Amyotrophic lateral sclerosis, frontotemporal dementia and beyond: the TDP-43 diseases

Abstract

Ever since the significance of pathological 43-kDa transactivating responsive sequence DNA-binding protein (TDP-43) for human disease has been recognized in amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration with ubiquitin positive inclusions (FTLD-U), a number of publications have emerged reporting on this pathology in a variety of neurodegenerative diseases. Given the heterogeneous and, in part, conflicting nature of the recent findings, we here review pathological TDP-43 and its relationship to human disease with a special focus on ALS and FTLD-U. To this end, we propose a classification scheme in which pathological TDP-43 is the major disease defining pathology in one group, or is present in addition to other neurodegenerative hallmark pathologies in a second category. We conclude that the TDP-43 proteinopathies represent a novel class of neurodegenerative disorders akin to α-synucleinopathies and tauopathies, with the concept of ALS and FTLD-U to be widened to a broad clinico-pathological multisystem disease, i.e., TDP-43 proteinopathy.

Introduction

In 2006, the notion that pathological TDP-43 is involved in human disease was raised when it was identified as the major disease protein in frontotemporal lobar degeneration with ubiquitinated inclusions (FTLD-U), FTLD-U with motor neuron disease (FTLD-MND), and amyotrophic lateral sclerosis (ALS), suggesting a common pathogenesis linked to TDP-43 abnormalities in these disorders [3, 75]. Recently, detailed clinico-pathological studies on the whole spectrum of TDP-43 related neurodegeneration have become available, and have contributed to establishing the significance of pathological TDP-43 for ALS, FTLD-U and other neurodegenerative diseases [32].

TDP-43 (TAR DNA Binding Protein 43) is a 414 amino acid nuclear protein encoded by the TARDBP gene on chromosome 1 (for reviews, see [11, 15, 64, 94, 105]); its functions are not yet completely understood. TARDBP was first identified as a gene encoding a 43 kD protein that binds to the transactive response (TAR) DNA sequence of human immunodeficiency virus type 1. Subsequently, TDP-43 was also shown to be involved in the splicing of the cystic fibrosis transmembrane conductance regulator gene, the apolipoprotein A-II gene and possibly others [1, 6, 15–17, 67, 79, 95, 102–105]. TDP-43 is a highly conserved protein ubiquitously expressed in many tissues including the central nervous system (CNS) where it is present in neuronal and glial nuclei and to a lesser extent in the cytoplasm. It contains two RNA-recognition motifs and a glycine-rich carboxy terminal region that may be required for exon skipping and splicing inhibitory activity. This is consistent with the finding that the carboxy terminal domain binds to several proteins of the heterogeneous nuclear ribonucleoprotein family involved in the biogenesis of mRNA. Other studies also identified TDP-43 as a multi-functional RNA binding protein involved in: (1) exon-skipping of cystic fibrosis transmembrane conductance regulator and apolipoprotein A-II genes [15–17, 67]; (2) exon-inclusion of the survival of motor neuron gene [12]; (3) stabilization of low molecular weight neurofilament protein mRNA through a direct interaction with its 3′UTR [95]; (4) modulation of cyclin-dependent kinase 6 expression [6, 7] and microRNA biogenesis [37]. TDP-43 has also been shown to bind to the proximal promoter of the mouse SP-10 gene (acrosomal vesicle protein 1), which is involved in spermatogenesis, thereby implicating TDP-43 in the regulation of its expression [1]. Finally, other reported functions of TDP-43 include: (1) acting as scaffold for nuclear bodies (i.e. Gemini of coiled bodies) through interaction with survival of motor neuron protein [103]; (2) cell cycle regulation and apoptosis [6, 7] and (3) mRNA transport and regulation of local translation at synapses [102]. Thus, the physiological functions of TDP-43 are diverse but incompletely characterized, and they likely involve the regulation of multiple biological processes through TDP-43 binding to DNA, RNA, and/or proteins.

Since pathological TDP-43 also occurs in disorders other than ALS and FTLD-U, it is useful to separate these disorders, in which TDP-43 aggregations are the main feature, from other neurodegenerative diseases with secondary TDP-43 pathologies, yielding the following two groups: First, the major TDP-43 proteinopathies or the “TDP-43 multisystem diseases”, which include FTLD-U, FTLD-MND, ALS with (ALS-D) or without dementia, and others; and second, the neurodegenerative disorders with TDP-43 inclusions being present to a lesser degree or more localized, representing “additional” or “secondary” pathology to other protein aggregations such as tau, α-synuclein, or others. In addition to this dichotomy, we include a third group encompassing all those relevant neurodegenerative diseases characterized by minor or no TDP-43 pathology. Table 1 summarizes the current status of pathological TDP-43 related diseases, which shall be reviewed in this article. We focus especially on the spectrum of the major TDP-43 proteinopathies with ALS and FTLD-U being the conceptual starting points, as the clinical significance of the secondary TDP-43 pathology in other neurodegenerative diseases is yet to be fully established.

Table 1.

Overview of human diseases according to their pathological 43 kDa nuclear trans-active response DNA-binding protein (TDP-43) status

Major TDP-43 diseases

Amyotrophic lateral sclerosis with/without dementia, frontotemporal degeneration with motor neuron disease, frontotemporal degeneration with ubiquitin positive, tau and α-synuclein negative inclusions

Frontotemporal dementia with inclusion body myopathy and Paget disease of bone

Perry syndrome

Diseases with secondary TDP-43 pathology

Parkinsonism-dementia complex/amyotrophic lateral sclerosis on geographic isolates

Alzheimer's disease/hippocampal sclerosis

Pick's disease

Corticobasal degeneration

Argyrophilic grain disease

Parkinson's diseases, dementia with Lewy bodies, Parkinson's disease dementia

Huntington's disease

Myopathies (inclusion body myositis, oculopharyngeal muscular dystrophy, distal myopathies with rimmed vacuoles, polymyositis with mitochondrial pathology, polymyositis)

Diseases with minor or no TDP-43 pathology

Superoxide dismutase-1 linked amyotrophic lateral sclerosis

Primary lateral sclerosis

TDP-43 negative frontotemporal lobar degeneration with ubiquitin positive inclusions

Charged multivesicular body protein 2B linked frontotemporal dementia

Progressive supranuclear palsy

Multiple system atrophy

Basophilic inclusion body disease/motor neuron disease with basophilic inclusions

Neuronal intermediate filament inclusion disease

Tangle only dementia

Hereditary diffuse leucoencephalopathy with spheroids

Prion disease

Schizophrenia

Normal aging

Anoxia/Neoplasma

Other disorders yet to be studied

Major TDP-43 diseases

The concept of linking ALS and frontotemporal dementia (FTD) precedes the discovery of TDP-43 based on the shared accumulation of ubiquitin positive cellular inclusions, as well as on the clinical evidence of overlapping symptoms. Initially, in 2006, TDP-43 pathology in FTLD-U, FTLD-MND or ALS was reported only in select CNS areas [3, 75]; however there has been increasing evidence of the deposition of pathological TPD-43 aggregates in multiple brain areas in various studies [13, 23, 25, 33, 66, 77, 78]. Detailed reviews on the significance of clinical overlap and transition forms between ALS, ALS-D, FTLD-MND and FTLD-U have recently been published [26, 58, 93] and consensus criteria on the diagnosis of frontotemporal cognitive and behavioral syndromes in ALS have emerged [94]. Likewise, the spectrum of TDP-43 pathology and the associated recommended neuropathological nomenclature, including the morphological subtypes, have been the subject of detailed reviews [10, 18, 24, 54, 55, 63, 73]. We recently published a study on whole CNS TDP-43 pathology in a large cohort of cases with all the major TDP-43 proteinopathies, both according to their clinical phenotype (pure MND, FTD, or a combination of both), as well as to their morphological subtypes, in order to describe clinico-pathological findings adopting a broad scope on these disorders [32]. Herein, we conclude that there is significant overlap of both clinical and pathological features as denoted schematically in Fig. 1. In fact, ALS, ALS-D/FTLD-MND and FTLD-U may be situated at different points on one continuous and broad clinico-pathological spectrum of multisystem degenerations. We demonstrated that all the clinical groups showed widespread CNS TDP-43 pathology; however, the presence of MND was associated with a higher burden of inclusions in lower motor neurons [32]. Likewise, the presence of cognitive dysfunction was interrelated with the degree of cortical TDP-43 pathology. Upper and lower motor neurons were affected to an overall mild degree in the FTD group relative to the ALS group, and a certain degree of pathology was present in cortical brain areas in the ALS group, pointing toward a subclinical or preclinical involvement in both cases. Other than the defining clinical syndromes in the voluntary motor and cognitive domains, extrapyramidal signs were the most common clinical features, consistent with the robust pathology found in the striatum. Affect disturbances were observed, as well, and appeared to be associated with severe pathology in the amygdala. Further, we illustrated that the clinical syndrome of FTD with and without MND is associated with TDP-43 pathology accompanied by various degrees of neuronal loss and gliosis. Subcortical degeneration, such as pathology in the basal ganglia or amygdala, is usually present to a degree similar to that of cortical pathology, thereby suggesting that FTLD-U is not solely a cortical degeneration as implied by its designation. FTLD-U subtype 1 (characterized by frequent long neuritic profiles predominantly in the superficial cortical layers) appears to represent the most “cortical variant of degeneration” in comparison with subtypes 2 (with neuronal cytoplasmic inclusions in superficial and deep cortical layers) and 3 (with abundance of small neuritic profiles and neuronal cytoplasmic inclusions predominantly in the superficial cortical layers) [85]. In terms of lower motor neuron pathology, the latter two subtypes are closer to the motor neuron disease phenotype when compared with subtype 1. Similarly, it has been shown that cases with predominantly neuronal intracytoplasmic inclusions (subtype 2 [85]) can present with clinical MND in addition to FTD, whereas cases with predominantly dystrophic neurites (subtype 1 [85]) tend to show semantic dementia, and when neuronal cytoplasmic inclusions and dystrophic neurites are coupled with neuronal intranuclear inclusions (subtype 3 [85]), FTD or progressive nonfluent aphasia is common [59]. Further, FTLD-U patients with numerous neuronal cytoplasmic inclusions, as occurring in subtypes 2 or 3, have shorter survival times than those with subtype 1 [32, 38], potentially representing a link to the involvement of the lower motor neurons in decreased survival.

Fig. 1.

43 kDa nuclear trans-active response DNA-binding protein (TDP-43) multisystem diseases: clinico-pathological spectrum. Schematic illustration of the concept of a clinico-pathological spectrum of the major TPD-43 diseases extending from frontotemporal degeneration with ubiquitin positive, tau and α-synuclein negative inclusions (FLTD-U) at one end to amyotrophic lateral sclerosis (ALS) at the other. Blue arrowhead-like triangles denote clinical syndrome with motor neuron disease decreasing and dementia increasing from left to right. Color change from yellow to red in central box denotes increasing spread and severity of TDP-43 pathology in the brain and spinal cord, as an approximate estimation. Specifically, yellow represents predominant involvement of the spinal cord and red represents predominant involvement of cortical areas. Other brain areas are not as explicitly represented in this color coded diagram. MND motor neuron disease, ALS-D ALS with dementia, FTLD-MND frontotemporal degeneration with MND

From the biochemical perspective, the profile of the TDP-43 proteinopathies has been shown in sporadic and familial FTLD-U and ALS tissue to comprise ubiquitination, variable hyperphosphorylation and N-terminal truncation of the TDP-43 protein. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of sarkosyl-insoluble extracts isolated from affected cortical regions showed disease-specific bands at ∼45 kD, ∼25 kD, as well as high molecular weight aggregate smears in addition to the normal band at 43 kD [3, 19, 41, 75].

Since pathological TDP-43 is abnormally hyperphosphorylated, we and others have investigated the sites of TDP-43 phosphorylation. Notably, TDP-43 has 41 serine, 15 threonine and 8 tyrosine residues. By predictive in silico analysis some of these are potential phosphorylation sites. Using this approach, Hasegawa and colleagues made phosphorylation-specific antibodies and demonstrated that TDP-43 becomes abnormally phosphorylated at residues 379, 403, 404, 409, and 410 in a small number of cases of FTLD-U and ALS [41, 47]. In our study, we used a similar approach, and we developed and characterized rat monoclonal antibodies (mAbs) 1D3 and 7A9 raised to diphosphopeptide S409/410 of TDP-43 (CSMDSKSpSpGW) [72]. These mAbs were used to study the presence of S409/410 phosphorylation by immunohistochemistry and immunoblots in a large series of FTLD-U cases with/without MND, including familial cases with progranulin (PGRN) or valosin containing protein (VCP) mutations or linkage to chromosome 9p, as well as 18 ALS cases and other neurodegenerative disease cases with or without concomitant TDP-43 pathology. Our data demonstrating that phosphorylation of S409/410 of TDP-43 is a highly consistent feature in TDP-43 inclusions of ALS and FTLD-U confirmed the initial findings of Hasegawa and colleagues [41, 47]. Further, we extended these studies by showing phosphorylated TDP-43 in the inclusions of a more diverse and larger group of sporadic and familial forms of TDP-43 proteinopathies [72]. Physiological nuclear TDP-43 was not detectable with these mAbs by immunohistochemistry, while accumulations of phosphorylated C-terminal fragments were readily seen in Western blots in affected cortical brain regions from ALS and FTLD-U. At least 4 fragments were detected suggesting that they may represent the same C-terminal fragment with different degrees of phosphorylation, different C-terminal fragments with same sites of phosphorylation, or a combination of both. Indeed, these C-terminal fragments were often more abundant than phosphorylated full length TDP-43 in the cortex. However, in the spinal cord the predominate p409/410 species is full length TDP-43 [72], which is consistent with our previous findings that different TDP-43 species may form distinct inclusions in cortical versus spinal cord cells [46].

Ultrastructurally, several studies focusing on the TDP-43 inclusions showed granular and filamentous material (with an average width of ∼15 nm) that labeled for TDP-43 to a variable degree [41, 57, 69, 76, 77, 99].

Recent findings of mutations in the TARDP gene in cases of familial autosomal dominant and rare sporadic ALS patients further corroborate the significance of pathological TPD-43 as being mechanistically implicated in the disease process [9, 22, 35, 36, 39, 50, 52, 84, 91, 101, 108]. However, no mutations in the TARDP gene have yet been reported in either familial or sporadic FTLD-U [9, 35, 83, 87]. Significantly, many of the TARDBP variants display autosomal dominant inheritance in familial ALS patients, suggesting that they may be pathogenic mutations. To date, >25 TARDBP genetic variants have been identified, and the majority of these (including the G290A and G298S mutations we identified) are in the glycine-rich domain of TDP-43, suggesting that they may disrupt the normal exon skipping and splicing functions of TDP-43. Moreover, a number of variants involved the substitution of serine and threonine residues suggesting the possibility of aberrant phosphorylation. PGRN gene mutations cause TDP-43 pathology which is restricted to subtype 3 [8, 18, 19, 21, 90]. Despite the lack of apparent clinical or neuropathological differences between cases with and without PGRN gene abnormalities, a distinct phenotype appears to be present at the molecular level [20]. Given that the proportion of patients with a family history of dementia and/or MND (or other associated clinical features) is higher than that of patients with PGRN or TARDBP mutations, other genetic abnormalities, such as mutations in chromosome 9p or other genetic loci, might play a role as well [32]. Accordingly, TDP-43 pathology in the CNS and muscle have been shown to be associated with mutations in the VCP gene, which presents with FTD, Paget's disease of bone and inclusion body myopathy [18, 19, 74, 106]. Similarly, a recent report on Perry syndrome (characterized by early-onset parkinsonism, depression, severe weight loss and hypoventilation) with TDP-43 pathology in a pallidonigral distribution (with sparing of the cortex, hippocampus and motor neurons) links this rare TDP-43 proteinopathy with pathogenic mutations in the dynactin (DCTN1) gene [27, 107].

Recent studies reporting on elevated TDP-43 plasma levels in clinically diagnosed FTD (and AD) and increased cerebrospinal fluid TDP-43 levels in FTLD-U and ALS patients as compared with controls are intriguing, as they might offer a diagnostic ante-mortem tool and a biomarker for interventional clinical trials, but they need to be verified in larger cohorts and, ideally, confirmed by post-mortem follow up studies [28, 51, 92].

Diseases with secondary TDP-43 pathology

The recognition of TDP-43 proteinopathies as a new class of neurodegenerative disorder has raised the question as to the existence of this pathology in other disorders, since it is known that neurodegenerative diseases frequently exhibit overlapping morphological features (for review see [5]). Indeed, there now are numerous reports on the presence of these alterations as an additional pathology in entities determined by aggregation of other altered proteins. Notably, the degree of the TDP-43 pathology in these disorders is usually lower and its localization usually tends to be somewhat more restricted, as opposed to the widespread distribution found in the major TDP-43 proteinopathies.

Thus, it was shown that TDP-43 pathology is present predominantly in the medial temporal lobe limbic structures in a number of cases of Alzheimer's disease and hippocampal sclerosis [2–4, 19, 44, 45, 100]. The data on the TDP-43 immunoreactivity of Pick's bodies in Pick's disease are conflicting, but TDP-43 pathology is uncommon in this disorder and not yet verified by western blot analysis [3, 29, 43, 44, 57, 100]. For other tauopathies, namely corticobasal degeneration and progressive supranuclear palsy, TDP-43 pathology has been reported to be present in the former [3, 100], but not in the latter [100]. Argyrophilic grain disease also shows concomitant pathological TDP-43 with the distribution of these TDP-43 positive structures being largely overlapping with the tau-positive argyrophilic grains [30]. We and others recently published on pathological TDP-43 in parkinsonism-dementia complex and amyotrophic lateral sclerosis of ethnic Chamorros of Guam [34, 40]. We demonstrated that in contrast to tau pathology, pathological TPD-43 was virtually absent in ethnic Chamorro controls indicating that pathological TPD-43 distinguishes disease from control better than tau-pathology in this cohort. Biochemical analyses showed the presence of FTLD-U-like insoluble TDP-43 in Guam-parkinsonism dementia complex, but not in Guam controls. For the α-synucleinopathies—Parkinson's disease, Parkinson's disease dementia, and dementia with Lewy bodies—concomitant TDP-43 pathology has been reported [4, 44, 57, 71]. Recently, it was also shown that in Huntington's disease TDP-43 frequently co-localized with huntingtin in dystrophic neurites and various intracellular inclusions, but not in intranuclear inclusions [88]. Further, the concept of pathological deposits of TDP-43 and disease has to be widened to include sporadic or familial myopathies (i.e., inclusion body myositis, oculopharyngeal muscular dystrophy, distal myopathies with rimmed vacuoles, polymyositis with mitochondrial pathology, polymyositis) [53, 98, 106]. Taken together, the co-occurrence of TDP-43 pathology in tauopathies, α-synucleinopathies and other neurological diseases is of a variable degree and frequency compared to ALS and FLTD-U. Also, the co-localization of TDP-43 with other pathological proteins is present only in a subset of inclusions. In summary, the pathogenetic and clinical significance of the variable degree of co-existence in a given brain and the rare co-localization in a given inclusion of pathological TDP-43 with other hallmark proteins remains to be established.

Diseases with minor or no TDP-43 pathology

About 10% of ALS cases are familial with the most common genetic abnormality being mutations in superoxide dismutase-1 (SOD1). Soon after the discovery of TPD-43 as the main pathological protein in FTLD-U and ALS, it was reported by independent groups that the ubiquitinated inclusion pathology of the SOD-1-positive familial ALS does not reflect the much more common sporadic disease cases and is different from other familial cases, as the SOD-1 positive familial cases largely lack TDP-43 immunoreactivity [60, 97], not excluding the possibility that this may rarely occur [81, 96]. Thus, motor neuron degeneration in these apparently similar entities may result from different mechanisms [60]. In the few cases of primary lateral sclerosis and FTLD with upper motor neuron disease (i.e., FTLD-primary lateral sclerosis) reported so far, TDP-43 pathology in the pyramidal motor system is not a consistent feature [25, 49]. Recent publications have reported FTLD-U cases that are negative for TDP-43 [42, 62, 82]. These have been described to be heterogeneous in terms of their clinical and pathological nature, and comprise sporadic early-onset FTD with predominant behavioral and personality abnormalities. The term “atypical FTLD-U” has been suggested to define this entity, reflecting the idea of a new clinico-pathological subtype of FTLD-U occurring in 7–20% of cases with a diagnosis of FTLD-U or dementia lacking distinctive histopathology [62, 82]. Additionally, there are other rare TDP-43 negative entities, such as FTD linked to chromosome 3 due to mutations in the charged multivesicular body protein 2B gene [18, 89]. Data on the remaining neurodegenerative conditions are limited, but the few multiple system atrophy cases, assessed with only restricted CNS examinations, have been reported to be negative for pathological TDP-43 [19]. Prion diseases, basophilic inclusion body disease/motor neuron disease with basophilic inclusions, neuronal intermediate filament inclusion disease, tangle only dementia, hereditary diffuse leucoencephalopathy with spheroids have also been reported to be negative for pathological TDP-43, as is the case in a recent small report on schizophrenia [18, 19, 31, 48, 65]. Further, cardiovascular (anoxic or ischemic) brain injury and brain neoplastic conditions have not been associated with significant TDP-43 pathology [56]. Finally, the frequency or degree of TDP-43 pathology in normal or “elderly” individuals has not been definitely determined yet, but minor, or occasionally a higher, sub-clinical degree of pathology might occur [33, 34, 71] similar to α-synuclein, tau, or amyloid-beta deposition in aged individuals. Global gene expression analysis using total RNA harvested from the frontal cortex showed that controls and FTLD-U differ in over 100 networks [68]. However, the significance of TDP-43 pathology in normal controls remains to be established.

Whole CNS evolution of pathology

Many studies show that the evolution of neurodegenerative pathology throughout the CNS follows a pattern that is typical for each of the various disease groups, as has been defined for pathological aggregations of α-synuclein or tau (and others), and respective grading or staging schemes have been established to account for this temporal and spatial development of neurodegeneration. In the case of the TDP-43 proteinopathies, data on whole brain pathology with clinical correlations have become available [32, 33, 76, 77]. It appears that brain areas afflicted early on or that are associated with a rapid disease course include basal ganglia, medulla, amygdala, hippocampal formation, and spinal cord and primary motor cortex. On the other hand, the cerebellum and the occipital cortex appear to be affected to a lesser/minor degree or later in the disease course. In fact, the distribution of the pathology in ALS with/without dementia has been split into two different types by cluster analysis with one of them being the more severe and FTLD-U-like “dementia” morphological type involving the hippocampal formation, neocortex and basal ganglia, and the other one being more restricted [77]. An additional hint for a concept of temporal and spatial evolution of TDP-43 pathology throughout the brain might emerge from studying diseases in which TDP-43 is present as an additional feature, “secondary” to other disease defining pathologies, such as in Alzheimer's disease in which a progression of TDP-43 pathology was suggested, with higher order association cortices being affected following the involvement of medial temporal lobe structures such as amygdala or hippocampus [45].

Conclusions

The TDP-43 diseases represent a novel class of neurodegenerative disorders akin to α-synucleinopathies and tauopathies, and thus can be categorized according to the predominant accumulated pathological protein into two groups: the major TDP-43 diseases and those with TDP-43 aggregations being present in addition to other, disease defining, pathological proteins. Notwithstanding this, the definite pathogenetic role of TDP-43 inclusions in disease is not yet established [70, 80, 86]. Many neurodegenerative diseases other than ALS and FTLD-U can show variable degrees of TDP-43 pathology. As to the major or primary TDP-43 proteinopathies, the widespread distribution of pathological TDP-43 establishes the diffuse involvement of the CNS. The idea of a predominantly frontotemporal degeneration pattern in FTLD-U and a primarily pyramidal tract degeneration pattern in MND should be refined in favor of a broad clinico-pathological spectrum disorder involving multiple systems with differences being present at group level, but not necessarily in a given case, denoting that they may share similar disease mechanisms linked to pathological TDP-43 [14, 61, 75]. However, despite initial data suggesting that certain brain areas are particularly vulnerable to TDP-43 pathology, including basal ganglia, brainstem, mesolimbic and the corticospinal system, a staging or grading scheme of the evolutionary pattern(s) of pathological TDP-43 throughout the CNS has not yet been developed, and hence represents a major task for future research.