TDP-43 in aging and Alzheimer's disease - a review

Abstract

Transactive response DNA-binding protein of 43 kDa (TDP-43), an RNA and DNA binding protein involved in transcriptional repression, RNA splicing and RNA metabolism during the stress response, is the major component of neuronal inclusions in amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration with ubiquitin inclusions, now referred to as FTLD-TDP. While initially thought to be relatively specific to ALS and FTLD-TDP, TDP-43 pathology has now been detected in a number of other neurodegenerative diseases, many associated with tau pathology, including Guam Parkinson dementia complex and Alzheimer's disease (AD). TDP-43 pathology is detected in 25% to 50% of AD cases, especially those with more severe clinical phenotype and greater Alzheimer type pathology, as well as AD cases with hippocampal sclerosis (HS). HS is characterized by selective neuronal loss affecting CA1 sector of the hippocampus, and most cases of HS, with or without AD, have TDP-43 pathology. Whether TDP-43 pathology is merely an incidental finding in AD or actually contributing to the more severe clinical phenotype remains unresolved. Presence of TDP-43 in normal elderly, who are at increased risk for AD, would strengthen the argument that it is not merely a secondary or incidental finding in end stage AD. Limited studies suggest that TDP-43 pathology is infrequent in neurologically normal elderly (3% or less). We provide an overview of what is known about TDP-43 in AD, normal aging and in other disorders and suggest that TDP-43 proteinopathies be considered in two classes - primary and secondary.

Introduction

Transactive response DNA-binding protein 43 (TDP-43) was initially described based upon its binding to the regulatory element, TAR, in the human immunodeficiency virus type 1 (HIV-1) long terminal repeat, ultimately affecting HIV-1 gene expression [1]. A decade later, its functions include, but are not limited to, acting as a transcriptional repressor, binding both RNA and DNA, and modulating gene splicing [2-6]. It is also involved in RNA metabolism during the stress response [7-9]. The marked interest in TDP-43 is related to its role in neurodegenerative diseases [10-16].

Structure and function

TDP-43 is a predominantly nuclear protein, 414 amino acids in length, with an estimated molecular mass of approximately 43 kDa [1]. The TARDBP gene on chromosome 1, which encodes TDP-43, is 6 exons in length and has up to 11 different alternative splice forms [17], the predominant being the 43 kDa form [4, 17]. Both mRNA and protein expression seem to be ubiquitous, as TDP-43 is detected in the pancreas, placenta, spleen, testis, ovary, lung, kidney, spinal cord and brain [3]. This distribution holds true for both rodents and humans although the actual levels of expression may vary amongst these tissues and also between species [3]. Evolutionarily speaking, the TARDBP gene is highly conserved and has been found in all higher species, as well as in Drosophila melanogaster and Caenorhabditis elegans [18], signifying the importance of its function. In addition, TARDBP knockouts are embryonic lethal due to peri-implantation defects [19].

The primary structure of TDP-43 resembles that of a heterogeneous nuclear ribonucleoprotein family member [1]. This type of structure includes two RNA recognition motifs and a glycine -rich C-terminal tail [17]. One of the RNA recognition motifs has been shown to bind to the gene for the cystic fibrosis transmembrane conductance regulator (CFTR), allowing for skipping of exon 9 through alternative RNA splicing, contributing to cystic fibrosis [17]. The glycine rich C-terminal tail contains most of the known mutations, suggesting that neurotoxic effects of TDP-43 are driven by this domain [20-23]. Im-munohistochemical staining of C-terminal fragments are enriched in TDP-43 inclusions [24]. In vitro work has also shown these fragments to be toxic [21, 25].

Numerous functions have been proposed for TDP-43 through studies in cell culture experiments, animal models and biochemical assays [26-29]. Most functions suggest a role of TDP-43 in transcriptional repression, RNA metabolism and gene splicing. These roles involve -TDP -43 binding to both RNA and DNA. These interactions converge around a conserved poly- UG sequence contained in RNA [30]; however, DNA binding domains have not been elucidated, suggesting a more indirect effect. Recent studies have suggested that it is also a component of stress granules induced by cell stress, such as oxidative or osmotic stress [7-9].

Pathology of TDP-43 in FTLD-TDP and ALS

In affected neurons and glia in neurodegenerative disorders, TDP-43 is absent from its normal nuclear location and found in the cytoplasm in the form of inclusion bodies, which are associated with insoluble forms of the protein in biochemical extracts of affected tissue [12]. Pathological aggregates in FTLD-TDP with or without motor neuron disease and in amyotrophic lateral sclerosis (ALS) contain protein with posttranslational modifications, including phos-phorylation, ubiquitination and proteolytic cleavage [12, 24, 31-33]. These forms of TDP-43 have been shown to accumulate in cytosolic and nuclear fractions [34]. Abnormal forms of TDP-43 have been shown with immunoelectron microscopic to accumulate as intracellular filamentous inclusions in neurons and glia [35, 36].

The morphology and anatomical pattern of TDP-43 inclusions shows disease specificity that correlate with clinical and genetic phenotypes [14, 37]. Table 1 summarizes features of FTLD-TDP subtypes as originally defined by Mackenzie and colleagues based on clinical features and distribution of abnormal TDP-43 [37] . More recently, this scheme has been validated and extended to subcortical regions [14]. Table 1 also includes limited studies of TDP-43 pathology in normals. The commonality in all FTLD-TDP subtypes is pathology in frontal and temporal neocortical regions, but not in occipital cortex or cerebellum [38]. The relative amount of inclusions in different cellular structures differs among the FTLD-TDP types. For example, Type 1 is associated with widespread cortical and subcortical neuronal cytoplasmic inclusions (NCI), dystrophic neurites (DN) and neuronal intraneuronal inclusions (NII), while Type 2 has predominantly DN in the cortex and Pick-body like NCI in the dentate fascia, amygdala and basal ganglia, and Type 3 has mainly NCI in medial temporal lobe structures and in cases with motor neuron disease, NCI in motor neurons [14].

Table 1.

TDP-43 in neurodegenerative disorders adapted from [15, 50, 62, 71]

Pathologic classification	Clinical presentation	Genetic links	TDP-43 cellular locations	Frequency in disease
FTLD-TDP type 1	FTDbv or PNFA	Progranulin (RGN) mutations	DN, NCIs and NII in layer II of frontal and temporal cortex.	32% of FTLD-TDP43
FTLD-TDP type 2	Semantic dementia	None	DN predominant in lower layers of cerebral cortex	27% of FTLD-TDP43
FTLD-TDP type 3	Motor neuron disease	Familial cases linked to chromosome 9	NCI in cerebral cortex and hippocampus	42% of FTLD-TDP43
AD	Unknown	Unknown	DN and NCI in amygdala and hippocampus	23%
Normal	N/A	Unknown	DN, NCI and NII in amygdala and hippocampus	3%

Classification according to Mackenzie, et al. [72] Abbreviations: dystrophic neurites (DN), neuronal cytoplasmic inclusions (NCI), intraneuronal inclusions (NII), progressive non-fluent aphasia (PNFA), FTDbv = behavioral variant of frontotemporal dementia; N/A = not appropriate.

TDP-43 and Alzheimer's disease

In it increasingly clear that abnormal TDP-43 immunoreactivity is common in AD [15, 16, 38-41]. Presence of TDP-43 neuronal and glial inclusions is estimated to be approximately 25-30% of sporadic AD cases [15, 16, 39], but perhaps lower (14%) in familial AD and Down's syndrome [41]. The wide range in frequency of TDP -43 pathology in various disorders is in part related to methodology, since higher frequency of TDP-43 pathology is detected in AD when immunohistochemistry is performed with antibodies that are specific to pathologic forms of TDP-43, such as abnormal phospho-epitopes [42] or carboxyl-terminal epitopes [21, 24].

The potential clinical implications of TDP-43 immunoreactivity in AD have been explored by investigating associations between presence of abnormal TDP-43 and both imaging and behavioral features. These studies suggest that presence of TDP-43 is associated with greater brain atrophy (in particular hippocampal atrophy) and more severe clinical deficits [40]. Hippocampal atrophy is also associated with hippocampal sclerosis in the elderly [43, 44], a relatively common coexistent finding in elderly with dementia [45]. Given that TDP-43 is frequent in hippocampal sclerosis [16, 46-48] its role in the worse hippocampal atrophy needs to be considered. After controlling for concomitant pathology, hippocampal atrophy continued to be greater in AD with TDP-43 pathology compared to AD without TDP-43 pathology [40].

In the setting of AD the most common pattern of TDP-43 pathology is that in which it is limited to limbic regions of the brain, including the hippocampus, amygdala and adjacent cortices [15, 49, 50]. These distribution overlaps with tau pathology in AD in the form of neuropil threads and neurofibrillary tangles (NFT) [51]. In fact, some of the TDP-43 pathology in AD has been shown to be within neurons with NFT using double labeling immunofluorescent microscopy and double labeling immunoelectron microscopy [16]. Such a relationship has not been found for amyloid plaques. Several studies have addressed the proportion of TDP-43 pathology that is associated with NFT, with results ranging from more than 14% [50] to almost none [15]. Clearly, there is a wide range of TDP-43 pathology in AD, with some indistinguishable from that seen in FTLD-TDP, including presence of neuronal cytoplasmic and intranuclear inclusions as well as dystrophic neurites to cases in which the brunt of the TDP-43 is in neurons vulnerable to NFT. Given that other tauopathies such as Guam Parkinson dementia complex [10], corticobasal degeneration [15] and argyrophilic grain disease [52] have TYDP-43 pathology, more biochemical evidence is needed to determine if there is protein-protein interaction between tau and TDP-43.

Genetics

Common and rare variants in the gene for TDP-43 (TARDBP) have not been studied extensively in AD and other disorders with TDP-43 pathology. With respect to AD, one recent study investigated 8 different TARDBP single-nucleotide polymorphisms (SNPs) in a Japanese population [53] and found no significant association between 181 AD patients and 130 age-matched controls. In addition, no synergistic effects were observed between APOE genotypes and the TARDBP SNPs for the AD cohort [53]. The most common cause of familial FTLD-TDP is mutation in the gene for progranulin (GRN) [54-56]. Of interest is the fact that common variants in the 3'-untranslated region of GRN in a possible micro-RNA binding site have been shown to associated with risk of FTLD-TDP [57] and hippocampal sclerosis in the elderly [46]. It remains to be determined if it is associated with TDP-43 pathology in AD, although one study suggested a trend for this [46]. How changes in progranulin levels lead to TDP-43 pathology remains an unanswered research question, but given that progranulin is a growth factor and that growth factor withdrawal is associated with programmed cell death [58], it is tempting to speculate that activation of programmed cell death pathways may play a role, as suggested by some in vitro experiments [31]. Further studies are needed to determine if this is a viable mechanism in human neurodegenerative disorders.

TDP-43 in normal brain

The majority of the studies on TDP-43 have focused on its abnormal distribution in disease, including FTLD-TDP, ALS and a range of other disorders [10, 15, 38, 48, 52, 59]. There have been far fewer studies looking at changes in TDP-43 associated with aging. To better understand the frequency of TDP-43 inclusions in normal individuals, TDP-43 immunohistochemistry was used to study 63 neurologically normal individuals ranging in age from 23 to 94 years at death (median 71). Immunohistochemistry was performed as in other studies [16, 60] and the region used for screening was the amygdala, a region known to be predominately affected in AD [61]. Two of the 63 controls (3%) had TDP-43 inclusions - an 81-year-old man and an 83-year-old woman. Both were cognitively normal without any clinical features of motor neuron disease. Additional immunohistochemistry of medial and superior frontal cortex, superior temporal, cingulate cortex, anterior and posterior hippocampus, striatum, midbrain, pons, medulla and spinal cord of the two positive cases showed that abnormal TDP-43 was restricted to limbic regions (Figure 1). These findings are similar to the only other report on the frequency of abnormal TDP-43 in normal controls, which demonstrated 1 case out of 33 controls (3%) and TDP-43 inclusions limited to hippocampus and entorhinal cortex [62]. To study possible biochemical changes in TDP-43 expression in AD and normals, frontal cortical tissue samples from four neurologically normal controls and five AD cases were homogenized and extracts subjected to western blot analysis. No significant differences were observed in relative expression levels between AD and controls (Figure 2). None of the AD or controls had abnormal lower molecular weight (25 kDa and 35 kDa) cleavage fragments comparable to those found in FTLD-TDP and ALS [12].

Figure 1.

TDP-43 immunoreactivity in normals. TDP-43 immunohistochemistry in the amygdala of the positive cases at a 40X magnification. Black arrows indicate pathological TDP-43 inclusions, while blue arrows indicate normal nuclear TDP-43. A. An 83-year-old woman with medial temporal tangles and grains. B. An 81-year-old man with minimal entorhinal tangles.

Figure 2.

TDP-43 expression in age-matched AD and control subjects. A. Five age-matched AD and four control brain samples were homogenized and run on a 10-20% Tricine Gel. The membrane was probed with the monoclonal TDP-43 antibody (EnCor Biotechnology, Gainesville, FL). One clean band migrating at the expected molecular weight was observed for all samples. The membrane was stripped and re-probed with the goat polyclonal GAPDH antibody (Santa Cruz, CA) as a loading control. B. Intensities of the signal were quantified using the NIH Image J software. There was no difference in TDP-43 expression between AD and control brain samples, after normalization to GAPDH.

Conclusion

Studies examining both genetic and pathological influences of TDP-43 in FTLD-TDP and ALS suggest that it is a primary factor in these disorders (i.e. primary TDP-43 proteinopathies), but also that it is a pathological hallmark of a neurodegenerative process that can occur in association with other distinct pathologic processes (i.e. secondary TDP-43 proteinopathies). Intriguing evidence that genetic variants in GRN may be associated with TDP-43 pathology in FTLD-TDP [57] need to be explored in other conditions in which pathologic TDP-43 is found [46]. Other genetic factors that influence progranulin levels, such as genetic variants in the gene (SORT1)[63] for the neuronal receptor for progranulin, sortilin [64], and for genes associated with risk of FTLD-TDP, including TMEM106B [65, 66], need to be explored in AD and other conditions in which TDP-43 pathology is found as part of the growing family of secondary TDP-43 proteinopathies. Additional studies are needed in clarifying biochemical changes in TDP -43 and how they may be exploited as bio-markers for differential diagnosis and early detection of TDP-43 proteinopathies [67-70] and to study the influence of this disease process in other disorders.