TDP-43 functions and pathogenic mechanisms implicated in TDP-43 proteinopathies

Abstract

Given the critical role for TDP-43 in diverse neurodegenerative diseases including amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration (FTLD-TDP), there has been a recent surge in efforts to understand the normal functions of TDP-43 and the molecular basis of dysregulation that occurs in TDP-43 proteinopathies. Here, we highlight recent findings examining TDP-43 molecular functions with particular emphasis on stress-mediated regulation of TDP-43 localization, putative downstream TDP-43 target genes and RNAs, as well as TDP-43 interacting proteins, all of which represent viable points of therapeutic intervention for ALS, FTLD-TDP, and related proteinopathies. Finally, we review current mouse models of TDP-43 and discuss their similarities and potential relevance to human TDP-43 proteinopathies including ALS and FTLD-TDP.

TDP-43 and neurodegeneration

Trans-activation response DNA-binding protein of 43 kDa (TDP-43), encoded by the TARDBP gene on chromosome 1, is a major component of tau-negative and ubiquitin-positive inclusions that characterize amyotrophic lateral sclerosis (ALS; see Glossary) and frontotemporal lobar degeneration (FTLD) linked to TDP-43 pathology (FTLD-TDP) [1]. TDP-43 aggregation and neuropathology have been observed in a spectrum of distinct neurodegenerative disorders collectively known as the TDP-43 proteinopathies, suggesting a central role for TDP-43 in neurodegenerative disease pathogenesis [2, 3]. Indeed, the identification of more than 35 missense mutations in the TARDBP gene has further implicated abnormal TDP-43 function as a cause, rather than a consequence, of neurodegeneration in ALS and FTLD-TDP [3–5]. Thus, a more detailed understanding of TDP-43 functions will be necessary to devise therapeutic strategies aimed at increasing TDP-43 activity or preventing the potential toxic effects of TDP-43 aggregates to alleviate disease onset or progression.

TDP-43 regulates mRNA expression and gene transcription

TDP-43 is abundantly expressed in nearly all tissues and is well conserved among mammals and invertebrates [6]. Structural analysis has identified two RNA-recognition motifs (RRMs) termed RRM1 and RRM2 that are capable of binding nucleic acids [7], and a glycine-rich region harboring the majority of ALS-associated mutations (Figure 1). TDP-43 binds both mRNA and DNA, thereby regulating mRNA splicing, stability and translation as well as gene transcription. Although early in vitro studies showed that TDP-43 preferentially bound RNAs via a GU dinucleotide repeat element [7], a clear consensus TDP-43 binding site in vivo has not been firmly established. However, in agreement with GU element binding specificity recent analyses of global TDP-43 binding sites showed that the most significant RNA binding occured at either uninterrupted GU repeats, or a GU-rich motif interrupted by a single adenine [8, 9]. In an attempt to identify putative TDP-43 targets, initial studies showed that TDP-43 bound and stabilized human neurofilament (hNFL) mRNA [10], promoted exon skipping of the cystic fibrosis transmembrane conductance regulator (CFTR) gene [7, 11], and promoted exon 7 inclusion of the SMN2 gene [12]; however, the significance of these TDP-43 targets with regard to disease pathogenesis is currently unclear.

Figure 1. Schematic of TDP-43 domain structure, ALS-associated mutations, and C-terminal fragments.

TDP-43 protein contains a nuclear localization sequence (NLS), two RNA recognition motifs (RRMs), a nuclear export sequence (NES), and a glycine-rich domain that harbors the majority of ALS-associated genetic mutations. More than 35 ALS mutations have been identified, with the majority of mutations found within the glycine-rich region. Those mutations mentioned in the text are highlighted in the expanded view above. TDP-43 C-terminal truncated protein was identified within insoluble aggregates from FTLD-TDP brain tissue, and the precise fragment lengths identified are indicated.

The fused in sarcoma protein, or FUS/TLS, possesses a similar domain structure to TDP-43, consisting of RRM and glycine-rich domains. FUS/TLS mutations have been identified in both familial and sporadic ALS cases, and evidence indicates that FUS/TLS aggregates accumulate in a subset of tau and TDP-43-negative ALS and frontotemporal lobar degeneration, or FTLD-FUS [13]. The possibility that these two RNA-associated proteins could physically and/or functionally interact was supported by studies showing that both TDP-43 and FUS/TLS bind and stabilize the mRNA encoding histone deacetylase 6 (HDAC6) [14, 15], a previously described microtubule deacetylase [16]. Although not fully characterized, TDP-43 and/or FUS/TLS could promote neuron survival or neuroprotection via increased HDAC6 expression, as HDAC6 has recently been implicated in mitophagy and the clearance of misfolded protein aggregates [17, 18]. Although it is conceivable that TDP-43 aggregation could directly cause mitochondrial defects similar to those observed with mutant Huntingtin and Aβ fragments [19, 20], the improper regulation of HDAC6 and other TDP-43 regulated targets could additionally compromise mitochondrial integrity. Supporting an indirect role for TDP-43 in mitochondria function, abnormal expression of the mitochondria fission/fusion proteins MFN1 and Fis1 was observed in transgenic (Tg) mice that constitutively express a human wild-type TDP-43 transgene driven by the mouse prion promoter [21]. These results support the notion that TDP-43 is either directly or indirectly linked to mitochondrial function.

A recent study performed transcriptional profiling of embryonic stem (ES) cells lacking TDP-43 and identified the RAB GTPase Tbc1d1 as a putative target, suggesting TDP-43 is critical for cellular metabolism and obesity [22], a phenotype previously linked to Tbc1d1 expression [23]. Given that Tbc1d1 is expressed in most tissues, its regulation by TDP-43 could be extended to highly metabolic non-CNS tissues, including skeletal muscle, in which TDP-43 pathology has been observed in patients with Inclusion Body Myositis (IBM) [24–27]. Thus, TDP-43 regulation of metabolic gene expression could have distinct tissue-specific effects and may explain the exquisite sensitivity of affected neurons in TDP-43 proteinopathies. Similarly, TDP-43 could uniquely regulate gene expression in other non-CNS tissues such as germ cells; TDP-43 was recently shown to repress testis specific expression of the ACRV1 gene, thereby regulating spermatogenesis [28, 29].

TDP-43 Tg mice expressing a TDP-43 mutant that is localized to the cytoplasm by virtue of a defective nuclear localization sequence (TDP-43-ΔNLS) were recently generated, and it was shown that endogenous mouse TDP-43 was depleted, providing a putative TDP-43 loss of function mouse model [30]. Surprisingly, mRNA expression of histones and histone regulatory factors, typically expressed in mitotic cells, were among a subset of genes whose mRNA expression profile was significantly altered in the brain tissue of TDP-43-ΔNLS mice. Although the significance of histone gene expression in neurons remains unclear, altered cell cycle profiles have been observed in neurodegenerative diseases [31], including the abnormal expression of mitotic cyclins and cyclin-dependent kinases (CDKs) [32] as well as cell cycle regulatory proteins such as Pin1 [33] and cdc25 [34]. Increased expression of cell cycle genes in post-mitotic neurons, in contrast to rapidly dividing cells, may promote apoptosis and neurodegeneration. If true, this suggests an exciting link between TDP-43, neuronal cell cycle gene expression, and the neuronal cell death observed during disease pathogenesis. This possibility is supported by TDP-43-mediated regulation of cyclin-dependent kinase 6 (cdk6) [35]. A more detailed analysis of TDP-43 target genes including cell-cycle regulated gene expression, particularly in diseased brains, is an ongoing area of investigation.

Two recent studies have demonstrated a more global role for TDP-43 in the regulation of RNA expression and splicing. Polymenidou et al. showed that TDP-43 is crucial for sustaining mRNA levels of synaptic proteins, choline acetyltransferase, the disease-related proteins FUS/TLS and progranulin as well others implicated in various neurologic diseases [8]. Tollervey et al. showed that TDP-43 bound RNAs that encode proteins which regulate neuronal development and survival, and identified the noncoding RNAs MALAT1 and NEAT1 as disease relevant TDP-43 targets in FTLD [36]. Future studies may further clarify the splicing regulatory role of this exciting class of non-coding RNAs in both normal and pathological states. The direct binding of TDP-43 to a wide range of critical target RNAs presented in these studies suggests that TDP-43 activity promotes and/or sustains proper neuronal function and integrity. As such, the factors that alter TDP-43 function or expression levels, including environmental stress and genetic mutations, could have profound indirect consequences on global neuronal protein expression and splicing patterns.

Finally, several reports have indicated that cells are exquisitely sensitive to TDP-43 levels and thus tightly regulate TDP-43 expression. For example, increased TDP-43 expression in Tg mouse models or cultured cells led to a prominent reduction in endogenous TDP-43 levels [30, 37]. Two recent studies have provided evidence that TDP-43 auto-regulation occurs via direct binding of the TDP-43 protein to its own 3'UTR sequence, and contributes to TDP-43 mRNA instability and degradation [8, 38]. Such a negative auto-regulatory feedback loop could be critical to maintain proper levels of functional TDP-43 protein. Indeed, ALS-associated mutations or environmental stressors that perturb such auto-regulation could promote abnormal TDP-43 expression and initiate aberrant expression of TDP-43 targets. In this regard, it was recently shown that TDP-43 regulates expression of a subset of miRNAs, including let-7b and miR-663 [39], which could feedback to regulate TDP-43 expression levels, suggesting that a complex signaling network exists to fine-tune TDP-43 expression levels, which, if de-regulated, could contribute to neuronal cell death. Thus, genetic and environmental factors that perturb the TDP-43 auto-regulatory network may contribute to neuronal dysfunction and disease pathogenesis.

TDP-43, stress granules, and pathological aggregates

Although TDP-43 regulates mRNA levels under normal physiological conditions, accumulating evidence indicates that TDP-43 is a stress-responsive RNA-associated factor. Exposure to a variety of stressors including zinc [40] or the lipid mediator 15d-PGJ(2) [41], both of which promote oxidative stress, led to accumulation of insoluble TDP-43 aggregates similar to those observed under pathological conditions [1]. However, any definitive role for TDP-43 in initiating or executing an oxidative stress response had not been characterized. More recent studies using a variety of cultured cells have demonstrated that exposure to stress causes TDP-43 re-localization to stress granules (SGs) [42–47], cytoplasmic foci that represent both sites of translationally stalled RNAs as well as active sites of RNA regulation and sorting [48, 49]. SG assembly involves a physiological protein aggregation mechanism whereby polyglutamine-rich regions of the SG initiator T-cell intracellular antigen-1 (TIA-1) promotes prion-like TIA-1 aggregation [50], providing a scaffold that drives SG formation. Moreover, SG formation is a reversible process, in which TIA-1 aggregation is inhibited by chaperone-mediated refolding via heat shock protein 70 (HSP70) [50]. In fact, the mechanism of TIA-1 protein aggregation resembles that of pathological prion protein or huntingtin, highlighting the similarity between SG protein assembly dynamics and pathological protein aggregation that occurs in neurodegenerative diseases.

Using HEK293 cells, HeLa cells, human neuroblastoma cells, NSC-34 cells (a motor neuron-like cell line), or cultured primary glia, recent studies have shown re-localization of endogenous or ectopically expressed TDP-43 to SGs in response to a variety of stresses including arsenite and sorbitol [43, 45]. Although the descriptive nature of SG localization reveals a physical incorporation of TDP-43 within SGs, whether TDP-43 has an active role within SGs or a causal role in SG assembly remains uncertain. Initial evidence in cell-based studies indicated that loss of TDP-43 function had no effect on SG assembly [42, 45]; however, Macdonald et al. suggest the kinetics of SG assembly in response to oxidative stress are significantly delayed in the absence of functional TDP-43 [46]. In addition, indirect evidence showing TDP-43 interacts with TIA-1 and regulates the expression of SG components including GTPase activating protein binding protein 1 (G3BP), which also localizes to SGs, further supports a potential role in SG assembly [45, 46]. These conflicting views as to whether TDP-43 is implicated in SG formation could reflect the dynamic and transient nature of SG formation, in which the particular cell type and stress paradigm are critical variables to be considered.

The specific type of oxidative insult analyzed introduces additional complexity underlying SG regulation, which could lead to the formation of distinct types of SGs that incorporate a unique set of RNA-binding proteins including TDP-43. For example, TDP-43 was prominently observed in sorbitol-induced SGs but showed less pronounced re-localization to arsenite-induced SGs, suggesting TDP-43 controls a subset of RNAs in response to specific types of stressors [43]. The enigmatic role of TDP-43 in SG dynamics may be further clarified with the identification of TDP-43 target RNAs that are recruited into SGs. Novel genomic approaches using RNA sequencing methods have generated potential TDP-43 target RNAs whose active regulation within SGs may be critical for neuron survival during exposure to stress [8, 9, 36]. One tantalizing TDP-43 target mentioned above is the mRNA encoding HDAC6, whose protein product itself is required for SG formation, potentially through the reversible deacetylation of tubulin [51–53]. Thus, whether TDP-43 directly or indirectly regulates SG dynamics awaits further investigation.

Similar to TIA-1 aggregation, incorporation of TDP-43 into SGs could occur via an identified poly-glutamine/asparagine (Q/N)- rich region that is potentially critical for TDP-43 self-aggregation [50, 54]. The abnormal or aberrant localization of TDP-43 to SGs could initiate pathological TDP-43 aggregation or pathogenic interactions with other SG-localized proteins, and a similar phenomenon could underlie aggregation of the related RNA-binding protein FUS/TLS [55, 56]. This suggests a potential link between SG assembly and the formation and/or emergence of pathological aggregates or inclusions in neurons and glia of ALS and FTLD-TDP CNS tissues. However, until recently, biochemical or immunohistological analysis of SGs in the brain had not been well characterized, although SGs were proposed to have overlapping functions with reported neuronal RNA granules, which are storage compartments for mRNAs composed of Staufen and Fragile-X mental retardation protein (FMRP) [57]. Using cultured neurons, Wang et al. have shown that TDP-43 incorporates into neuronal granules that co-localize with FMRP in response to repetitive stimuli [58], suggesting that significant accumulation of TDP-43-containing granules occur in the brain.

The detection of putative pathological SGs in ALS spinal cord and FTLD-TDP brain tissues was recently observed [45], characterized by phosphorylated and aggregated TDP-43 protein. Indeed, it was suggested that these structures are in fact SGs, and this was supported by co-localization studies of pathological TDP-43 aggregates with the SG markers TIA-1 and eIF3 [45]. These studies suggest that recruitment of misfolded TDP-43 into SG aggregates could be linked to disease pathogenesis. Supporting this possibility, several ALS-associated missense mutations in TDP-43 alter SG dynamics, both increasing (Gly 294Ala, Gln331Lys, Gly348Cys) and decreasing (Arg361Ser) the propensity to form SGs in cell culture models [43, 45], providing a potential pathogenic mechanism for a subset of ALS mutations. Domain mapping from several studies has indicated that the glycine-rich region harboring the majority of pathogenic mutants is required for TDP-43 association with SGs [42, 43, 45]. Indeed, expression of an insoluble, aggregated TDP-43 C-terminal fragment (a.a. 216–414) is sufficient for recruitment to SGs [45], suggesting that pathological TDP-43 species alter SG dynamics. Thus, SGs could represent a functional intersection of normal TDP-43 function and the pathological accumulation of TDP-43 inclusions, reflecting the potential for both loss and gain of function toxicity associated with TDP-43 aggregation in ALS and FTLD-TDP (Figure 2).

Figure 2. Stress-granule mediated regulation of TDP-43 and its target RNAs.

RNA bound TDP-43 complexes regulate global expression of genes required for diverse functions including auto-regulation of TDP-43 mRNA itself. Such regulation could potentially occur in the context of other RNA-binding proteins including FUS and RNA-associated factors (represented by X). Pathogenic ALS-associated mutations or conditions of oxidative insult promote recruitment of TDP-43 containing RNA complexes into stress granules via a TDP-43 Q/N rich C-terminal domain. TDP-43 is selectively recruited into specific stress granules (SGs) depending on the type of stress condition. Once recruited to SGs, TDP-43 may have a physiological role in the regulation of TDP-43-dependent RNAs during acute stress. However, pathological factors including prolonged stress or ageing, for example, could induce the formation of `pathological stress granules', representing persistent accumulation of coalesced SGs that form TDP-43 positive inclusions, characterized pathologically as TDP-43 aggregates in ALS and FTLD-TDP. The formation of pathological SGs could inhibit TDP-43 function via cytoplasmic sequestration of normal TDP-43 and/or initiate a toxic gain of function leading to mitochondrial damage and other neuronal dysfunctions.

Physiological and pathological TDP-43 interacting proteins

The identification of TDP-43 interacting proteins could provide critical insight into the functions described above for TDP-43 in gene transcription, RNA binding, and SG formation. Buratti et al. showed that the C-terminal region of TDP-43 is capable of binding directly to several proteins of the heterogeneous nuclear ribonucleoprotein (hnRNP) family including hnRNP A1/A2/B1 [59]. Recent proteomics analyses verified and extended these TDP-43 interactions to include several hnRNP family members and splicing factors [9, 44, 60]. Functional analysis determined that TDP-43 residues 321–366 are critical for TDP-43/hnRNP A2 interaction and demonstrated that a TDP-43/hnRNP complex regulates splicing activity [61]. Although the ALS-linked mutations within this region (Gln331Lys, Met337Val, and Gly348Cys) did not affect formation of a TDP-43/hnRNP complex, more subtle TDP-43 complex perturbations could result from these pathogenic mutations. For example, a slightly reduced TDP-43/hnRNP interaction occurring throughout a lifetime could progressively manifest an ALS phenotype in individuals with these mutations. Although TDP-43 physically and functionally associates with hnRNP complexes, pathological incorporation into protein aggregates is a unique aspect of TDP-43, as accumulation of hnRNPs was not observed in TDP-43 positive inclusions in FTLD-TDP brain tissue [62].

While studies have shown that hNFL mRNA is bound and stabilized by a complex of proteins that includes TDP-43, SOD1, and 14-3-3 proteins [10], the intervening RNA scaffold accounts for the observed binding between several of these components. However, immunoprecipitation analysis showed that the TDP-43 and SOD1 interaction was direct and independent of RNA. The exact SOD-1 species bound by TDP-43 is unclear, as Higashi et al. showed that mutant SOD-1 preferentially interacts with TDP-43 via a unique SOD-1 dimer interface that is exposed specifically in ALS spinal motor neurons [63]. The unique and specific binding of mutant SOD-1 to TDP-43 correlates with reduced TDP-43 solubility, suggesting that mutant SOD-1 could alter TDP-43 conformation and promote the accumulation of TDP-43 aggregates. However, given that spinal cord pathology in SOD1-positive fALS cases were demonstrated to be TDP-43-negative [64], it is currently unclear whether mutant SOD1 would in fact functionally interact with TDP-43 in pathological human brain tissue.

In addition to hNFL expression, a TDP-43 complex likely binds and regulates a large subset of mRNAs with diverse cellular functions [8, 9, 36]. Recent evidence indicates that TDP-43 associates with FUS/TLS in a 300–400 kDa complex, which regulates expression of the HDAC6 mRNA [14, 60]. It is conceivable that disruption of a TDP-43/FUS complex and abnormal mRNA expression could be linked to an ALS phenotype. Supporting this possibility, the pathogenic TDP-43 mutants Q331K and M337V showed dramatically elevated binding to FUS/TLS, potentially causing misregulation of HDAC6 expression as well as other TDP-43 targets [60]. These observations suggest that the shared association of TDP-43 and FUS/TLS with the ALS phenotype may reflect a common function in the regulation of a subset of critical target RNAs, including HDAC6.

Using a yeast screen to identify proteins that modify TDP-43-induced toxicity, Elden et al. demonstrated that TDP-43 interacts with the cytoplasmic localized Ataxin-2 protein in an RNA-dependent manner, and this interaction occurs in the cytoplasm where TDP-43 abnormally accumulates into misfolded aggregates [65]. Moreover, mutant polyglutamine (polyQ) expansions present within Ataxin-2 enhanced binding to TDP-43, suggesting a pathological interaction in which expanded polyQ repeats may facilitate accumulation of TDP-43 inclusions in ALS brain tissue [65]. Supporting a more global role for polyQ containing proteins in modifying TDP-43 function, TDP-43 co-localizes with expanded mutant Huntingtin in cell-based studies, and this recruitment required at least part of the TDP-43 C-terminal Q/N rich region [54]. TDP-43 binding to polyQ aggregates appears to be specific, as prominent co-localization or binding to other types of misfolded protein aggregates has not been observed. Extending these observations to humans, TDP-43 was detected within polyQ aggregates in Huntington's disease patients [66]. These findings raise the possibility that physical interactions specifically between TDP-43 and pathological polyQ-containing species could trigger abnormal TDP-43 conformational changes, altered solubility, and/or aberrant TDP-43 aggregation.

In addition to interactions with splicing factors and RNA regulatory proteins, TDP-43 readily self-associates to form dimers in vitro and in cell-based studies, and these findings are supported by structural analysis of TDP-43 RRM2 dimers [67]. Shiina et al. confirmed these observations demonstrating that ectopically expressed TDP-43 could associate with endogenous TDP-43, and importantly, that TDP-43 self-interaction occurs in the cytoplasm where TDP-43 inclusions are prominently detected in TDP-43 proteinopathies [68]. Protein analyses in this study using human brain indicated that dimerized TDP-43 could be detected in ALS brain lysates, suggesting multimeric species may occur pathologically in the diseased brain, although the authors suggest that more detailed analysis of control and diseased brains is required. Thus, protein self-association, commonly observed with other neurodegenerative disease-associated proteins including tau and alpha-synuclein, could underlie the pathological changes observed with TDP-43 in ALS and FTLD-TDP.

Mouse models of TDP-43 proteinopathies

TDP-43 protein levels are abundantly and widely expressed in most human tissues, including cells that comprise central nervous system (CNS) tissues such as neurons and glia. Genetic and biochemical studies suggest that the abnormal sequestration of TDP-43 within neuronal and glial inclusions observed in diseased brains could reflect loss of TDP-43 functions resulting in disease onset or progression. To address the physiological role for TDP-43, several laboratories have generated Tardbp null mice and in all instances TDP-43 was essential for embryonic development, with lethality occurring between embryonic day 3.5 and 8.5 [69–71]. Interestingly, heterozygous Tardbp mutant mice showed deficits in motor function, including decreased grip strength, without any apparent degeneration of motor neurons, suggesting that mild reductions in TDP-43 levels are sufficient to cause a motor weakness phenotype [69]. However, a more recent study using conditional Tardbp knock-out mice demonstrated that postnatal deletion led to whole body deficits including loss of body fat and a transcriptional profile linked to metabolic dysfunction and obesity in which one critical target of TDP-43 was identified as Tbc1d1 [22]. Given the lack of clear pathological changes in motor neurons in the mouse models described above, Tardbp null mice are likely insufficient models to recapitulate the pathogenesis of ALS. This insufficiency necessitated the subsequent generation of transgenic animal models of TDP-43 toxicity. The section below highlights TDP-43 transgenic mouse models (see Table 1), however, we reference a recent review for a more detailed summary of other animal models of TDP-43 [72].

Table 1.

Comparison of transgenic mouse models of TDP-43

	TDP-43 Transgene	Promotera	TDP-43 Pathology (C=cytoplasmic, N=nuclear)	Phospho-TDP-43 Positiveb	Neurodegeneration	Nuclear TDP-43 Clearingc	Phenotype	Refs.
Wegorzewska et al. 2009	A315T	mouse PrP	no	no	yes	yes	motor impairments, axonal degeneration, denervation	80
Wils et al. 2010	WT	mouse Thy-1	yes (C/N)	yes	yes	yes	motor impairments, paralysis, reduced survival	73
Tsai et al. 2010	WT	CAMKII	yes (C/N)	not reported	yes	yes	reduced learning/memory, reduced LTP, motor impairments	77
Shan et al. 2010	WT	Thy1.2	intranuclear granules	no	no	no	growth retardation, decreased large-caliber axon length, muscle atrophy	78
Xu et al. 2010	WT	mouse PrP	yes (C/N)	yes	yes	no	axonal degeneration, gait abnormalities, early lethality	21
Stallings et al. 2010	WT, A315T, M337V	mouse PrP	yes (C/ rare N)	yes	no	no	motor impairments, denervation, myopathy (in WT TDP-43 mice)	79
Igaz et al. 2011	WT, ΔNLS	CAMKII	yes (C) (ΔNLS)	yes	yes (WT and ΔNLS)	yes	motor impairments, more severe in ΔNLS mice	30

^aCell-type specific promoter driving TDP-43 transgene expression

^bAs indicated by immunoreactivity using antibodies against serines 409/410 or 403/404

^cNuclear TDP-43 clearing due to the presence of a TDP-43 transgene

Tg mice expressing a wild-type human TARDBP transgene directed TDP-43 protein expression to neurons using a Thy-1 promoter, which caused prominent degeneration of cortical and spinal motor neurons giving rise to impaired motor function and neuronal cell death, phenotypes reminiscient of ALS and FTLD-TDP [73]. Neuropathological analysis of cortical layers and spinal cord detected a small number of cytoplasmic and nuclear TDP-43 inclusions (NCIs and NIIs, respectively), which were ubiquitinated and phosphorylated on serines 409 and 410 (P409/410), both of which are indicators of pathological TDP-43 found in ALS and FTLD-TDP. Interestingly, these pathological changes were accompanied by increased caspase-3 activation and accumulation of distinct 25 kDa C-terminal TDP-43 cleavage products within the nucleus, a prominent species detected in human FTLD-TDP brain that may promote neuronal toxicity [74, 75]. However, the exact role for these fragments in driving disease pathogenesis of TDP-43 proteinopathies is unclear given that they are present in FTLD-TDP but reduced in ALS spinal cord, in which TDP-43 pathology comprises predominantly full-length TDP-43 protein [76].

A separate study directed wild-type TDP-43 expression to the forebrain using a CAMKII–promoter driven transgene [77]. Analysis of these mice indicated impaired learning and memory, attenuated long-term potentiation (LTP), and progressive motor abnormalities. Interestingly, the levels of several proteins required for synaptic signaling including phospho-ERK (pERK) and phospho-CREB (pCREB) were reduced in these Tg mice, providing a potential pathogenic mechanism associated with the observed deficits. Similar to Wils et al. [73], Tg mice expressing wild-type TDP-43 from a CAMKII-promoter driven transgene showed prominent hippocampal atrophy and neurodegeneration accompanied by activation of apoptosis and accumulation of ubiquitin-positive TDP-43 inclusions, similar to that observed in FTLD-TDP. Moreover, neurons containing NCIs in these mice showed depletion of normal nuclear TDP-43, consistent with a nuclear clearing mechanism and potential loss-of-function previously suggested [1]. Indeed, using a mouse-specific TDP-43 antibody, Igaz et al. demonstrated endogenous mouse TDP-43 clearing in either wild-type TDP-43 or TDP-43-ΔNLS expressing mice, which was accompanied by neurodegeneration [30], suggesting that perturbations of endogenous TDP-43 levels via autoregulation (see above) could represent a TDP-43 loss-of-function mechanism leading to neuronal cell death.

Tg mice expressing wild-type TDP-43 generated by Shan et al. [78] and Xu et al. [21] caused prominent motor phenotypes characterized by loss of body weight and hindlimb clasping. Xu et al. observed nuclear and cytoplasmic TDP-43 aggregates that were phosphorylated on serines 403/404, which correlated with elevated levels of ubiquitination in the nucleus and cytoplasm of spinal cord neurons. Surprisingly, although TDP-43 itself was not ubiquitinated, TDP-43 physically associated with ubiquitinated proteins. Further analysis indicated that the eosinophilic aggregates present in motor neurons were composed of abnormal clusters of mitochondria present within swollen axons of spinal cord motor neurons, suggesting Tg expression of TDP-43 causes mitochondria damage and dysfunction. Supporting this possibility, abnormal expression of mitochondria fission/fusion machinery including mitofusin 1 (MFN1) and Fis1 was observed in TDP-43 expressing homozygous mice [21], suggesting either direct or indirect alterations of mitochondrial dynamics. In accordance with mitochondrial abnormalities, Shan et al. observed cytoplasmic inclusions within motor neurons composed of accumulated mitchondria, and this phenomenon was attributed to impaired mitochondrial trafficking reflected by a marked reduction of mitochondria at nerve terminals [78].

Although mitochondria dysfunction has been linked to ALS pathogenesis, the role of TDP-43 in mediating these deleterious effects is not yet clear. In addition to the fusion/fission machinery described above, TDP-43 regulates components of mitochondria quality control including HDAC6, which could be critical for the clearance of damaged mitochondria, as evidenced in cell-based studies [14, 15]. Similarly, components of the trafficking machinery including kinesin-associated proteins such as Kif3a and KAP3, previously implicated in ALS disease progression, could be regulated by TDP-43 [78]. The future identification of the full scope of TDP-43 targets could shed light on putative TDP-43 targets involved in mitochondria dynamics and quality control mechanisms.

Given the identification of ALS-associated mutations linked to TARDBP [3], two recent mouse models were generated in which the prion promoter directed expression of TDP-43 containing the Ala315Thr or Met337Val mutations, previously identified in a small number of familial ALS (fALS) kindreds [79, 80]. In both studies, TDP-43 mutant expressing mice developed a progressive and severe motor phenotype with reduced survival. Wegorzewska et al. observed ubiquitin-positive aggregates selectively in cortical layer V and spinal cord motor neurons, but in all cases these structures were TDP-43 negative [80]. In contrast, Stallings et al. observed phosphorylated TDP-43 (P409/410) within ubiquitin-positive inclusions in spinal cord motor neurons especially from late stage mice, suggesting the TDP-43 pathology in these mice is due to accumulation of abnormal mutant TDP-43 proteins [79]. Thus, the mechanisms underlying TDP-43 mediated neurodegeneration in ALS and FTLD-TDP are currently unclear. However, future studies may clarify whether mutant TDP-43 itself is abnormally sequestered within inclusions, thereby suggesting that disease is mediated by a loss of nuclear TDP-43 function, or whether mutant TDP-43 undergoes a gain of toxic function that might have a number of deleterious consequences, such as impairments in the protein degradation machinery causing neurodegeneration (see Table 1 for a summary of mouse models)

Concluding remarks

Biochemical, neuropathological, and genetic studies have shown that TDP-43 is a major component of ubiquitinated inclusions in ALS and FTLD-TDP and has sparked intense efforts to elucidate the role of RNA-binding proteins in neurodegenerative diseases [81]. Although it is emerging that TDP-43 nuclear functions, including gene transcription and mRNA regulation, are critical for cell survival, the exact mechanisms by which TDP-43 regulates these processes remains uncertain. A detailed understanding of TDP-43 function in normal and pathological contexts is required to identify future therapeutic avenues to pursue for these devastating diseases. Figure 2 integrates the putative functions of TDP-43 with the deleterious consequences of TDP-43 pathology in neurodegenerative TDP-43 proteinopathies. The generation of TDP-43 mouse models that faithfully recapitulate key aspects of the pathogenesis of ALS and FTLD-TDP has proved challenging. However, significant progress has been made in understanding TDP-43 molecular functions including its role as a stress responsive RNA-associated factor, the identification of TDP-43-associated proteins and RNAs, and illuminating the pathological processes that post-translationally alter TDP-43 properties in the disease state (e.g., phosphorylation, C-terminal cleavage, and aggregation). Proteomic and genomic methods have spurred novel approaches to further understand TDP-43 function, and will undoubtedly provide a platform to identify mechanisms associated with TDP-43 function and dysregulation in disease. Ultimately, diverse approaches and methodologies will hopefully foster the identification of TDP-43 targeted therapeutics that could both increase TDP-43 protein activity and/or prevent the potential toxic effects associated with TDP-43 aggregation.

Glossary

ALS

Amyotrophic Lateral Sclerosis, characterized by progressive degeneration of motor neurons

dinucleotide repeat element

repetitive nucleic acid sequences consisting of two nucleotides present in DNA or RNA that provide specificity for protein binding

Eosinophilic

cytoplasmic structures staining positive with the acidic dye eosin, which can be used to characterize cytoplasmic protein aggregates in ALS and other neurodegenerative diseases

FTLD-FUS: Frontotemporal Lobar Degeneration linked to FUS pathology

a clinical syndrome characterized by progressive changes in behaviour, personality and/or language with the common feature being relatively selective degeneration of the frontal and temporal lobes. FTLD-FUS neuropathology represents a distinct subset of tau and TDP-43 negative FTLD containing pathological FUS protein aggregates

FTLD-TDP

Frontotemporal Lobar Degeneration linked to TDP-43 pathology: clinically similar to FTLD-FUS, however neuropathological analysis has shown that TDP-43 represents the pathological aggregated species

FUS/TLS

Fused in Sarcoma (FUS) protein, RRM and glycine-rich domain containing protein found in protein aggregates of a specific subtype of TDP-43-negative ALS

HDAC6

histone deacetylase 6, a microtubule deacetylase implicated in protein quality control mechanisms including removal of misfolded protein aggregates and damaged mitochondria

Hindlimb clasping

an indicator of neurological impairment in which hindlimbs are retracted towards the abdomen

Inclusion Body Myositis

an inflammatory muscle disease characterized by progressive muscle weakness and wasting

RAB GTPase Tbc1d1

a member of the TBC1 Rab-GTPase family of proteins that regulates glucose transport. Tbc1d1 has been linked to severe obesity in humans.

RRM

RNA recognition motif present in TDP-43 and FUS, which facilitates protein binding to RNA

Stress Granules (SGs)

cytoplasmic structures that recruit stalled protein-RNA complexes as a cellular response to environmental stressors

TDP-43 proteinopathies

spectrum of neurodegenerative diseases including ALS and FTLD-TDP characterized by hallmark TDP-43 inclusions in affected brain regions