate hallmarks of ALS. (B) The M323K mutation of TDP‐43 (LCDmut) modifies its splicing activity leading to excision of otherwise normally conserved exons. The loss of these so‐called skiptic exons is associated with the dysfunction of a number of cellular processes, including regulatory mechanisms of proteostasis (ubiquitin/proteasome and p62) that could underlie neurodegeneration.

These studies illustrate that elucidating the pathological role of TDP‐43 in ALS and FTD will depend on methods that preserve endogenous expression levels. The work by Fratta et al (2018) published in this issue of The EMBO Journal reports two new mouse models carrying endogenous point mutations in the murine Tardbp gene. The first mutation, F201I (or RRM2mut), is located within the second RNA recognition motif of TDP‐43 and impairs its RNA binding capacity and splicing activity, leading to the aberrant expression of otherwise non‐conserved cryptic exons (i.e., exons that are normally excised under physiological conditions). As expected, homozygous RRM2mut mutation is embryonically lethal, similar to that observed in full Tardbp knock‐out mice (Kraemer et al, 2010; Sephton et al, 2010). RRM2mut thus represents a dose‐dependent loss‐of‐function mutation, reminiscent of the situation generated after reduced TDP‐43 expression.

The second mutation, M323K (or LCDmut), is located within the C‐terminal low complexity glycine‐rich domain of TDP‐43, in the vicinity of most of the TARDBP gene mutations related to ALS (Gao et al, 2017; Fratta et al, 2018). Interestingly, LCDmut mice display mild progressive neuromuscular modifications, including decreased muscle strength and altered electrophysiological properties at the neuromuscular junction, as well as loss of motor neurons in the presence of markers of neurodegeneration, such as ubiquitin and p62. These findings are not without precedent and are indeed similar to those described in comparable mouse lines bearing mutations within the endogenous Fus gene, which encodes another ALS‐related RNA binding protein (Scekic‐Zahirovic et al, 2016, 2017; Devoy et al, 2017). In contrast, a very recent report on a new Q331K Tardbp knock‐in mouse strain did not reveal any motor neuron degeneration, even in old mice, but rather showed prominent FTD‐like alterations, including hyperphagia and attentional deficits (White et al, 2018). Taken together, these studies describe different pathological conditions of mice that likely reflect the phenotypic diversity of the expression of identical or almost identical mutations causing ALS or FTD, or both, in humans.

Following the characterization of the motor phenotype of LCDmut mice, Fratta et al (2018) attempted to understand the mechanism whereby LCDmut causes neurodegeneration. Unexpectedly, it could not be attributed to an increase in the aggregative potential of the protein in vivo, since no TDP‐43 aggregates were observed (Fig 1A). Once again, this finding is reminiscent of what has been recently reported for Fus and Tardbp knock‐in mouse lines (Scekic‐Zahirovic et al, 2016, 2017; Devoy et al, 2017; White et al, 2018). Moreover, the motor phenotype observed in LCDmut mice could not be attributed to TDP‐43 nuclear depletion neither, since protein levels in the nucleus of the mutant animals stayed in the physiological range (Fig 1A). Combining several transcriptomics approaches, the authors analysed splicing events on a large scale. Interestingly, a number of constitutive exons (i.e., exons that are normally retained during the splicing process), most of them flanked by TDP‐43 binding sites, were found consistently skipped in LCDmut mice compared to wild‐type animals (Fig 1B). The absence of these so‐called skiptic exons led to a generalized decrease in a variety of transcripts, likely as a consequence of their increased instability. This phenomenon particularly affected transcripts involved in the ubiquitin proteasome pathway, which could account for the ubiquitin and p62 alterations observed in the spinal cord of LCDmut mice (Fig 1A and B). Most importantly, a subset of these aberrant splicing events was observed in patient‐derived cells with TDP‐43 mutations, suggesting that this new gain‐of‐function mechanism could, at least in part, contribute to disease pathogenesis in cases of ALS (or FTD) caused by mutant TDP‐43.

Finally, another important original aspect of the study by Fratta et al (2018) is that the newly identified gain‐of‐function mechanism, leading to increased skipping of constitutive exons, mirrors previous studies documenting loss of TDP‐43 function whereby loss‐of‐function leads to decreased excision of non‐conserved cryptic exons (Ling et al, 2015). While both gain‐of‐function and loss‐of‐function are likely to have strong biological consequences on the cellular proteome, they affect two different sets of genes. It is thus tempting to speculate that mutations in TDP‐43 may induce a two‐step neurodegeneration process. First, an exclusion of skiptic exons may affect central cellular processes such as protein degradation and drive the first stage of neurodegeneration. This, indirectly, may lead to protein (including TDP‐43) aggregation, TDP‐43 nuclear depletion and loss of TDP‐43 function, all associated with increasing occurrence of cryptic exons, along with late stage of neurodegeneration. In ALS patients that do not carry any TARDBP mutation but display TDP‐43 pathology, TDP‐43 nuclear depletion and cytoplasmic aggregation, leading to TDP‐43 loss‐of‐function and increased inclusion of cryptic exons, could be the primary driver of neurodegeneration. Such working models now need to be experimentally challenged in order to determine their biological relevance and potential, and to allow the design of therapeutic strategies for the devastating ALS and FTD.

The EMBO Journal (2018) 37: e99645