|Home | About | Journals | Submit | Contact Us | Français|
TDP-43 is a highly conserved and ubiquitously expressed member of the heterogeneous nuclear ribonucleoprotein (hnRNP) family of proteins. Recently, TDP-43 was shown to be a major disease protein in the ubiquitinated inclusions characteristic of most cases of amyotrophic lateral sclerosis (ALS), tau-negative frontotemporal lobar degeneration (FTLD), and inclusion body myopathy. In these diseases, TDP-43 is redistributed from its predominantly nuclear location to ubiquitin-positive, cytoplasmic foci. The extent to which TDP-43 drives pathophysiology is unknown, but the identification of mutations in TDP-43 in familial forms of ALS and FTLD-U suggests an important role for this protein in pathogenesis. Little is known about TDP-43 function and only a few TDP-43 interacting proteins have been previously identified, which makes further insight into both the normal and pathological functions of TDP-43 difficult. Here we show, via a global proteomic approach, that TDP-43 has extensive interaction with proteins that regulate RNA metabolism. Some interactions with TDP-43 were found to be dependent on RNA-binding, whereas other interactions are RNA-independent. Disease-causing mutations in TDP-43 (A315T and M337V) do not alter its interaction profile. TDP-43 interacting proteins largely cluster into two distinct interaction networks, a nuclear/splicing cluster and a cytoplasmic/translation cluster, strongly suggesting that TDP-43 has multiple roles in RNA metabolism and functions in both the nucleus and the cytoplasm. Finally, we found numerous TDP-43 interactors that are known components of stress granules and, indeed, we find that TDP-43 is also recruited to stress granules.
The RNA binding protein TDP-431 was recently identified as the major disease protein in the ubiquitinated inclusions characteristic of sporadic and familial forms of amyotrophic lateral sclerosis (ALS), tau-negative frontotemporal lobar degeneration (FTLD), and inclusion body myopathy. TDP-43 pathology also frequently accompanies the pathognomonic pathology of Parkinson’s and Alzheimer’s diseases2-4. In these diseases, TDP-43 is redistributed from its predominantly nuclear location to ubiquitin-positive, cytoplasmic foci. The extent to which TDP-43 drives pathophysiology is unknown, but the identification of mutations in TDP-43 underlying rare familial forms of ALS and FTLD suggests an important role for this protein in pathogenesis5-9.
TDP-43 is a highly conserved and ubiquitously expressed member of the heterogeneous nuclear ribonucleoprotein (hnRNP) family of proteins10. TDP-43 contains two RNA recognition motifs (RRMs) and binds RNA primarily through the first of these1. The glycine-rich C-terminus of TDP-43 has been shown to mediate interaction with several other hnRNP proteins, specifically hnRNPs A1, A2/B1, C1/C2, and A311, although the full extent of TDP-43 interactions has not been previously described. Predominantly a nuclear protein, TDP-43 has been shown to shuttle between the nucleus and cytoplasm12. Interestingly, TDP-43 redistributes to cytosolic granules as a physiological response to neuronal injury, and nuclear localization is restored after recovery13, 14.
Little is known about TDP-43 function, although there is evidence from experimental systems that TDP-43 can negatively regulate expression of target genes at multiple levels, including transcription, splicing and translation15-17, although the full extent of TDP-43 target genes and the influence of TDP-43 on their expression is not known. Additionally, there is no clear consensus of how pathological TDP-43 functions within diseased cells.
To date, only a few TDP-43 interacting proteins have been identified, which makes further insight into both the normal and pathological functions of TDP-43 difficult. Here we show, via a global proteomic approach, that TDP-43 has extensive interaction with proteins that regulate mRNA metabolism. TDP-43 interacting proteins largely cluster into two distinct protein interaction networks. The first is a network of nuclear proteins that regulate RNA splicing and other aspects of nuclear RNA metabolism, and the second is a network of cytoplasmic proteins that regulate mRNA translation. Additionally, we show that TDP-43 interaction with some proteins is dependent on TDP-43 interaction with RNA, whereas other interactions are RNA-independent. Surprisingly, the disease-causing mutations A315T and M337V do not alter the profile of TDP-43 interactions. Numerous proteins in translational regulation cluster are known to accumulate in stress granules and, indeed, we find that TDP-43 is also recruited to stress granules.
FLAG-TDP-43 was subcloned into the mammalian expression vector pcDNA 3.1(+) (Invitrogen). FLAG-TDP-43(A315T), FLAG-TDP-43(M337V) and FLAG-TDP-43(mutRRM) with the W113A and R151A mutations were generated using PCR to perform site-directed mutagenesis.
10 cm2 plates of HEK-293T or HeLa cells grown in a 1:1 mixture of DMEM/F12 culture media were transfected with 5μg of FLAG-TDP-43 or relevant TDP-43 mutant plasmid for 48 hours. Cells were then lysed in gentle lysis buffer (1X PBS, 5mM EDTA, 0.2% NP-40, 10% glycerol + Roche complete EDTA-free protease inhibitor cocktail Cat# 11836170001), passed five times through a 21-gauge needle, and spun at 20,000g for 10 minutes. The supernatant was pre-cleared using Protein G affinity gel (Sigma, Cat# E3403) for 30 minutes and then immunoprecipitated using Anti-FLAG M2 affinity gel (Sigma, Cat# F2426) for 1.5 hours at 4C. The immunoprecipitate was then eluted using FLAG peptide (Sigma, Cat# F3290) at 4C for 30 minutes. 330 μg of RNase A (Sigma, Cat# R4642) was added immediately following lysis prior to immunoprecipitation where indicated. For immunoprecipitation from mouse brain tissue, mouse brain homogenate was lysed as described above and then immunoprecipitated with 2.5 μg of TDP-43 polyclonal antibody (Proteintech, Cat# 10782-2-AP). As a control, half of the homogenate was immunoprecipitated using normal rabbit IgG.
Lysates/immunoprecipitates were separated on a 8-16% gradient tris-glycine gel. M2 monoclonal antibody (Sigma, Cat# F1804) and TDP-43 polyclonal antibody (Proteintech, Cat# 10782-2-AP) were used to visualize TDP-43. Polyclonal antibodies were also used to visualize PABPC1, hnRNP H and hnRNP U respectively (Abcam Cat# ab21060 and ab10374, Bethyl Laboratories Cat# A300-689A).
HEK-293T cells grown on chamber slides (Lab-Tek Cat#154917) were transiently transfected with FLAG-TDP43 or FLAG-TDP-43(mutRRM) using FuGENE 6 (Roche Diagnostics). After 48 h, HEK-293T cells were fixed in 4% formaldehyde in PBS for 10 min at room temperature. The cells were then permeabilized with 0.5% Triton-X in PBS and incubated with primary antibodies for 1 hr to visualize TDP-43, hnRNP H, PABPC1, EIF4G and G3BP1. Cells were then washed and proteins were visualized using secondary antibodies conjugated to Rhodamine Red-X and FITC (Jackson Immunoresearch). Cells were then washed, stained with DAPI and visualized on a Leica DMIRE2 fluorescent microscope using a 63X objective.
The following primary antibodies were used to visualize proteins: mouse anti-FLAG M2 (1:1000 for western blot and immunofluorescence) (Sigma Cat# F1804), rabbit anti-TDP-43 (1:350 for immunofluorescence) (Proteintech Group Cat# 10782-2-AP), rabbit anti-PABPC1 (1:1000 for western blot, 1:200 for immunofluorescence) (Abcam Cat# ab21060-100), rabbit anti-hnRNP H (1:10,000 western blot, 1:500 for immunofluorescence) (Abcam, Cat# 10374-50), mouse anti-G3BP1 (1:200 for immunofluorescence) (BD Transduction Laboratories Cat #611126), and rabbit anti-EIF4G (1:200 for immunofluorescence) (Santa Cruz Biotechnology Cat# sc-11373)
FLAG epitope-tagged TDP-43 constructs were transfected into HEK293T cells and immunoprecipitated as described above. The sample was then run on an 8-16% gel, and analyzed as described below.
The gel lane containing the immunoprecipitated sample was manually excised into 24 bands in the molecular weight range between 14 kDa and greater than 200 kDa. Each of the protein bands was then digested individually as below. The protein bands were cut into small plugs, washed with 50% acetonitrile, and destained by several incubations in 100 mM ammonium bicarbonate pH 8 containing 50% acetonitrile. Reduction (10 mM, DTT for 1 hour at 37°C) and alkylation (50 mM iodoacetamide for 45 min at room temperature in the dark) were performed, followed by washing of the gel plugs with 50% acetonitrile in 50mM ammonium bicarbonate twice. The gel plugs were then dried using a speedvac (Savant) and rehydrated in 10 μl of 0.2ug trypsin. 25uL of 25 mM ammonium bicarbonate pH 8 was added to the tube after 10 minutes. The peptides were extracted from the gel plugs using 20 to 30uL of 0.2% formic acid after an overnight (approx 12 hours) enzymatic reaction at 37°C. The solution was then transferred to a sample vial for LC-MS/MS analysis. Non-transfected cells were used as a control and treated in an identical manner to determine non-specific interactions.
LC-MS/MS analysis was performed using a ThermoFisher LTQ XL linear ion trap mass spectrometer in line with a nanoAcquity ultra performance LC system (Waters Corporation, Milford, MA). Tryptic peptides generated above were loaded onto a “precolumn” (Symmetry C18, 180μm i.d X 20mm, 5μm particle) (Waters Corporation, Milford, MA) which was connected through a zero dead volume union to the analytical column (BEH C18, 75μm i.d X 100mm, 1.7μm particle) (Waters Corporation, Milford, MA). The peptides were then eluted over a gradient (0-70% B in 60 minutes, 70-100% B in 10 minutes, where B = 70% Acetonitrile, 0.2% formic acid) at a flow rate of 250nL/min and introduced online into the linear ion trap mass spectrometer (ThermoFisher Corporation, San Jose, CA) using electrospray ionization (ESI). Data dependent scanning was incorporated to select the 10 most abundant ions (one microscan per spectra; precursor isolation width 3.0Da, 35% collision energy, 30ms ion activation, exclusion duration: 30s; repeat duration: 15s; repeat count: 2) from a full-scan mass spectrum for fragmentation by collision activated dissociation (CAD).
Product ions generated above (b/y-type ions) were used in an automated database search against the Swissprot (Swissprot 57.1, Homo Sapiens subset) database by the Mascot search algorithm18 using trypsin (1 missed cleavages) as the digestion enzyme. The following residue modifications were allowed in the search: carbamidomethylation on cysteine and oxidation on methionine. Mascot was searched with a precursor ion tolerance of 1.0 Da and a fragment ion tolerance of 0.6 Da. Using the automatic decoy database searching tool in the Mascot, a false discovery rate for peptide matches above the identity threshold was estimated to be 4%. In addition, searches were also performed on two mgf files (one for IP lane and one for the control lane) that were generated by merging data from all the bands in each lane. The identifications from the automated search were further validated through Scaffold (Proteome Software, Portland, OR) and manual inspection of the raw data. Peptide identifications were accepted if they could be established at greater than 95% probability as specified by the Peptide Prophet algorithm19. Protein identifications were accepted if they could be established at greater than 99% probability and contained at least 2 identified peptides. Protein probabilities were assigned by the Protein Prophet algorithm20.
TDP-43 interacting proteins in human epithelial kidney (HEK-293T) cells were isolated by immunoprecipitation of FLAG-TDP-43 followed by identification of co-purified proteins by mass spectrometry (Figure 1A, Sup. Figure 1). We found 261 proteins to be enriched in the FLAG-TDP-43 immunoprecipitate relative to control (Table 1). Of these 261 proteins, 126 were found exclusively in association with TDP-43. Sixty-eight proteins were found to be enriched in the control relative to the immunoprecipitate indicating that our immunoprecipitation was highly specific ( Supplementary Table 1).
Analysis of the TDP-43 interactors reveals extensive interaction with proteins that associate with RNA, consistent with previously described roles for TDP-43 in RNA metabolism. These include hnRNPs, RNA helicases, splicing factors, translation regulatory proteins, as well as proteins involved in mRNA transport and stability (Figure 1B and Table I). TDP-43 was found to interact with a smaller number of DNA binding proteins such as transcription factors, consistent with a previously described role for TDP-43 in transcriptional repression1, but also interacts with DNA repair proteins such as Ku70 suggesting that TDP-43 may have roles in DNA metabolism beyond transcriptional regulation (Figure 1B and Table I). Notably, although TDP-43 is predominantly a nuclear protein, we found interaction with both cytoplasmic and nuclear proteins, as well as many proteins that are known to shuttle between the nucleus and cytoplasm. This likely reflects a functional role for TDP-43 in both the nucleus and the cytoplasm consistent with the observation that TDP-43 itself undergoes nucleocytoplasmic shuttling12.
To gain a better understanding of the relationships between TDP-43 interacting proteins, we employed the STRING interaction database21. To minimize the chance of including false positives, our analysis included only those proteins in which the spectral count was at least two-fold enriched in the TDP-43 immunoprecipitate relative to control. Furthermore, only high confidence interactions as determined by the STRING database were accepted. This analysis reveals that TDP-43 interactors cluster largely into two distinct protein interaction networks (Figure 2). The “Nuclear/Splicing Cluster” is comprised entirely of nuclear proteins including many hnRNPs, but also serine/arginine-rich (SR) proteins, small nuclear ribonucleoproteins (snRNPs), an ATP-dependent RNA helicase, and nuclear RNA export factors. These proteins are all involved in nuclear RNA metabolism, primarily RNA splicing but also export of mRNA to the cytoplasm (Table 2). The “Cytoplasmic/Translation Cluster” is comprised entirely of cytoplasmic proteins, including translation initiation and elongation factors, and ribosomal subunits (Table 3). Interestingly, PABPC1 was found to link these two distinct protein interaction networks (Figure 2).
The missense mutations A315T and M337V are causative of dominantly inherited ALS5, 7. To investigate whether disease-associated mutations alter the complement of proteins that interact with TDP-43, we introduced each of these mutations into TDP-43 by site-directed mutagenesis and examined their interaction profiles. We found that TDP-43 variants harboring either the A315T or M337V mutation have interaction profiles that are qualitatively indistinguishable from that of wild type TDP-43 by examination of Sypro-Ruby-stained gel (Figure 3A). This finding suggests that the mechanism by which TDP-43 missense mutants are pathogenic may be due to cell type-specific interactions that do not occur in 293T cells or that disease-causing mutations do not grossly alter the function of TDP-43 or its binding partners.
Since TDP-43 and many of its interacting proteins are RNA binding proteins, we sought to determine how RNA binding influences the TDP-43 interactome. RNA binding by TDP-43 is mediated by its first RRM domain22. Two specific point mutations, W113A and R151A, have been previously shown to abolish RNA binding by TDP-4322. We introduced both of these mutations into FLAG-TDP-43 to generate the RNA binding mutant FLAG-TDP-43(mutRRM). In comparison with FLAG-TDP-43, some TDP-43 interactions are lost with FLAG-TDP-43(mutRRM) indicating that many TDP-43 interacting proteins/complexes are strongly influenced by RNA binding (Figure 3B, lane 4). To further examine the role of RNA binding in determining TDP-43 interactions, we performed immunoprecipitation of TDP-43 in the presence of RNase A to degrade RNA. This approach yielded an almost identical interaction profile to FLAG-TDP-43(mutRRM), further demonstrating the strong influence of RNA binding on TDP-43 interactions (Figure 3B, lane 3). Many of the RNA-dependent interactions are proteins with molecular weights between ~14 and 35 kDa, a cohort largely comprised of ribosomal subunits, which suggests that the association of TDP-43 with ribosomes is indirect and mediated by interaction with the same transcript. However, other proteins are likely to interact with TDP-43 independent of its ability to bind RNA. Such proteins are more likely to be present in a multimeric protein complex with TDP-43.
We verified a subset of TDP-43 interacting proteins by co-immunoprecipitation followed by Western blot. hnRNP H is one of a large number of hnRNPs found to interact with TDP-43 in our proteomic analysis. Similar to TDP-43, this protein has been shown to be involved in the regulation of splicing23. Immunoprecipitation followed by Western blot confirms an interaction between TDP-43 and hnRNP H (Figure 4A). This interaction is not altered in the disease-associated point mutations A315T or M337V (Figure 4A). The interaction between TDP-43 and hnRNP H is at least partially influenced by TDP-43 binding to RNA because treatment with RNase A strongly mitigates interaction (Figure 4A). Consistent with this finding, hnRNP H shows reduced interaction with the TDP-43(mutRRM) mutant (Figure 4A). To determine the subcellular compartment in which the interaction between TDP-43 and hnRNP H occurs, immunofluorescence was performed in HeLa cells to simultaneously visualize TDP-43 and hnRNP H. TDP-43 and hnRNP H both show pan-nuclear localization and are found to co-localize in nuclear puncta (Figure 4B).
Verification of the interaction between TDP-43 and PABPC1 was also performed. PABPC1 is a predominantly cytoplasmic protein that associates with and stabilizes poly(A) mRNA and is regulates RNA translation24, 25. Immunoprecipitation followed by Western blot confirms that PABPC1 associates with TDP-43 and that this association is not affected by either the A315T or M337V mutation (Figure 4A). Immunoprecipitation in the presence of RNase A reveals that the association between TDP-43 and PABPC1 is also dependent upon RNA since binding is strongly mitigated by treatment with RNAse A (Figure 4A). Consistent with this finding, PABPC1 shows reduced interaction with the TDP-43(mutRRM) mutant (Figure 4A). Thus, hnRNP H and PABPC1 interaction with TDP-43 is completely abolished by RNase A treatment, but only partially mitigated by selectively impairing the ability of TDP-43 to bind RNA (TDP-43-(mutRRM)). RNase A treatment is likely to completely disassemble ribonucleoprotein complexes, thus abolishing both direct and indirect interactions between TDP-43 and RNA binding proteins. On the other hand, the residual binding exhibited by TDP-43(mutRRM) indicates limited ability to associate with multimeric ribonucleoprotein complexes independent of its ability to bind RNA, although the interaction is clearly stabilized by RNA binding. Co-immunoprecipitation experiments were also performed in HeLa cells, confirming the interaction between PABPC1 and hnRNP H with TDP-43 and associated mutants (Sup. Figure 2A) and providing a second cell type in which these novel interactions are observed. Furthermore, we performed co-immunoprecipitation from mouse brain homogenate to confirm that interactions between TDP-43 and PABPC1 and hnRNP U occur with the endogenous TDP-43 protein in one tissue that is frequently affected in TDP-43-related disease (Sup Figure 2B).
Although TDP-43 is predominantly a nuclear protein, in some HeLa cells TDP-43 can be visualized in discrete cytoplasmic puncta in addition to diffuse nuclear staining (Figure 5A). These puncta do not stain for hnRNP H (data not shown) although they stain strongly for PABPC1, a marker for cytoplasmic RNA granules26 (Figure 5A). Our findings are consistent with previous evidence indicating that TDP-43 co-purifies with cytoplasmic RNA granules27. Cytoplasmic RNA granules, including stress granules, processing bodies and germ cell (or polar) granules are cytoplasmic structures believed to represent physiological accumulations of mRNA and ribonucleoproteins that modulate gene expression by influencing translation, trafficking and stability28. PABPC1 is a specific marker of stress granules26, suggesting that TDP-43 is present in this specific subtype of RNA granule. Further extensive evidence that TDP-43 associates with stress granules was the identification of TDP-43 interaction with numerous additional protein components of stress granules28, 29 (Table 4).
To confirm the association of TDP-43 with stress granules, we performed immunofluorescence to examine two additional stress granule proteins (EIF4G and G3BP1) that are also known to associate with stress granules. FLAG-TDP-43 was found to strongly co-localize with these proteins in discrete cytoplasmic puncta clearly indicating that cytoplasmic TDP-43 associates with stress granules (Figure 5B-C). Furthermore, endogenous TDP-43 was found to localize to stress granule markers following challenge with the proteasome inhibitor MG-132, a well-established stimulus of stress granule formation (Figure 5D).
To determine whether RNA binding is necessary for TDP-43 localization to stress granules, we visualized the localization of FLAG-TDP-43(mutRRM) and EIF4G as a marker for RNA granules. The localization of FLAG-TDP-43(mutRRM) remains predominantly nuclear, although the presence of tiny discreet puncta is observed in many cells that have both nuclear and cytoplasmic localization that do not co-localize with stress granules (Figure 6). FLAG-TDP-43(mutRRM) was found to be present in only 6.5% of stress granules whereas FLAG-TDP-43 was found to be present in 84.7% of stress granules (Figure 6). This indicates that the association of TDP-43 with stress granules is strongly impaired by an inability to interact with RNA.
Using a global proteomic approach we have demonstrated that TDP-43 has extensive interaction with proteins that regulate mRNA metabolism. These include nuclear proteins, cytoplasmic proteins, and proteins known to undergo nucleocytoplasmic shuttling. Among TDP-43’s interactors are hnRNPs, RNA helicases, splicing factors, translation regulatory proteins, as well as proteins involved in mRNA transport and stability. TDP-43 was found to interact with a smaller number of DNA binding proteins such as transcription factors, consistent with a previously described role for TDP-43 in transcriptional repression1, but also interactions with DNA repair proteins such as Ku70 suggesting that TDP-43 may be involved in other aspects of DNA metabolism.
Disease-associated mutations in TDP-43 are nearly all located within a C-terminal glycine-rich domain that has previously been found to interact with some hnRNPs5-9,11. Surprisingly, the disease-causing mutations A315T and M337V do not alter the profile of TDP-43 interactions in 293T cells. Analysis using the STRING database of protein-protein interactions demonstrates that TDP-43 associates with two distinct protein interaction networks. The first is a network of nuclear proteins that regulate RNA splicing and other aspects of nuclear RNA metabolism, consistent with prior evidence that TDP-43 can influence transcription and RNA splicing10. The second is a network of cytoplasmic proteins that regulate mRNA translation. Although a predominantly nuclear protein, it has been previously shown that TDP-43 shuttles between the nucleus and cytoplasm12. Moreover, TDP-43 has been found to redistribute to cytoplasmic RNA granules in response to neuronal injury13, 14. This is consistent with our finding that TDP-43 has extensive interaction with components of stress granules and that TDP-43 colocalizes with stress granules.
TDP-43 is a relatively new player in a growing list of RNA binding proteins that are associated with disease30. In addition to TDP-43, there are at least two other RNA binding proteins in which mutations lead to motor neuron disease. Loss of function mutations affecting the SMN gene cause spinal muscular atrophy31 whereas mutation in the SR protein FUS/TLS also leads to dominantly inherited ALS32, 33. Furthermore, a large number of additional neurodegenerative diseases are also associated with mutations in RNA binding proteins indicating that defects in RNA metabolism may be a common underlying mechanism causing neurodegeneration30. Our work suggests that TDP-43 may play a role in regulation of mRNA at multiple levels that may include transcription, stability, trafficking and translation. Other RNA binding proteins mutated in neurodegenerative disease are similarly multifunctional, including SMN, FUS/TLS, and FMRP. It remains to be determined whether any one particular aspect of RNA metabolism is perturbed in common amongst these diseases.
TDP-43 has a well described role in the nucleus in the negative regulation of splicing, specifically it has been shown to promote exon skipping by direct interaction with the CFTR mRNA34. In the cytoplasm, TDP-43 has been shown to stabilize the mRNA of the neurofilament light chain through direct interaction with mRNA17. Recently, it has been shown that TDP-43 interacts with 14-3-3 protein subunits (also identified in our screen) to modulate the stability of the NFL mRNA35. Another intriguing possibility is that TDP-43 is required for site specific translation of specific mRNAs. Previous work has shown localization of TDP-43 in RNA granules in the dendrites of hippocampal neurons and repression of translation in vitro36. Altered regulation of site specific translation of mRNAs in motor neurons may prove to be an important mechanism leading to development of TDP-43 proteinopathies. Thus, future studies will be required to determine specific mRNAs that associate with TDP-43 in neurons.
TDP-43 pathology in ALS, FTLD-TDP and IBMPFD is typically characterized by clearance of TDP-43 from the nucleus and accumulation in the cytoplasm of affected cells2. Thus, diseases mediated by TDP-43 could involve loss of TDP-43 nuclear function or gain of a toxic of function in the cytoplasm. Given the dominant mode of inheritance of ALS associated with TDP-43 mutations5-7, and insight derived from our Drosophila model of TDP-43-related disease (Ritson et al. submitted) we hypothesize that toxic gain of cytoplasmic function is more likely.
TDP-43 associates with two distinct protein interaction networks, one implicated in RNA metabolism nucleus and the other involved in mRNA metabolism in the cytoplasm. Many of these interactions are dependent upon the ability of TDP-43 to bind RNA. TDP-43 interactions are not altered by two mutations that are causative of ALS. The association of TDP-43 with translational machinery, as well as histological evidence of TDP-43 assocaition with stress granules, strongly suggests that TDP-43 plays a role in translational regulation.
The authors thank members of the Taylor lab, particularly Natalia Nedelsky, for helpful discussion and advice. This work was supported by NIH grant AG031587 and a grant from the Packard Foundation for ALS Research at Johns Hopkins to JPT.