|Home | About | Journals | Submit | Contact Us | Français|
Mutations in TARDBP, encoding TAR DNA-binding protein-43 (TDP-43), are associated with TDP-43 proteinopathies, including amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration (FTLD). We compared wild-type TDP-43 and an ALS-associated mutant TDP-43 in vitro and in vivo. The A315T mutant enhances neurotoxicity and the formation of aberrant TDP-43 species, including protease-resistant fragments. The C terminus of TDP-43 shows sequence similarity to prion proteins. Synthetic peptides flanking residue 315 form amyloid fibrils in vitro and cause neuronal death in primary cultures. These data provide evidence for biochemical similarities between TDP-43 and prion proteins, raising the possibility that TDP-43 derivatives may cause spreading of the disease phenotype among neighboring neurons. Our work also suggests that decreasing the abundance of neurotoxic TDP-43 species, enhancing degradation or clearance of such TDP-43 derivatives and blocking the spread of the disease phenotype may have therapeutic potential for TDP-43 proteinopathies.
FTLD is a common form of dementia, second in incidence only to Alzheimer’s disease among individuals with early onset dementia1. Pathologically, individuals with FTLD show atrophy often limited to the prefrontal and anterior temporal lobes, although clinically they may display significant heterogeneity. FTLD can be classified into distinct groups on the basis of immunohistochemistry, as either tauopathy with tau-positive inclusions (FTLD-tau) or ubiquitinopathy with tau-negative but ubiquitin-positive (ub+) neuronal inclusions (FTLD-U; ref. 2 and references within). Recent studies show that most ub+ inclusions contain TDP-43 (refs. 3,4), and most of the remainder contain fused in sarcoma/translocated in liposarcoma (FUS/TLS)5,6. TDP-43 and FUS/TLS are both DNA- and RNA-binding proteins involved in numerous aspects of gene regulation. Collectively, neurodegenerative diseases with TDP-43–immunoreactive pathology have been named TDP-43 proteinopathies7.
Encoded by the TARDBP gene, TDP-43 is a multifunctional DNA- and RNA-binding protein that is involved in many cellular processes, including RNA transcription, alternative splicing and mRNA stability regulation8–10. Since the landmark discovery of TDP-43 as an important component of inclusion bodies in ALS and FTLD, more than 30 TDP-43 mutations have been identified in individuals affected by ALS and FTLD with TDP-43–immunoreactive pathology (FTLD-TDP)9. The C-terminal fragments of human TDP-43 (hTDP-43) are detected in tissue samples from individuals with ALS and FTLD-TDP3,4,11–13, suggesting that the C-terminal domain may have a role in TDP-43 proteinopathy.
Transient expression of the mutant hTDP-43 gene leads to apoptotic death of spinal motor neurons in chicken embryos14. Overexpression of wild-type or mutant hTDP-43 causes motor neuron degeneration in mice and rats15–18. Our previous studies show that simply overexpressing the wild-type human TDP-43 in cultured cells or in transgenic flies is sufficient to cause pathology mimicking that found in individuals with TDP-43 proteinopathy19,20. However, molecular mechanisms by which TDP-43 mutations lead to neurotoxicity remain to be elucidated. Here we examined wild-type and A315T mutant TDP-43 both in cultured cells and in the Drosophila melanogaster model of TDP-43 proteinopathy. Synthetic TDP-43 peptides flanking amino acid residue 315 form amyloid fibrils in vitro and cause neuronal death with axonal damage when added to cultured neurons. Our work identifies an amyloidogenic and neurotoxic region in the C-terminal domain of TDP-43, flanking residue 315. These experiments reveal previously unknown similarities between TDP-43 and prion proteins in their peptide sequences and biochemical properties.
To understand how mutations in the TDP-43 gene cause neurodegeneration, we generated transgenic flies expressing human TDP-43 (hTDP-43) containing the ALS-associated mutation A315T. We chose two wild-type lines and three A315T mutant lines that showed similar expression of hTDP-43 for further characterization (Supplementary Fig. 1). Use of strong promoters such as the actin-Gal4 driver caused death of flies expressing mutant TDP-43 during the embryonic stages. When we used the OK371-Gal4 driver to drive specific expression of the mutant hTDP-43 in subsets of motor neurons, we noticed that flies often failed to eclose and that surviving flies were smaller than flies expressing either the vector control or wild-type hTDP-43 (Fig. 1a–c; compare right panels showing larvae).
Expression of membrane-bound green fluorescent protein (mGFP) in these flies allowed us to examine their motor neurons via their mGFP-marked axons. As compared to the control group, flies expressing either wild-type or A315T mutant TDP-43 showed axonal abnormalities (Fig. 1). Motor neurons expressing wild-type or mutant TDP-43 showed axonal swelling (Fig. 1b,c, white arrows). In the A315T mutant group, flies frequently died before the third instar stage. All of the larvae expressing A315T mutant TDP-43 that survived to the third instar stage showed a marked axonal loss (Fig. 1c). In the remaining axons we detected severe damage, including axon swelling (white arrows), axon thinning and defects in axonal integrity (purple arrows). We examined larval movement in different lines of transgenic flies with consistent results. Both the wild-type and A315T mutant groups showed substantial movement impairment as compared with the control flies expressing the red fluorescent protein (RFP) vector. Flies expressing the A315T mutant hTDP-43 showed more severe movement deficits than did flies expressing wild-type hTDP-43 (Fig. 1d), although expression of hTDP-43 in the wild-type and A315T groups was comparable (Supplementary Fig. 1).
Motor neurons in the control RFP vector–expressing flies showed well-organized clusters in the ventral nerve cord (Fig. 2a), whereas motor neurons in larvae expressing wild-type or mutant hTDP-43 showed disorganized motor neuron clusters and neuronal loss, especially in the posterior abdominal segments (Fig. 2b,c). By the third instar stage, condensed nuclei appeared in motor neurons in the posterior abdominal segments, consistent with our previous study20. Quantification of motor neurons in the last three abdominal segments showed that ~79% of neurons expressing A315T mutant TDP-43 showed signs of cell body swelling or nuclear condensation (marked by the arrowhead and arrow, respectively), as compared to about 32% in flies expressing wild-type TDP-43 (Fig. 2b,c). Flies expressing the A315T mutant had more severe motor neuron loss and axonal damage, accompanied by more severe movement deficits, than did the wild-type hTDP-43 group (Figs. 1 and and2).2). We noticed that the motor neuron damage in these transgenic flies often occurred in a group fashion with clustered cell loss or axonal loss.
We also tested effects of A315T mutant TDP-43 expression in mammalian cells. When expressed in cultured neurons, the A315T mutant formed cytoplasmic protein aggregates. Neurons expressing A315T mutant TDP-43 showed substantially lower survival than those expressing wild-type TDP-43 protein (Supplementary Fig. 2). Thus, expression of the ALS mutant A315T TDP-43 in both cultured mammalian neurons and fly motor neurons in vivo leads to significantly increased neurotoxicity as compared to wild-type TDP-43.
We examined postmortem brain tissues from the brain bank of the Cognitive Neurology & Alzheimer’s Disease Center (CNADC) at Northwestern University. Seven control samples from subjects without clinical cognitive impairment or obvious TDP-43–positive pathology (control group) and seven FTLD-U samples with TDP-43–immunoreactive inclusions (FTLD-TDP) were lysed in radioimmunoprecipitation assay (RIPA) buffer and analyzed by western blotting using a polyclonal antibody specific for TDP-43. As expected, the control brain samples showed moderate levels of TDP-43 (migrating at the expected molecular weight of 43 kDa), which were detectable in four of the seven control brain samples on western blots after long exposures (Fig. 3a). In contrast, FTLD-TDP brain samples showed substantially higher total amounts of different TDP-43 species, although input protein quantities were similar, as shown by western blotting using anti-actin antibody (Fig. 3a,b). In addition to the predicted 43-kDa band and lower-molecular-weight species (marked by and the arrowhead, respectively), we detected an abnormal species migrating around 74 kDa in all seven subjects with FTLD-TDP. We observed this 74-kDa species only at low level, after a long exposure of the western blot, in three of the seven control brain samples (marked by the arrow). This suggests that the 74-kDa TDP-43 derivative may represent a common aberrant TDP-43 product associated with FTLD-TDP.
We next tested whether a similar aberrant TDP-43 protein species was produced in TDP-43–expressing cells. We prepared stable HEK293 cell lines expressing either wild-type or A315T mutant TDP-43 as hemagglutinin (HA) epitope–tagged proteins. These stable cell lines were used for biochemical experiments. We then separated protein lysates from TDP-43–expressing cells into Sarkosyl-soluble and Sarkosyl-insoluble pellet fractions (Supplementary Methods), then analyzed them by SDS-PAGE under denaturing conditions. Western blotting using a specific antibody against the HA tag revealed a band migrating at around 75 kDa, detectable only in cells expressing the A315T mutant TDP-43 but not those expressing wild-type TDP-43 (Fig. 4a, marked by the arrow). The 75-kDa band was much more abundant in the Sarkosyl-insoluble fraction (Fig. 4a; compare lane 2 with lane 4).
Following 3% SDS extraction of Sarkosyl-insoluble pellets, we analyzed Sarkosyl-soluble and Sarkosyl-insoluble fractions by electrophoresis under semidenaturing conditions21 (Fig. 4b). We detected an SDS-resistant TDP-43–immunoreactive oligomer species in cells expressing wild-type TDP-43 and those expressing the A315T mutant, with the SDS-resistant oligomeric species being much more abundant in Sarkosyl-insoluble fractions (compare lane 2 with lane 1 and lane 4 with lane 3; Fig. 4b). There was a substantially higher concentration of SDS-resistant TDP-43–positive oligomer species in the cells expressing the A315T mutant than in those expressing TDP-43 (compare lane 4 with lane 2; Fig. 4b).
We examined the biochemical behavior of TDP-43 protein species by size-exclusion chromatography using RIPA-soluble protein lysates prepared from HEK293 stable cells expressing either wild-type or A315T mutant TDP-43-HA proteins. Following gel filtration, we analyzed the different fractions by western blotting with anti-HA antibody. The 43-kDa TDP-43 protein was present in fractions ranging from >667 kDa to <23 kDa (Fig. 4c–e; Supplementary Figure 3), as marked by molecular size markers in conjunction with the 23-kDa cytosolic protein of thioredoxin peroxidase (TPX) family, a protein that interacts with presenilin22,23. In addition to the 43-kDa monomer species, we detected an aberrant 75-kDa TDP-43 species (marked by the arrow) lysates from cells expressing the A315T mutant TDP-43 in the high-molecular-weight fractions between the 440-kDa and 67-kDa molecular weight markers (fractions 24–29) (Fig. 4d,e). In contrast, we detected the 60-kDa TDP-43 species in lysates from cells expressing either wild-type TDP-43 or the A315T mutant almost exclusively in fractions 31–33 of gel filtration (marked by the arrowhead in Fig. 4c,d). These data suggest that, in solution, TDP-43 exists in multiple oligomeric conformations.
To test whether the aggregation tendency of TDP-43 was intrinsic to TDP-43 or dependent on the presence of other mammalian cellular proteins, we used recombinant TDP-43 protein purified from Escherichia coli. We purified glutathione S-transferase (GST)-tagged TDP-43 (either wild type or A315T) to more than 95% homogeneity and subjected it to gel filtration analyses (Supplementary Fig. 4). GST-tagged proteins were used because the TDP-43 protein became unstable when the GST tag was cleaved. We detected both the wild-type and A315T mutant TDP-43 proteins predominantly as a 70-kDa band (the expected molecular weight of a monomeric GST–TDP-43 fusion protein) in a wide range of fractions between the 667-kDa and 13.7-kDa molecular weight markers. This suggests that the purified TDP-43 protein alone may form high-molecular-weight oligomers and/or adopt different conformations in solution. In addition, we detected a ladder of lower-molecular-weight species ranging from 65 kDa to 29 kDa by western blotting using anti–TDP-43 antibody in A315T TDP-43 fractions, suggesting that these TDP-43 species may be more resistant to degradation.
We further investigated aberrant TDP-43 species in mammalian cells. We prepared RIPA-soluble cell lysates from the wild-type TDP43-HA or A315T TDP-43-HA stable expression cells and separated them by SDS-PAGE. Western blotting analyses of these cell lysates with anti-HA antibody revealed 75-kDa (detected as 75–76 kDa doublets on some gels), 43-kDa and lower-molecular-weight bands (Fig. 5a). The 75-kDa species was detected at a substantially greater abundance in cell lysates expressing A315T mutant TDP-43 (Fig. 5a). This 75-kDa species was phosphorylated because treating cells with the phosphatase inhibitor okadaic acid (OA) increased its concentration (Fig. 5b; compare lane 4 with lane 3). Conversely, alkaline phosphatase treatment substantially reduced the amount of the 75-kDa species, with a concomitant increase in the 43-kDa and lower-molecular-weight bands (Fig. 5b, compare lane 6 with lane 5). Consistent with this, the 75-kDa aberrant TDP-43 species was resistant to heating (up to 100 °C for 15 min), reducing conditions of 200 mM DTT and denaturation with 6 M urea, suggesting that the 75-kDa species is covalently modified (Fig. 5c). MS analysis confirmed that the purified 75-kDa band contained TDP-43 protein without detectable peptides of other proteins.
The above experiments suggest that the 75-kDa species is a hyperphosphorylated form of the TDP-43 protein. To test whether the 75-kDa species was also ubiquitinated, we immunoprecipitated the wild-type and A315T mutant TDP-43-HA proteins using anti-HA monoclonal antibody, separated them by SDS-PAGE and blotted with a specific anti-ubiquitin antibody. Although the high-molecular-weight species (>100 kDa) were recognized by both anti–TDP-43 and anti-Ub, the 75-kDa species was not recognized by anti-Ub, supporting the idea that the 75-kDa species was not ubiquitinated (Supplementary Fig. 5).
We detected multiple lower-molecular-weight bands on western blots, including those migrating between 43 kDa (marked by ) and 15 kDa (for example, those marked by double arrowheads in Fig. 5a), suggesting that some TDP-43 fragments or derivatives may be resistant to proteases present in the cell lysates. To test this, we carried out western blotting following digestion of cell lysates using different concentrations of protease K. Western blotting by anti-HA revealed ladders of TDP-43–reactive bands (Fig. 5d, lanes 1–6). After treatment with protease K at the concentration of 1.2 μg ml−1 for 30 min, we detected, using specific anti–TDP-43 antibodies, protease K–resistant TDP-43 species smaller than 10 kDa that were not reactive to anti-HA antibody (Fig. 5d, lanes 9–10 and lanes 7–8), indicating that the HA tag was cleaved off these TDP-43 fragments. Lysates from cells expressing A315T mutant TDP-43 showed consistently higher abundance of protease K–resistant species of different sizes (Fig. 5d; compare lanes 2, 4 and 6 with lanes 1, 3 and 5). Under similar conditions, even at the 0.3 μg ml−1 protease K concentration, none of the other proteins examined—including TPX, actin or GAPDH—showed any protease K–resistant bands. We quantified different protease K–resistant bands following protease K treatment at a moderate concentration (1 μg ml−1) and found that the A315T mutant consistently produced higher amounts of protease K–resistant fragments, including 24–25 kDa and 10–11 kDa, although we used the same amounts of input protein as detected by either anti-actin or anti–TDP-43 controls (Fig. 5e,f). These data indicate that TDP-43 is partially resistant to protease and that the A315T mutation further increases this protease resistance.
The above biochemical analyses show that the TDP-43 protein has the propensity to form high-molecular-weight aberrant species that are heat stable and detergent resistant. Moreover, the lower-molecular-weight TDP-43 fragments are partially resistant to protease K. These biochemical features are reminiscent of prion proteins. Consistent with this, our sequence analysis indicates that the C-terminal fragments of both human and chimpanzee TDP-43 have moderate sequence similarity to human and chimpanzee prion proteins (Fig. 6a).
Using a molecular dynamics simulation24, we analyzed different regions at the C-terminal domain of TDP-43. A 46-amino-acid fragment flanking residue 315, with either the wild-type alanine or mutant threonine at this position, showed notable features. For both the wild-type and the A315T mutant peptides, the N-terminal half of the 46-mer TDP-43–derived peptide seems to be more flexible, sampling multiple conformations, including a partially collapsed globular conformation. In contrast, the C-terminal half of the peptide probably adopts an extended β-sheet conformation.
The A315T mutant peptide shows a more extended conformation at its C terminus (Supplementary Fig. 6 and Supplementary Movie 1). Consistent with this, analyses of the flexibility index, β-sheets and β-turns of the peptide sequence in this region of TDP-43 using the Protscale server (http://expasy.org/tools/protscale.html)24–26 revealed that the N-terminal half of the TDP-43 peptide seems to be more flexible and more prone to form β-turns than is the C-terminal half, whereas the C-terminal half (especially residues 313–321) has a higher propensity of forming β-sheets (Fig. 6b–d). This propensity to form β-sheets is further increased by the A315T mutation (Fig. 6c).
We also analyzed the TDP-43 peptides using a Ramachandran plot27. Again, the A315T mutation is predicted to increase the probability to form β-sheets (Fig. 6e) and has a higher probability of staying in an extended conformation at 14–16 Å of a radius of gyration (Fig. 6f). Notably, when we analyzed the TDP-43 peptide containing the A315E phosphomimetic mutant, the A315E mutant showed an even higher propensity to form β-sheets around the mutation site (Fig. 6e), supporting the idea that phosphorylation of this residue may affect the conformation of the TDP-43 protein.
The prediction of β-sheet formation in these TDP-43 peptides led us to test whether such peptides could form amyloid fibrils. We prepared two TDP-43 peptides flanking amino acid residue 315, one with the wild-type sequence (wild-type TDP46mer: Gln286–Gln331) and the other with an identical TDP-43 peptide except that it contained phosphothreonine at residue 315 (A315T TDP46mer), together with a control peptide containing the reverse sequence of the amyloid peptide Aβ42. We tested these peptides for their interaction with thioflavin T (ThT). Binding of ThT to amyloid fibrils leads to increased fluorescence emission at 480 nm upon excitation at 440 nm28. Both the wild-type and A315T mutant synthetic 46-mer peptides showed significant fluorescence upon binding to ThT, whereas the control peptide had no detectable ThT fluorescence (Fig. 7a). Changes in ThT fluorescence intensity with time showed sigmoidal curves similar to those of other amyloid fibrils28,29 (Supplementary Methods).
We carried out EM to examine the fibrils formed with the TDP-43 peptides. Consistent with their binding to ThT, both the wild-type and A315T mutant TDP-43 synthetic peptides form fibril structures (Fig. 7b).
To examine the dynamic process of the fibril formation, we used time-lapse atomic force microscopy (AFM) (Fig. 7c–l). AFM images of the wild-type (Fig. 7c–f) or the A315T mutant peptide (Fig. 7g–j) aggregates formed on mica surface were taken following incubation times of 0, 7 h, 13 h and 17 h. We did not detect aggregates at the zero time point for either the wild-type or A315T mutant peptides (Fig. 7c,g). Aggregation of the wild-type TDP-43 peptide followed the process from initially granular oligomers, to mixtures of oligomers and short protofibrils, and finally thin fibrils (Fig. 7c–f). The A315T mutant peptide, however, showed a faster initial phase of protofibril formation, with numerous short thin protofibrils detectable by 7 h, thicker longer fibrils by 13 h, and finally long, thick fibers by 17 h of incubation (Fig. 7g–j). The corresponding cross-sectional profiles confirmed that the average height of the A315T mutant fibrils (7.7 nm) was larger than that of the wild-type fibrils (0.29 nm) (Fig. 7k,l).
We next examined whether these TDP-43–derived peptides affected neuronal survival when added to primary cultures. In addition to the TDP-43 synthetic peptides and the control peptide described above, we included the Aβ42 peptide as a positive control. After treating cultured primary mouse cortical neurons with corresponding peptides at different concentrations, we monitored neuronal death by immunostaining with the neuronal marker Tuj-1 antibody and nuclear staining with Hoechst dye 33342, followed by fluorescence microscopy. When neurons were treated with the control peptide, we detected only a baseline level of cell death. Consistent with previous studies30,31, treatment using Aβ42 peptide led to a marked increase in neuronal death (Aβ42 panels, Fig. 8). Treatment with the 46-mer wild-type peptide increased neuronal death, with cells showing positive terminal deoxynucleotidyltransferase dUTP nick end-labeling (TUNEL) staining and the appearance of condensed or fragmented nuclei (as indicated by arrowheads in ‘TUNEL’ and ‘Nu’ panels, respectively). This was accompanied by increased axonal varicosity and reduced axonal integrity, with only a fraction of axons showing relatively normal morphology (marked by arrows in the Tuj-staining panels, Fig. 8a). Conversely, in A315T peptide–treated neurons, such neurotoxicity was further increased, with more than 30% of neurons showing TUNEL signals (Fig. 8b). Most axons in the A315T mutant peptide–treated group showed abnormal morphology, including increased varicosity and reduced axonal integrity (marked by arrowheads in the Tuj-staining panels, Fig. 8a). Neurotoxicity caused by these TDP-43 peptides was dose dependent (Fig. 8c). These data suggest that TDP-43 derivatives present in the extracellular environment cause significant neuronal damage.
Our results reveal molecular and biochemical features that are shared between TDP-43 and other amyloid proteins, including prions. Notably, the 46-mer TDP-43 peptide shows sequence similarity to the neurotoxic prion peptide PrP106–126 (Fig. 6)32. Our study thus identifies an amyloidogenic and neurotoxic region at the C-terminal domain of TDP-43. Our work also uncovers previously unrecognized similarities between TDP-43 and abnormal prion proteins, including partially protease-resistant protein fragments and amyloid fibril formation.
Cushman et al. recently proposed that the glycine-rich C-terminal domain (Gln277–Met414) of TDP-43 acts as a putative ‘prion domain’ with sequence similarity to yeast prion proteins33. This domain and additional sequences in the RNA-recognition motif were shown as the minimal region necessary to induce TDP-43 aggregation and toxicity when expressed in yeast or neuroblastoma cells34–36. Purified wild-type and ALS mutant TDP-43 proteins formed thread-like or filament-like non-amyloid structures37. Our study, in contrast, shows that synthetic peptide containing Q286-Q331 was sufficient not only to form amyloid fibrils but also to induce neurotoxicity. Additionally, A315T mutant phosphopeptide further increased the neurotoxicity (Fig. 8). This suggests that TDP-43 proteolytic peptides may cause neurotoxicity. Molecular modeling suggests that the C-terminal part of the TDP-43 peptide (Gln286–Gln331) adopts an extended β-sheet conformation. Mutating residue 315 from alanine to threonine further increases the β-sheet propensity and accelerates the formation of protofibrils (Figs. 6 and and7).7). This may help to explain why the A315T mutant peptide causes enhanced neurotoxicity.
It is interesting that, in human tissues, TDP-43 inclusions are not thioflavin S positive, whereas neurofibrillary tangles and amyloid plaques are. This suggests that the conformation of TDP-43 inclusions is different from that of tangles and plaques. It is conceivable that β-sheets formed by TDP-43 peptide(s) may not be as extensive as in tangles and plaques. Alternatively, such TDP-43 β-sheets may be buried inside TDP-43 inclusions and therefore not accessible be for thioflavin S binding.
A previous report suggested that TDP-43 in yeast did not form SDS-resistant species34. Our analyses reveal a heat-stable 75-kDa TDP-43 species that is resistant to detergents, reducing agents and urea treatment, suggesting that the 75-kDa species is covalently modified, possibly by phosphorylation (Fig. 5). Consistent with this, we did not detect similar high-molecular-weight species with the A315T mutant TDP-43 protein purified from E. coli (Supplementary Fig. 4). The aberrant 75-kDa band in lysate from cells expressing the A315T mutant was not recognized by anti-ubiquitin antibodies (Supplementary Fig. 5). The 75-kDa species in cells expressing the HA-tagged A315T mutant TDP-43 was protease sensitive, although multiple forms of lower-molecular-weight TDP-43 species, especially those 10 kDa, are protease resistant (Fig. 5d,e). It should be noted that TDP-43–immunoreactive species larger than 43 kDa have been detected in some samples from affected individuals4,38,39. In our experience, the 74-kDa TDP-43 species detected in such samples is unstable, possibly degraded by endogenous proteolytic enzymes. This may explain why this 74-kDa TDP-43 species in brain samples from affected individuals escaped notice in previous studies. The exact molecular nature of the 74-kDa TDP-43 species in brain samples and the 75-kDa aberrant species in cells expressing the HA-tagged A315T mutant TDP-43 remains to be elucidated.
In TDP-43 proteinopathy tissues, TDP-43 is often found cleared from the nuclei of neurons containing cytoplasmic inclusion bodies, suggesting that pathogenesis may be driven by a loss of normal TDP-43 function in the nucleus. We examined wild-type and A315T mutant TDP-43 in a splicing assay using a previously reported cystic fibrosis transmembrane conductance regulator (CFTR) exon 9 minigene40. When co-transfected with the CFTR mini-gene, the C-terminal fragment T202 led to a substantial change in its ability to suppress exon 9 inclusion. In contrast, both wild-type and A315T mutant TDP-43 proteins showed similar activity in suppressing exon 9 splicing (Supplementary Fig. 7). This suggests that the neurotoxic effect of the A315T mutant TDP-43 may not be caused by the loss of its splicing regulatory activity.
Several lines of evidence support the gain-of-function toxicity model. Most TDP-43 mutations identified are missense mutations. Deleting or knocking down expression of the TDP-43 homolog in flies20,41,42, or deleting the TARDBP gene in mice43, did not lead to the formation of inclusion bodies—characteristic neuropathological features of TDP-43 proteinopathy—although movement deficits and abnormal dendritic development were detected in TDP-43 deficient animals41–43. Simply overexpressing the wild-type TDP-43 is sufficient to induce protein aggregation in yeast or mammalian cells33,34,44,45 and leads to neurodegeneration with TDP-43 pathology in transgenic animals16,17,20,46. It is possible that both ‘loss of normal TDP-43 function’ and ‘gain-of-function toxicity’ mechanisms contribute to molecular pathogenesis in TDP-43 proteinopathy.
Procedures for plasmids, peptides and antibodies are described in the Supplementary Methods.
Human brain samples were collected from autopsied tissues at the Neuropathology Core of the Northwestern University following US National Institutes of Health (NIH) and institutional guidelines. Brain samples were evaluated for atrophy, for pathology by hematoxylin-eosin staining and immunostaining using multiple antibodies. For details, see Supplementary Methods.
To determine the time course of ThT binding of fibrils formed with synthetic TDP-43 peptide, the wild-type or A315T mutant TDP-43 peptide or the control peptide was incubated at 250 μM in PBS (pH 7.0) with ThT. Dynamic changes in ThT binding at 37 °C with time were detected as fluorescence intensity in arbitrary units (AU). EM images of fibrils formed with the wild-type or A315T TDP-43 synthetic peptides were taken after incubation at 1.25 mM in PBS (pH 7.0) at 37 °C for 10 d with rotation at 240 r.p.m.
Time-lapse AFM images of the wild-type and A315T mutant TDP-43 synthetic peptide were taken during the formation of protofibrils and fibrils after incubation of peptides in aqueous solution at a concentration of 20 mM for 0, 7 h, 13 h and 17 h. All the images were obtained on a mica surface. The corresponding cross-sectional profiles for the wild-type and A315T mutant TDP-43 peptides were measured to determine the thickness of the fibrils with z scale bars included for the AFM images.
Details are provided in the Supplementary Methods.
We thank F. Baralle, E. Buratti, B. Cui, D. Kuo, N. Jayaram, Y. Li, M. Mishra, S. Perrett, M.-Y. Shen, M.-J. Zhang and T. Siddique for providing invaluable suggestions and reagents and for critical reading of the manuscript. We thank members of the Wu laboratory for stimulating discussions and suggestions. We thank L. Guo and L. Wang for technical assistance and A. Joselin for help in the early stage of the work. W.G. (grant 2009CB825402) and Y.C., H.Y. and Q.X. (grant 2010CB529603) are supported by the Ministry of Science and Technology (MOST) China 973 Project. J.Y.W. is supported by funds from Northwestern University and the Chinese Academy of Science (CAS). Y.Y. and C.W. are supported by CAS and MOST. We also thank the US National Institutes of Health (grant AG13854 to M.M. and E.H.B.) for support.
Supplementary information is available on the Nature Structural & Molecular Biology website.
AUTHOR CONTRIBUTIONSJ.Y.W., W.G., X.Z., K.F. and E.J.R. designed the study; W.G., Y.C., X.Z., A.K., P.R., X.C., E.J.R., M.Y., L.Z., J.L., M.X., Y.Y., C.W., D.Z., K.F., E.J.R. and J.Y.W. performed the experiments and analyzed the data; H.Y., L.Z., J.L., Y.S., K.F., Q.X. and J.Y.W. supervised the experiments and discussed and analyzed the data; E.H.B. and M.M. provided crucial tissue samples and revised the manuscript; W.G., E.J.R. and J.Y.W. wrote the paper.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.
Reprints and permissions information is available online at http://www.nature.com/reprints/index.html.