The Tau Tubulin Kinases TTBK1/2 Promote Accumulation of Pathological TDP-43

Pathological aggregates of phosphorylated TDP-43 characterize amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration (FTLD-TDP), two devastating groups of neurodegenerative disease. Kinase hyperactivity may be a consistent feature of ALS and FTLD-TDP, as phosphorylated TDP-43 is not observed in the absence of neurodegeneration. By examining changes in TDP-43 phosphorylation state, we have identified kinases controlling TDP-43 phosphorylation in a C. elegans model of ALS. In this kinome-wide survey, we identified homologs of the tau tubulin kinases 1 and 2 (TTBK1 and TTBK2), which were also identified in a prior screen for kinase modifiers of TDP-43 behavioral phenotypes. Using refined methodology, we demonstrate TTBK1 and TTBK2 directly phosphorylate TDP-43 in vitro and promote TDP-43 phosphorylation in mammalian cultured cells. TTBK1/2 overexpression drives phosphorylation and relocalization of TDP-43 from the nucleus to cytoplasmic inclusions reminiscent of neuropathologic changes in disease states. Furthermore, protein levels of TTBK1 and TTBK2 are increased in frontal cortex of FTLD-TDP patients, and TTBK1 and TTBK2 co-localize with TDP-43 inclusions in ALS spinal cord. These kinases may represent attractive targets for therapeutic intervention for TDP-43 proteinopathies such as ALS and FTLD-TDP.

TDP-43 undergoes a number of pathological modifications in disease-affected neurons including ubiquitination, phosphorylation, and proteolytic processing. These modifications may promote aggregation and the formation of detergent-insoluble inclusions. The precise molecular cause underlying neurotoxicity in most TDP-43 proteinopathies remains unclear, although the toxicity of mutant TDP-43 expressed in multiple model systems indicates it may be acting through a gain-of-function mechanism via aberrant interactions with proteins and/or nucleic acids [20]. Phosphorylation is a robust and consistent hallmark of pathological TDP-43, and detection of phosphorylation at tandem serines 409 and 410 characterizes virtually all TDP-43 proteinopathy cases [21,22].
In order to investigate the causes driving pathological TDP-43 phosphorylation, we have developed a C. elegans model of TDP-43 proteinopathy exhibiting TDP-43 phosphorylation dependent neurodegeneration and neurotoxicity; in C. elegans, phosphorylation of TDP-43 at serines 409 and 410 suffices to promote TDP-43 mediated neurotoxicity [14]. Further, we have used the model to previously identify the kinase CDC7 as a direct modulator of TDP-43 motor phenotypes [23]. This work also showed multiple kinases regulate TDP-43 phosphorylation in C. elegans, because detectable phosphorylated TDP-43 remains in the absence of CDC7. Inhibition of the kinases CDC7 or CK1 has also been shown to reduce but not eliminate TDP-43 phosphorylation in cultured cells [23,24]. Here we utilize the direct detection of changes in TDP-43 phosphorylation by immunoblot analysis of TDP-43 phosphorylation state to discover additional TDP-43 kinases in C. elegans. We have identified homologs of the tau tubulin kinases TTBK1 and TTBK2 and characterized their function as regulators of TDP-43 phosphorylation. TTBK1/2 may be attractive drug targets for therapeutic interventions in TDP-43 proteinopathies such as FTLD-TDP and ALS.

RNAi screen for TDP-43 kinases controlling pS409/410 TDP-43 levels
To identify TDP-43 kinases, we undertook a comprehensive survey utilizing kinase-targeting RNAi coupled with direct immunoblot detection of changes in TDP-43 phosphorylation in C. elegans. We have assembled an RNAi library targeting 451 predicted kinase genes in C. elegans (95% coverage of the predicted kinases found in the C. elegans genome, Table S1). This library has been previously employed to identify kinase modifiers of TDP-43 dependent behavioral phenotypes, and identified CDC7 as a direct TDP-43 kinase responsible for promoting TDP-43 neurotoxicity [23]. However, CDC7 is not solely responsible for the phosphorylation observed in our C. elegans model as detectable phosphorylation at S409/410 is still observed in a cdc-7(2/2) null mutant background. Thus other kinases play conserved roles phosphorylating TDP-43, and previous behaviorbased screening may have failed to uncover kinases with multiple roles in vivo, or kinases whose loss of function could adversely impact motor function or viability independent of TDP-43. To identify additional TDP-43 kinases, a direct biochemical assay of TDP-43 phosphorylation in TDP-43 transgenic C. elegans was used to screen for alterations in pS409/410 TDP-43 phosphorylation. Populations of transgenic C. elegans expressing ALSmutant M337V TDP-43 were grown on bacteria producing double stranded RNA targeting each kinase, then harvested and tested by immunoblot for changes in TDP-43 phosphorylation (S1 Figure). Transgenic C. elegans expressing ALS mutant TDP-43 exhibit post-translational modification of TDP-43 including prominent phosphorylation [14] in addition to altered proteolytic processing and ubiquitination. Candidate TDP-43 modifying kinases were selected whose knockdown by RNAi robustly reduced the observed TDP-43 phosphorylation relative to control treated animals. Apparent hits were retested by RNAi and immunoblot to confirm decreased TDP-43 phosphorylation, and the identity of positive RNAi clones was confirmed by direct DNA sequencing. Candidate kinases with human homologs acting on serine and/or threonine residues (S/T) were selected for further analysis. A total of 7 candidate S/T kinases were identified that consistently decreased TDP-43 S409/410 phosphorylation following RNAi treatment (Table 1). Interestingly, two of these kinases, cdc-7 and mlk-1, were identified previously in behavior-based screening for TDP-43 kinases [23]. Behavior-based screening also identified three additional homologs of the mammalian tau tubulin kinases TTBK1 and TTBK2, in the CK1 group. The CK1 group of kinases has greatly expanded in C. elegans, from 12 members in humans to 86 members in C. elegans, including 32 TTBK and TTBKL (TTBK-like) family members [25]. The dramatic expansion of the CK1 family of kinases in C. elegans suggests a diversification of functional roles for the TTBK1/2 like kinases in the nematode.
RNAi can inactivate multiple genes simultaneously depending on their sequence similarity, potentially confounding the identification of any single gene responsible for TDP-43 phosphorylation. To unambiguously determine the effects of single kinase gene loss of function on TDP-43 phosphorylation, we generated TDP-43 transgenic animals with viable deletion mutants eliminating the kinase active domain of each candidate gene of interest ( Table 1). Each of these kinase mutants was tested for changes in the amount of phosphorylated TDP-43 by immunoblot. Three of the kinase loss of function mutations tested, cdc-7(2/2), H05L14.1(2/2), and dkf-2(2/2), dramatically reduce TDP-43 phosphorylation with only moderate or no changes in total levels of TDP-43, consistent with the results from the initial RNAi screen (Fig. 1A-C and S2 Figure). We observed a slight decrease in levels of a shorter 37 kDa isoform of TDP-43 (Fig. 1A), but the appearance of higher or lower molecular weight species, including multimers, posttranslationally modified protein species, or translational variants, appears relatively unchanged (see S2 Figure for the full a-TDP-43 immunoblot), and after quantitation, only dkf-2(2/2) exhibited significant differences in total TDP-43 levels. cdc-7(2/2) has been previously characterized as a TDP-43 kinase [23], but we are including analysis of its mutant phenotypes in Fig. 1 for comparison with H05L14.1(2/2) and dkf-2(2/2).

Author Summary
Aggregated proteins are a hallmark of many neurodegenerative diseases. In ALS and FTLD-TDP, these aggregates contain abnormal TDP-43 modified by phosphorylation. Protein phosphorylation normally controls protein activity, stability, or location, but in some neurodegenerative diseases the phosphorylated proteins accumulate in excess. Kinases are the enzymes responsible for protein phosphorylation. We have identified two TDP-43 kinases, TTBK1 and TTBK2, using a novel approach combining reverse genetics and biochemical screening to identify the kinases responsible for changes in TDP-43 phosphorylation. We show TTBK1 and TTBK2 directly phosphorylate TDP-43 in vitro, and control TDP-43 phosphorylation in cellular and simple animal models of ALS. This has uncovered a molecular mechanism by which pathological phosphorylated TDP-43 can occur in disease. To determine whether changes in TTBK1/2 protein are contributing to TDP-43 pathology, we examined diseased brain and spinal cord tissue from patients with ALS or FTLD-TDP. We observed changes in the abundance of TTBK1 and TTBK2 in disease-affected neurons, and the coexistence of TTBK1/ 2 with phosphorylated TDP-43 aggregates in both FTLD-TDP and ALS. Therefore, increased abundance or activity of TTBK1 or TTBK2 may contribute to the neurodegeneration observed in ALS and FTLD-TDP. The human homologs of C. elegans genes are the best candidates identified by BLAST protein analysis (HUGO gene nomenclature). (C) C. elegans kinases are assigned to a kinase family and group based on protein sequence analysis [50].
The number of kinase family members identified as TDP-43 suppressors is compared to the total number of kinases within that family. alone. We assessed motor function by measuring the average dispersal velocity of the animals, and found significant improvements compared to TDP-43 (Fig. 1D). These results are consistent with the hypothesis that phosphorylation at S409/410 promotes TDP-43 toxicity, and decreased phosphorylation of TDP-43 will ameliorate the deleterious motor effects resulting from pathological TDP-43.

TTBK1/2 and PRKD2/3 are human homologs of TDP-43 kinases
To identify human homologs of H05L14.1 and dkf-2, we performed an unbiased search for related proteins from eukaryotes within the phylum chordata, including all vertebrate animals. This search employed a basic local alignment search tool (BLAST) [26], followed by automated construction of a phylogenetic tree with the top 50 hits from the search (S3A Figure) [27]. H05L14.1 is related to the human kinase TTBK1, although it is one of many members from an expanded family in C. elegans and other ecdysozoa. The H05L14.1 kinase domain has 40% sequence identity to the highly homologous tau tubulin kinases TTBK1 and TTBK2 at the amino acid level (S3C, D Figure) [28]. Variants in the gene coding for TTBK1 are associated with Alzheimer's disease, while mutation in TTBK2 causes spinocerebellar ataxia 11 (SCA11), both of which are characterized by pathologic alterations of tau [26][27][28]. dkf-2 is related to the conserved protein kinase D family, and is the major representative of the family in C. elegans (S3B Figure). The dkf-2 kinase domain has greater than 70% sequence identity to protein kinase D2 and D3 (PRKD2 and PRKD3) (S3E, F Figure). PRKD2/3 may be involved in cell proliferation, and dkf-2 has been shown to regulate C. elegans innate immunity (Table 1) [29,30]. Interestingly, our previous search for TDP-43 kinases identified another C. elegans homolog of TTBK1/2 [23]. This kinase, C55B7.10 also decreased TDP-43 phosphorylation and improved locomotion in C. elegans, although we were unable to determine a direct relationship between human TTBK1/2 and TDP-43 at that time. However, since our last study, we learned TTBK1/2 require millimolar levels of bivalent metal ions Mg 2+ or Mn 2+ in the reaction buffer for effective kinase activation [31]. We changed our in vitro kinase assay buffer composition, optimizing the reaction conditions for TTBK1/2 kinase assays. The quality of purified TTBK1/2 kinases also affects their activity in vitro. We compared purified TTBK1/2 from different commercial sources side by side in an in vitro kinase assay against a known target, tau, and found major differences in kinase activity (S4A Figure). Our previous characterization of TTBK1/2 as potential TDP-43 kinases used commercially available purified kinase with low activity against tau. Switching to a more active kinase preparation and modifying the buffer composition in the assay allowed a reassessment of these potential TDP-43 kinases in vitro.
Human TTBK1/2 directly phosphorylate TDP-43 TDP-43 kinases may act directly by phosphorylating TDP-43 S409/410 or may act indirectly by regulating the activity of other direct TDP-43 kinases. The amino acid sequence in the Cterminus of TDP-43 near S409/410 is consistent with the known CK1 family kinase consensus sequence S/TpXXS/T [32]. The PRKD kinase consensus sequence LXRXMSXXSFX [33], does not conform well with the sequence of human TDP-43. To empirically determine whether human TTBK1/2 or PRKD2/3 are direct TDP-43 kinases, we tested the ability of purified active kinase enzymes to phosphorylate TDP-43 at S409/410 and S403/ 404 in vitro (Fig. 1E, F, S4B Figure). We found that TTBK1 and 2 can directly phosphorylate both wild-type (WT) and familial ALS mutant TDP-43 (M337V TDP-43) under optimized reaction conditions that include magnesium. These conditions support robust phosphorylation of human tau protein, a known substrate of TTBK1/2 (S4A Figure, [31]). Although our preparation of PRKD2 kinase was enzymatically active against a known phosphorylation substrate, histone H1 [34] (S4C Figure), PRKD2 was unable to phosphorylate TDP-43 under any conditions tested, indicating its effect on TDP-43 phosphorylation may be indirect through the activation of other direct TDP-43 kinases or regulation of other downstream members of a TDP-43 regulatory pathway. If the kinases CDC7, TTBK1/2, or PRKD2/3 are in a common regulatory pathway, they may directly phosphorylate one another. Using an in vitro kinase assay with purified human kinases, we observed robust auto-phosphorylation by TTBK1 and modest auto-phosphorylation by TTBK2 and PRKD2, consistent with known activities of these kinases [28,31,34]. We also tested pairwise combinations of these kinases to determine any relative increases in phosphorylation. However, we did not see any significant increases in phosphorylation on these kinases (S4D Figure). Therefore, any indirect regulation of TDP-43 phosphorylation by PRKD2 may be through other unknown members of one or several regulatory pathways controlling TDP-43 phosphorylation.

TTBK1/2 promote TDP-43 phosphorylation in vivo
TTBK1/2 kinase hyperactivity may contribute to the pathological phosphorylated TDP-43 observed in both FTLD-TDP and ALS. To test whether increased cellular levels of TTBK1/2 activity suffice to drive TDP-43 phosphorylation, we transfected full-length TTBK1 and TTBK2 cDNAs into HEK293 cells. HEK293 cells have some neuronal characteristics and may be derived from a subpopulation of neuronal precursor cells in the embryonic kidney [35]. This cell line is especially useful for biochemical assays requiring high efficiency transfection rates. In the absence of other cellular stresses, we observed robust induction of TDP-43 phosphorylation by immunoblot following transfection with both TTBK1 and TTBK2 ( Fig. 2A-C). Likewise, we utilized SH-SY5Y cells, a human neuroblastoma-derived cell line, to determine the location of phosphorylated TDP-43 produced by TTBK2 transfection. The phospho-TDP-43 produced by TTBK2 overexpression is localized throughout the cytoplasm overlapping with TTBK2 (Fig. 2D, E). Further, TTBK2 and phospho-TDP-43 appear concentrated in apparent aggregates, producing a pattern of TDP-43 and TTBK1/2 expression reminiscent of the neuronal cytoplasmic inclusion pathology observed in FTLD-TDP and ALS. SH-SY5Y cells are relatively recalcitrant to transfection; we observed less than 5% transfection efficiency with TTBK2. However, all the cells with strong TTBK2::GFP expression also had inclusions of phosphorylated TDP-43. We observed a similar pattern of TTBK2 transfection overlapping with large phospho-TDP-43 positive aggregates in HEK293 cells (S5 Figure).
Decreasing TTBK1/2 kinase activity may prevent TDP-43 phosphorylation. To test this hypothesis, we employed small interfering RNAs (siRNAs) to decrease levels of TTBK1 gene expression in mammalian cultured cells. We have modeled pathological TDP-43 phosphorylation in the mouse motor neuron-like NSC-34 cell line using the chemical trigger ethacrynic acid (EA). EA acts by depleting cytosolic and mitochondrial glutathione, resulting in robust TDP-43 phosphorylation [36,37]. EA is a specific trigger of TDP-43 phosphorylation, because a variety of other cell stressors fail to induce phospho-TDP-43 (S6A Figure). NSC-34 cells were transfected with siRNAs targeting TTBK1, averaging 76% reduction in gene expression and 46% reduction in protein levels (S6B-D Figure). These cells were then treated with EA to induce TDP-43 phosphorylation. We observed a robust decrease in TDP-43 phosphorylation following treatment with TTBK1 siRNA (Fig. 3). We also tested siRNAs targeting TTBK2, but were unable to achieve significant reduction in gene expression.

TTBK1/2 co-localize with phospho-TDP-43 positive aggregates in FTLD-TDP and ALS
Both TTBK1 and TTBK2 are expressed in the brain, although TTBK2 is expressed in other tissues as well [31,38,39]. If TTBK1 or TTBK2 promote TDP-43 phosphorylation in patients with ALS or FTLD, there may be alterations in kinase abundance or localization, and there should be co-occurrence of the kinase with pathological TDP-43 aggregates. Immunohistochemistry for TTBK1, TTBK2 and phospho-TDP-43 was performed on frontal cortex sections from 6 FTLD-TDP cases, 6 ALS cases and 6 normal control cases to determine if there was overlap in the expression of these kinases and their purported target. Additionally, ALS spinal cord and hippocampus were also assessed. One FTLD case carried a progranulin mutation, the remaining 5 are of unknown etiology. FTLD cases were scored according to the harmonized FTLD-TDP classification of pathology [40]. Three of these cases were classified as Type A and three were Type B. All ALS cases were sporadic incidences of disease, and were negative for mutations in TDP-43, SOD1, FUS, and C9ORF72. TTBK1/ 2 antibody specificity was confirmed against purified substrate, and by antibody competition on fixed tissue (S7 Figure). Fig. 4A-D demonstrates that TTBK1 and TTBK2 immunoreactivity is present in a subset of pyramidal neurons in the frontal cortex of both normal and FTLD cases. Immunoreactivity is more prominent in cortical layers II-VI compared to cortical layer I, where immunoreactivity is relatively sparse, and the cellular localization appears both nuclear and cytoplasmic (Fig. 4, insets). Furthermore, the distribution of TTBK1 and TTBK2 immunoreactivity appears to be more widespread in FTLD cases compared to normal controls. Optical density measurements relative to the proportional area for TTBK1 and TTBK2 immunostaining in frontal cortex confirmed a statistically significant increase in both TTBK1 (Fig. 4E) and TTBK2 (Fig. 4F) immunoreactive distribution in disease-affected subjects. This increase was observed in all FTLD cases surveyed relative to controls. Importantly, the distribution of TTBK1 and TTBK2 in the frontal cortex is consistent with the distribution of phosphorylated TDP-43 pathology in FTLD cases, where aggregates are sparse in cortical layer I, and more abundant in cortical layers II-VI, depending on the FTLD classification (Fig. 4G). To further demonstrate this relationship, we performed double label immunohistochemistry to determine if the tau tubulin kinases and phosphorylated TDP-43 co-expressed within the same neurons. Most neurons immunoreactive for phospho-TDP-43 were also immunoreactive for TTBK1 and TTBK2 (Fig. 4H, I).
Of the six ALS cases examined, only two had phospho-TDP-43 aggregates in the frontal cortex and hippocampus, while all six demonstrated phospho-TDP-43 aggregates within spinal cord. ALS spinal cord motor neurons immunoreactive for phospho-TDP-43 pathology also co-labeled with TTBK1 and TTBK2 (Fig. 5A, B). Of the two ALS cases with pathologic changes in brain, a subset of neurons in the hippocampus and frontal cortex containing phospho-TDP-43 aggregates also co-expressed TTBK1 and TTBK2, while other neurons appeared to be immunoreactive for phospho-TDP-43 alone (Fig. 5 C-H). To test whether TTBK1/2 co-localize with phosphorylated TDP-43, we performed double-label immunofluorescence on ALS spinal cord sections ( Fig. 6 and S8 Figure). In general, more neurons were immunofluorescent for TTBK1/2 than for phosphorylated TDP-43. Similar to our double label immunohistochemical data, neurons immunofluorescent for phosphorylated TDP-43 usually co-localized with TTBK1/2, although some neurons labeled with phosphorylated TDP-43 alone. Taken together Figs. 4, 5 and 6 repeatedly demonstrate an overlapping expression pattern for TTBK1/2 and pS409/410 TDP-43 inclusions in ALS and FTLD-TDP consistent with TTBK1/2 participation in the genesis of TDP-43 lesions.

Discussion
Tandem phosphorylation at TDP-43 serines 409 and 410 (pS409/410) is a consistent and robust feature of TDP-43 pathology in ALS and FTLD-TDP. Our previous work in TDP-43 transgenic C. elegans demonstrated a causal relationship between neurodegeneration and S409/410 phosphorylation of TDP-43 [14,23]. We have utilized this model as a C. elegans behavior-based screening tool to identify TDP-43 kinases [23]. However, it is possible other relevant TDP-43 kinases remain unidentified. To uncover kinases responsible for the pathological phosphorylation of TDP-43, we have re-screened the C. elegans kinome by RNAi knockdown for modifiers of TDP-43 phosphorylation. For this survey we employed sensitive and specific S409/ 410 phosphorylation dependent TDP-43 antibodies [41] to directly detect changes in TDP-43 phosphorylation state following RNAi treatment. Confirmation of identified candidate kinases in C. elegans was conducted by testing deletion mutations within the kinase genes of interest. Three identified candidate kinases, cdc-7, dkf-2, and H05L14.1, reduced TDP-43 S409/410 phosphorylation and improved TDP-43 dependent behavioral phenotypes in C. elegans. TDP-43 is a known substrate of CDC7, as it was previously uncovered in a reverse genetic screen to identify modifiers of TDP-43 behavioral phenotypes [23], confirming the validity of this approach. We employed standard BLAST protein sequence homology searching algorithms [42] to identify the closest mammalian homologs of our novel TDP-43 kinases dkf-2 and H05L14.1. Interestingly, H05L14.1 was a homolog of the mammalian tau tubulin kinases TTBK1 and TTBK2. Our previous behavior-based screen for TDP-43 kinases identified a different C. elegans homolog of TTBK1/2 as a TDP-43 kinase, although at the time we were unable to demonstrate a direct   , and immunoreactivity appears to be more widespread in FTLD cases relative to normal controls. Cortical layers I-VI are indicated (C). Quantification of immunostaining demonstrated a statistically significant increase in both TTBK1 (E) and TTBK2 (F) in FTLD cases compared to normal controls (**P = 0.003; ***P,0.0001). The distribution of phospho-TDP-43 immunoreactivity in the cortex of an FTLD relationship between human TTBK1/2 and TDP-43 [23]. We decided to re-evaluate TTBK1/2 as we had identified both as candidate kinases in independent assays. Using optimized in vitro kinase reaction conditions, we demonstrate here that human TTBK1/2 are able to directly phosphorylate TDP-43. We then overexpressed TTBK1/2 in cultured human cells. TTBK1/2 overexpression in the absence of other stressors promoted robust phosphorylation of endogenous TDP-43. Furthermore, this phosphorylated TDP-43 localized to the cytoplasm in inclusionlike aggregates. We also found that reduction of TTBK1 mRNA levels attenuated TDP-43 phosphorylation in a chemically induced model of pathological phospho-TDP-43 accumulation. Finally, to explore whether TDP-43 kinase hyperactivity could underlie the etiology of TDP-43 proteinopathies, we immunostained tissue from FTLD-TDP and ALS for TTBK1/2. We observe increased TTBK1/2 in FTLD-TDP frontal cortex, and co-localization with TDP-43 positive aggregates in FTLD frontal cortex and ALS spinal cord. One possible explanation for these data is the observed differences in TTBK1/2 expression drives neurodegeneration in TDP-43 proteinopathies. Taken together these data support a pivotal role for TTBK1/2 hyperactivity in TDP-43 proteinopathy.
A number of kinases have been identified to date with the ability to phosphorylate TDP-43 in vitro and in vivo. The kinases CK1 [43,44], CKII [45], CDC7 ( [23], this study), TTBK1 and TTBK2 (this study) may all contribute to pathological TDP-43 phosphorylation in humans; regardless they all share target sequence conservation as the CK1 kinase domain is the prototypical model for this family of kinases. CK1 family kinases may act redundantly to regulate TDP-43 phosphorylation in a common signaling pathway. Alternatively, extracellular and intracellular signals may act as a trigger to specify kinase activity from one of the available TDP-43 kinases but not the others. We have observed that in the absence of each of these known TDP-43 kinases in C. elegans, mutant TDP-43 still exhibits varying but detectable degrees of phosphorylation ( Fig. 1 and [23]), indicating that no one kinase accounts for all observed TDP-43 phosphorylation. Exploring the functional relationships between and regulatory networks governing the TDP-43 kinases identified to date will be important future work.
TTBK1 and TTBK2 were originally purified on the basis of their kinase activity on the microtubule binding protein tau at several pathological phospho-tau epitopes known to accumulate in Alzheimer's disease [31,38,39,46]. Tangles composed of insoluble hyperphosphorylated tau are a pathological hallmark of Alzheimer's disease (AD), as well as a number of other neurodegenerative diseases including FTLD-tau, progressive supranuclear palsy (PSP), and chronic traumatic encephalopathy (CTE). Interestingly, phosphorylated TDP-43 is also present in a subset of patients with primary tauopathies such as AD, PSP, and CTE [8,47,48], and either tau or TDP-43 are the diagnostic pathologic changes in the vast majority of frontotemporal lobar degeneration cases [49]. The relationship between TDP-43 and tau neuropathologic changes remains unclear. One hypothesis is that both proteinopathy disorders share a common etiology in TTBK1/2 activation leading to either TDP-43 or tau neuropathology depending on the vulnerable cell population affected by TTBK activation. The downstream toxic mechanisms for tau and TDP-43 appear distinct; however, inappropriate TTBK1 and TTBK2 activity may constitute a shared mechanistic link in initiating both tau and TDP-43 neuropathologies.
Both TTBK1 and TTBK2 have been previously implicated in neurodegenerative diseases. Single nucleotide polymorphisms (SNPs) in TTBK1 are associated with decreased Alzheimer's disease risk in studies of Spanish and Han Chinese populations [50,51]. TTBK1 has been shown to co-localize with diffuse phospho-Ser422 tau in pre-tangle Alzheimer's disease neurons [52], and increased levels of TTBK1 have been observed in AD frontal cortex [53] and enhance the toxicity of tau in a P301L mouse model [54]. Mutations in TTBK2 have been shown to cause spinocerebellar ataxia type 11 (SCA11) [55], a progressive neurodegenerative disorder characterized by tau pathology. Mouse models heterozygous for mutant TTBK2 exhibit decreased TTBK2 kinase activity and altered TTBK2 localization, while homozygous mutant TTBK2 is embryonic lethal [56]. Our results are the first demonstration of a potential role for TTBK1 and TTBK2 in primary TDP-43 proteinopathies. Kinases regulating TDP-43 phosphorylation present an attractive target for therapeutic intervention in both ALS and FTLD-TDP. No specific small molecule inhibitors targeting TTBK1 or TTBK2 has been reported to date, despite their potential roles contributing to tauopathies by hyperphosphorylating tau. Development of brain penetrant TTBK1 and TTBK2 inhibitors may also provide a viable strategy for intervening in TDP-43 proteinopathy disorders including ALS and FTLD-TDP.

RNAi screen
The list of predicted kinase genes in C. elegans was derived from the C. elegans kinome project [58], with library construction as described [23]. Testing was done in an eri-1(2/2);lin-15(2/2) RNAi enhancing mutant background [59]. Staged embryos were plated, grown at 16uC for 8-9 days, and then a mixed population of 1 st generation gravid adults with 2 nd generation L2-L3 animals were harvested by washing with M9 buffer into 96 well plates and frozen at 280uC, for subsequent immunoblot analysis. Each RNAi treated population was evaluated semi-quantitatively for reduction in phospho-TDP-43 relative to control treated animals. Positives candidates were retested for effects on TDP-43 phosphorylation by independent RNAi treatment and immunoblot, and the RNAi gene target for each plasmid was confirmed by sequencing.

Immunoblotting
Equivalent mixed-stage worm lysate fractions were loaded and resolved on precast 4-15% gradient SDS-PAGE gels and transferred to PVDF membrane as recommended by the manufacturer (Bio-Rad). On immunoblots, human TDP-43 was detected with a commercially available monoclonal antibody ab57105 (Abcam) directed against human TDP-43 amino acids 1-261. TDP-43 phosphorylated at S409/S410 was detected by a monoclonal antibody called anti phospho TDP-43 (pS409/410) available from Cosmobio (catalog # TIP-PTD-M01). C. elegans b-tubulin levels were measured using monoclonal antibody E7 as a loading control as previously described [60,61]. TTBK1 was detected by Abcam rabbit polyclonal antibody ab103944 at 1:1000 dilution. TTBK2 was detected by Abgent rabbit polyclonal antibody AP12162a at 1:1000 dilution. HRP labeled goat antimouse IgG was the secondary antibody (GE Healthcare) and used at a dilution of 1:4000. Dilutions were: 1:7500 for ab57105, 1:1000 for pS409/410, and 1:10000 for E7. Immunoblots shown are representative of at least 3 independent experiments. Quantitation was performed using ImageJ image processing and analysis software.

Immunofluorescence for cultured cells
Cells were seeded onto poly-D-Lysine coated (Sigma Aldrich) 12 mm round glass cover slips in 24-well plates. Cells were transfected with the plasmid encoding TTBK2-GFP with GenePorter 2 (Genlantis) using the manufacturer's protocol. Cells were fixed for imaging in 4% formaldehyde 96 hours after transfection. Cells were washed 365 min in PBS/Ca 2+ /Mg 2+ , then blocked in antibody buffer (PBS, 0.5% Triton X-100, 1 mM EDTA, 0.1% BSA, 0.05% NaN 2 )+10% normal goat serum. Primary antibody was applied and incubated for 1 hour at room temperature (Cosmo Bio; 1:1000). Cells were washed 365 min in PBS/Ca 2+ /Mg 2+ , then re-blocked for 10 min. Appropriate secondary antibody was applied and incubated for 20 min at room temperature. Cells were again washed 365 min in PBS/ Ca 2+ /Mg 2+ , counterstained with 300 nM DAPI and mounted with ProLong Gold antifade. Microscopy was performed on a Delta Vision microscope (Applied Precision, Inc) using a 606 oil immersion objective, a sCMOS camera, and 262 binning. Image analysis was performed using softWoRx 6.0 Beta software.
Quantitative reverse-transcription PCR RNA was purified from flash-frozen cell pellets using TRIzol Reagent (Life Technologies) according to the manufacturer's protocol. cDNA was made using iScript Reverse Transcription Supermix (Bio-Rad). qPCR was performed on an 7900HT Real Time PCR System (Applied Biosystems) using iTaq Universal SYBR Green Supermix (Bio-Rad).

Ethics statement: Post mortem human tissue
De-identified post-mortem brain tissue used in this study was determined to be an exempt from IRB review by the VA Puget Sound Health Care System Human Research Protection Program Director on December 29, 2011. Tissue used for these studies was obtained from the University of Washington Alzheimer's Disease Research Center brain bank (Seattle, WA), and the Indiana Alzheimer Disease Center brain bank(Indianapolis, IN), where consent for autopsy and permission for use of tissue in scientific experiments was obtained. FTLD and ALS cases were selected on the basis of having an autopsy-confirmed diagnosis of FTLD and FTLDrelated disorders or ALS. Control samples were from de-identified neurologically healthy control participants, who were of a similar age.

Immunohistochemistry and immunofluorescence for tissue
Primary antibodies used for immunohistochemistry were anti-TTBK1 (Abcam, 1:100), anti-TTBK2 (Abgent, 1:200), and antiphospho TDP-43 409/410 (CosmoBio, 1:1000)). In order to minimize variability, sections from all cases (normal and affected subjects) were stained simultaneously for each antibody. Immunostained sections were analyzed using the computerized image analysis system, MicroComputer Imaging Device (MCID, Imaging Research, St. Catherines, Ontario, Canada). Blinded assessment of optical density measurements were obtained relative to the proportional area for TTBK1 and TTBK2 immunostaining in frontal cortex grey matter (three separate readings per case). Data were averaged and are represented as mean +/2 SEM. A two tailed Student's t-test was used to assess differences in TTBK1 and TTBK2 expression between cases and controls. For double label immunohistochemistry experiments, sections were first immunostained with anti-phospho TDP-43 and reaction product was visualized with nickel enhanced DAB (black). Sections were then immunostained with anti-TTBK1 or TTBK2 and visualized with DAB alone (brown). For double label immunofluorescence experiments, AlexaFluor 488 goat anti-rabbit and AlexaFluor 594 goat anti-mouse secondary antibodies (Molecular Probes) were used and autofluoresence was quenched with 0.1% Sudan Black [64]. To demonstrate specificity of the TTBK antibodies, TTBK1 and TTBK2 were blocked with 50-fold amount of immunizing peptide overnight at 4uC before proceeding with the immunostaining protocol (see S5 Figure).
Supporting Information S1 Figure Immunoblot results from primary kinase RNAi screen. Populations of RNAi treated C. elegans were harvested into 96-well plates prior to immunoblot analysis. Gene names and locations of kinases tested are presented in Table S1. Two rows from each plate were tested in alternating wells for each immunoblot. Labels above individual wells describe Row and Column information for each sample. Plate numbers are indicated at the left of the immunoblot. a-tubulin antibody is used as a load control. Candidates confirmed on repeat testing are boxed in blue. (PDF) The entire C. elegans amino acid sequences of H05L14.1 or dkf-2 were compared against non-redundant reference sequences from the RefSeq protein database (7-20-2014, NCBI). (A) 2878 active hits were identified by BLAST with similarity to H05L14.1, and subsequently filtered to include only sequences from C. elegans or phylum chordata. The top 50 hits underwent multiple sequence alignment, alignment refinement, phylogenetic reconstruction, and are displayed in a cladogram, with branch support values in red [27]. Related human gene and gene identifier is boxed. Homo sapiens GI# 58761548 is TTBK1. (B) 5000 active hits were identified with similarity to dkf-2, filtered, and graphed as above. Homo sapiens GI# 5031689 is PRKD3. (C) H05L14.1 kinase domain has 40% identity to human TTBK1 and TTBK2. (D) dkf-2 has more than 70% identity to human PRKD2 and PRKD3. Sequence identity was calculated using Clustal W method for multiple sequence alignment. (E) Alignment report for H05L14.1, TTBK1, and TTBK2 kinase domain, including boxes around sequence that matches the consensus. (F) Alignment report for dkf-2, PRKD2, and PRKD3. Reports were generated using Lasergene MegAlign software for protein sequence analysis and alignment. (PDF)

S4 Figure
In vitro kinase assay controls. (A) Tau is a known substrate of TTBK2 [65]. To test enzyme activity, approximately 1 mg of non-phosphorylated recombinant human tau purified from E. coli were incubated with equivalent amounts of TTBK2 enzyme purified from cultured cells by two commercial suppliers (Origene catalog #LY406582 and Signalchem #T18-11G). Phosphorylation was assessed by reactivity with AT270, a phospho-tau antibody recognizing tau phosphorylated at Thr181. (B) Purified TTBK1, TTBK2, and CDC7 can also phosphorylate TDP-43 at serines 403 and 404 (CosmoBio, #CAC-TIP-PTD-P05) in an in vitro kinase assay. (C) Histone H1 is a known substrate of PRKD2 [34]. To confirm PRKD2 activity, human PRKD2 (SignalChem #P76-10) purified from cultured cells was incubated with purified recombinant Histone H1. We observed phosphorylation of Histone H1 as detected by reactivity with pT146 specific antibody (Bioss Catalog # bs-3176R). (D) Purified CDC7, TTBK1, TTBK2, or PRKD2 were incubated singly or pairwise with radiolabeled phosphate. TTBK1 can robustly auto-phosphorylate, while TTBK2 and PRKD2 are also capable of auto-phosphorylation. Pairwise combinations of CDC7, TTBK1, TTBK2, and PRKD2 did not exhibit any increase or variety in phosphorylation beyond baseline levels of auto-phosphorylation for each kinase. Bafilomycin and wortmannin are inhibitors of autophagy, PSI is a general proteasome inhibitor, cadmium chloride is a heavy metal, taxol is an inhibitor of microtubule dynamics, rotenone blocks the mitochondrial electron transport chain (creating reactive oxygen species (ROS)), pepstatin inhibits aspartic proteases, and paraquat catalyzes formation of ROS. (B) TTBK1 is reduced by nearly 80% following siRNA treatment in NSC-34 cells. Quantitative PCR measurements (qPCR) for TTBK1 mRNA levels are displayed in arbitrary units for an untreated control and cells treated with TTBK1 siRNA.  Table Kinase genes tested by immunoblot. 96-well plate locations of each RNAi treated population of C. elegans prior to testing by immunoblot (S1 Figure). Control RNAi for each plate are highlighted in purple. L4440: empty vector RNAi control. unc-22: positive control for effective RNAi treatments, causing a strong paralyzed phenotype in treated TDP-43 worms. TDP-43: RNAi targeting the TDP-43 transgene is a positive control for suppression of TDP-43 phenotypes. Kinase RNAi treatments that caused C. elegans sterility or growth arrest, causing insufficient sample for protein detection by immunoblot, are highlighted in green. Kinase RNAi treatments that decreased TDP-43 phosphorylation in initial testing are highlighted in blue. Kinase RNAi treatments that reproducibly decreased TDP-43 phosphorylation in multiple independent experiments have names that are bolded and underlined. (XLSX)