Parkin Ubiquitinates Tar-DNA Binding Protein-43 (TDP-43) and Promotes Its Cytosolic Accumulation via Interaction with Histone Deacetylase 6 (HDAC6)*

Background: TDP-43 pathology and the role of E3 ubiquitin ligases are increasingly recognized in neurodegeneration. Results: Parkin ubiquitinates TDP-43 and forms a multiprotein complex with HDAC6 to sequester TDP-43 in cytosol. Conclusion: Parkin E3 ubiquitin ligase activity promotes TDP-43 inclusion formation and nuclear translocation. Significance: Parkin-TDP-43 interaction may be exploited as a therapeutic strategy in ALS/FTLD pathology. The importance of E3 ubiquitin ligases, involved in the degradation of misfolded proteins or promotion of protein-protein interaction, is increasingly recognized in neurodegeneration. TDP-43 is a predominantly nuclear protein, which regulates the transcription of thousands of genes and binds to mRNA of the E3 ubiquitin ligase Parkin to regulate its expression. Wild type and mutated TDP-43 are detected in ubiquitinated forms within the cytosol in several neurodegenerative diseases. We elucidated the mechanisms of TDP-43 interaction with Parkin using transgenic A315T mutant TDP-43 (TDP43-Tg) mice, lentiviral wild type TDP-43, and Parkin gene transfer rat models. TDP-43 expression increased Parkin mRNA and protein levels. Lentiviral TDP-43 increased the levels of nuclear and cytosolic protein, whereas Parkin co-expression mediated Lys-48 and Lys-63-linked ubiquitin to TDP-43 and led to cytosolic co-localization of Parkin with ubiquitinated TDP-43. Parkin and TDP-43 formed a multiprotein complex with HDAC6, perhaps to mediate TDP-43 translocation. In conclusion, Parkin ubiquitinates TDP-43 and facilitates its cytosolic accumulation through a multiprotein complex with HDAC6.

It has been reported that TDP-43 binds to Parkin mRNA and regulates its expression (29), suggesting a potential relationship between TDP-43 and Parkin. More recently, it was demonstrated that TDP-43 depletion results in down-regulation of Park2 mRNA in stem cell-derived human neurons and in motor neurons containing TDP-43 inclusions in sporadic ALS (30). Neurons bearing TDP-43 aggregates showed decreased cytoplasmic Parkin levels (30). Parkin is an E3 ubiquitin ligase involved in degradation of misfolded proteins (31), and Park2 mutations, leading to loss of the E3 ubiquitin ligase function, are linked to autosomal recessive early onset PD (32)(33)(34). Parkin functions as part of a number of multiprotein complexes, including PTEN-induced putative kinase 1 (PINK1) (35)(36)(37), the Skp1-Cullin-Fbox (SCF)-like complex to facilitate proteasomal degradation (38), the chaperone Hsp70, and the U-box protein C terminus of Hsc70-interacting protein (39). We previously demonstrated that wild type and not mutant loss-of-function Parkin increases proteasome activity (40 -42), leading to degradation of ubiquitinated proteins. Therefore, Parkin may be part of a protein complex that regulates, or perhaps is regulated by, TDP-43.
Accumulation of ubiquitinated TDP-43 in the cytosol may change the bioavailability of this predominantly nuclear protein (3)(4)(5)(6)(7)(8)(9), either leading to gain of cytosolic or loss of nuclear function. We previously showed that lentiviral expression of wild type TDP-43 can lead to pathological changes, including cleavage, aggregation, and phosphorylation (25,43). In these studies, we aimed to better understand the mechanisms of TDP-43 interaction with the E3 ubiquitin ligase Parkin. We used transgenic A315T mice, which express TDP-43 exclusively in the nucleus and display early motor symptoms (44), as well as lentiviral gene delivery that leads to expression of nuclear and cytoplasmic TDP-43 and allows examination of nondevelopmental early effects of TDP-43.

EXPERIMENTAL PROCEDURES
Stereotaxic Injection-Stereotaxic surgery was performed to inject the lentiviral (Lv) constructs encoding either LacZ, Parkin, and/or TDP-43 into the primary motor cortex of 2-monthold male Sprague-Dawley rats weighing between 170 and 220 g as described previously (40). Animals were injected into the left side of the motor cortex with 2 ϫ 10 9 m.o.i. Lv-LacZ and into the right side with 1 ϫ 10 9 m.o.i. Lv-Parkin ϩ 1 ϫ 10 9 m.o.i. Lv-LacZ, 1 ϫ 10 9 m.o.i. Lv-TDP-43 ϩ 1 ϫ 10 9 m.o.i. Lv-LacZ, or 1 ϫ 10 9 m.o.i. Lv-Parkin ϩ 1 ϫ 10 9 m.o.i. Lv-TDP-43. All animals were sacrificed 2 weeks post-injection, and the left cortex was compared with the right cortex. A total of eight animals in each treatment (32 animals) were used for WB, ELISA, and immunoprecipitation and eight animals in each treatment (32 animals) for immunohistochemistry. A total of n ϭ 64 animals were used. Transgenic hemizygous mice harboring human TDP-43 with the A315T mutation under the control of the prion promoter and C57BL6/J mouse controls were used (44). The colony was obtained from The Jackson Laboratory Repository (JAX stock no. 010700) and displayed a life span considerably shorter than previous reports (44), with almost 90% of all pups, including males and females, manifesting motor symptoms around 21-30 days. We bred hemizygous mice via mating of hemizygous with noncarrier wild type C57BL/6, and upon genotyping, half were identified as transgenic and the other half were nontransgenic control mice. All mice used are F1 generation from direct mating between hemizygous and C57BL/6 mice. These studies were approved and conducted according to Georgetown University Animal Care and Use Committee.
Parkin Enzyme-linked Immunosorbent Assay-ELISA was performed on brain-soluble brain lysates (in STEN buffer) or insoluble brain lysates (4 M urea) using a mouse-specific Parkin kit (MyBioSource) in 50 l (1 g/l) of brain lysates detected with 50 l of primary antibody (3 h) and 100 l of anti-rabbit antibody (30 min) at RT. Extracts were incubated with stabilized Chromogen for 30 min at RT, and solution was stopped and read at 45 0 nm, according to the manufacturer's protocol.
Parkin E3 Ubiquitin Ligase Activity-To determine the activity of Parkin E3 ligase activity, we used E3LITE customizable ubiquitin ligase kit (Life Sensors, UC 101), which measures the mechanisms of E1-E2-E3 activity in the presence of different ubiquitin chains. To measure Parkin activity in the presence or absence of substrates, we immunoprecipitated Parkin (1:100) with PRK8 antibodies and TDP-43 (1:100) with human TDP-43 (Abnova) from 100 mg of TDP43-Tg brain lysates. We used UbcH7 as an E2 that provides maximum activity with Parkin E3 ligase, and we added E1 and E2 in the presence of recombinant ubiquitin, including wild type containing all seven possible surface lysines, no lysine mutant (Lys-0), or Lys-48 or Lys-63 to determine the lysine-linked type of ubiquitin. We then added E3 as immunoprecipitation Parkin or recombinant Parkin (Novus Biologicals) to an ELISA microplate that captures polyubiquitin chains formed in the E3-dependent reaction, which was initiated with ATP at room temperature for 60 min. We also included an E1-E2-E3 and a polyubiquitin chain control in addition to E1, E2, and TDP-43 without Parkin and assay buffer for background reading. The plates were washed three times and incubated with detection reagent and streptavidin-HRP for 5 min, and the polyubiquitin chains generated by E1-E2-E3 machinery were read on a chemiluminescence plate reader.
Immunoprecipitation and Ubiquitination Assay-Either TDP-43 or Parkin was separately immunoprecipitated in 100 l (100 g of proteins) of 1ϫ STEN buffer using (1:100) humanspecific anti-TDP-43 monoclonal antibody (Abnova) or (1:100) anti-Parkin mouse monoclonal antibody (PRK8; Signet Labs; Dedham, MA), respectively. Following immunoprecipitation, 300 ng of each substrate protein (Parkin and TDP-43) were mixed in the presence of 1 g of recombinant human ubiquitin (Boston Biochem, MA), 100 mM ATP, 1 g of recombinant UbcH7 (Boston Biochem), 40 ng of E1 recombinant enzyme (Boston Biochem) and incubated at 37°C in an incubator for 20 min. The reaction was heat-inactivated by boiling for 5 min, and the substrates were analyzed by Western blot.
Immunohistology-Immunohistochemistry was performed on 20-m-thick sections of brain or cervical spinal cord. TDP-43 was probed (1:200) with rabbit polyclonal (ALS10) antibody (ProteinTech, catalog no. 10782-2-AP). Rabbit polyclonal anti-ubiquitin (Chemicon International) was used (1:100), and mouse monoclonal anti-Parkin (Millipore) antibody was used (1:200) for immunohistochemistry. Toluidine blue and DAPI staining were performed according to the manufacturer's instructions (Sigma). Counting of toluidine blue staining of centric axons within 10 random fields of each slide was performed by a blind investigator in n ϭ 8 animals from each treatment. All staining experiments were scored by a blind investigator to the treatments. 20 S Proteasome Activity Assay-Brain extracts of 100 g were incubated with 250 M of the fluorescent 20 S proteasome-specific substrate succinyl-LLVY-AMC at 37°C for 2 h. The medium was discarded, and homogenates were lysed in 50 mM HEPES (pH 7.5), 5 mM EDTA, 150 mM NaCl, and 1% Triton X-100, containing 2 mM ATP. The fluorophore AMC, which was released after cleavage from the labeled substrate succinyl-LLVY-AMC (Chemicon International, Inc.), was detected, and free AMC fluorescence was quantified using a 380/460-nm filter set in a fluorometer (absorption at 351 nm and emission at 430 nm). We measured nonproteasomal side reactivity by adding lactacystin as a specific proteasome inhibitor to the reaction mixture and subtracted these values from the total for an accurate measure of specific proteasome activity.
qRT-PCR in Neuronal Tissues-qRT-PCR was performed on PCR system (Applied Biosystems) with Fast SYBR Green PCR master mix (Applied Biosystems) in triplicate from reversetranscribed cDNA from control uninjected or lentiviral LacZ, Parkin, TDP-43, and TDP-43 ϩ Parkin injected rat cortical brain tissues. These experiments were repeated in human neuroblastoma M17 cells and A315T-Tg compared with nontrans-genic C57BL/6 controls. Human wild type Parkin forward primer CCA TGA TAG TGT TTG TCA GGT TC and a reverse primer GTT GTA CTT TCT CTT CTG CGT AGT GT were used. Gene expression values were normalized using GADPH levels.

TDP-43 Inhibits Proteasome Activity and Alters Parkin
Protein Levels-To determine the effects of TDP-43 on Parkin in transgenic animals, we used the A315T mutant TDP-43 transgenic mice (TDP43-Tg), which were reported to have aggregates of ubiquitinated proteins in layer five pyramidal neurons in frontal cortex, as well as spinal motor neurons, without cytoplasmic TDP-43 (44). This model is relevant to our studies because it shows nuclear TDP-43-driven pathology, independent of cytoplasmic TDP-43 inclusions (44). Western blot analysis showed accumulation of full-length and TDP-43 fragments (ϳ35 kDa) as well as higher molecular weight species with human TDP-43 antibody (Fig. 1A, 1st blot) compared with nontransgenic controls, suggesting TDP-43 pathology. Further analysis of the soluble brain lysate (STEN extract) showed increased Parkin levels by Western blot (Fig. 1A, 2nd blot, 82%, p Ͻ 0.05, n ϭ 8) and appearance of a lower molecular weight band, perhaps indicating Parkin cleavage. We also observed increased levels of ubiquitin smears (Fig. 1A, 3rd blot) using anti-ubiquitin antibodies, suggesting accumulation of ubiquitinated proteins. We previously showed that amyloid protein stress and aging alter Parkin solubility (45), so we aimed to determine whether Parkin solubility was altered in TDP43-Tg mice. We resuspended the protein pellet after STEN extraction in 4 M urea to detect the insoluble fraction, and we detected a significant increase (Fig. 1B, 95% by densitometry, p Ͻ 0.05, n ϭ 8) in insoluble Parkin in 1-month-old TDP43-Tg mice compared with control ( Fig. 1, B and C, p Ͻ 0.05, n ϭ 8), suggesting that TDP-43 aggregates are associated with altered Parkin solubility. The ratio of soluble over insoluble Parkin was not significantly changed (Fig. 1C, p Ͻ 0.05), suggesting that TDP-43 accumulation increases soluble and insoluble Parkin levels. We also probed for TDP-43 in 4 M urea extracts and detected increased levels of insoluble TDP-43 ( Fig. 1B, 2nd blot) in TDP43-Tg compared with control. To verify the changes in Parkin level observed by WB, we performed quantitative Parkin ELISA to determine the levels of both soluble (STEN extract) and insoluble (4 M urea) Parkin, using brain extracts from Parkin Ϫ/Ϫ mice as control for ELISA specificity (Fig. 1D, n ϭ 8). We detected a significant increase in both soluble (46%, p Ͻ 0.05) and insoluble (64%) Parkin in TDP43-Tg mice compared with control levels (Fig. 1D, p Ͻ 0.05, n ϭ 8), further suggesting an increase in Parkin level and insolubility in TDP43-Tg mice.
The seven in absentia homolog (SIAH) protein is another E3 ligase involved in ubiquitination and proteasomal degradation of specific proteins (46,47). SIAH is rapidly degraded via the proteasome (48). The activity of this ubiquitin ligase has been implicated in regulating cellular response to hypoxia, including hypoxia-inducible factor-1 (HIF-1␣) (49 -51). We used SIAH2 as an E3 ligase control to determine whether TDP-43 decreases Parkin solubility, leading to alteration of its E3 ligase function independent of other E3 ligases. Western blot analysis showed a significant increase (215%) in soluble SIAH2 levels (Fig. 1, E and F, p Ͻ 0.05, n ϭ 8) in TDP43-Tg mice compared with control, indicating lack of degradation of SIAH2 perhaps due to proteasomal impairment. However, SIAH2 was not detected in the insoluble fraction. We also observed a lower molecular mass band at 17 kDa (Fig. 1E) in transgenic mice, suggesting possible cleavage of SIAH2 dimeric structure (49 -51). Further examination of the level of SIAH2 target molecule HIF-1␣ showed a significant increase (76%, p Ͻ 0.05) in protein level (Fig. 1, E and  F), suggesting lack of proteasomal degradation.
To ascertain the effect of TDP-43 on Parkin level and proteasome activity, we expressed wild type TDP-43 (Fig. 1G, 1st blot) in the presence or absence of Parkin (Fig. 1G, 2nd blot) in human M17 neuroblastoma cells. Expression of TDP-43 alone led to the appearance of endogenous Parkin protein (Fig. 1G,  2nd blot), suggesting that TDP-43 regulates Parkin mRNA to induce protein expression (29). Co-expression of exogenous Parkin and TDP-43 led to a slight decrease in TDP-43 levels (Fig. 1G, 1st blot) and a noticeable decrease in ubiquitinated proteins (Fig. 1G, 3rd blot) compared with TDP-43 alone. SIAH2 was difficult to detect in control M17 cells (Fig. 1G, 4th blot), but it accumulated when TDP-43 was expressed despite the increase in endogenous Parkin; however, exogenous Parkin co-expression with TDP-43 led to disappearance of SIAH2 (Fig.  1G, 4th blot). We further compared the effects of Parkin expression alone (Fig. 1G, 2nd blot) compared with LacZ on TDP-43 and SIAH2 levels. No differences were observed between control ( Fig. 1F), LacZ-, and Parkin-transfected M17 cells (Fig. 1H) on endogenous TDP-43 expression levels (Fig. 1H, 1st blot). A higher level of ubiquitinated protein smears was observed with Parkin expression (Fig. 1H, 3rd blot), consistent with the role of Parkin as an E3 ubiquitin ligase, but the level of SIAH2 was significantly decreased (Fig. 1H, 4th blot, 74%, p Ͻ 0.05) compared with actin control. To determine whether SIAH2 accumulation is due to decreased E3 ligase activity or proteasomal function, we measured proteasome activity (Fig. 1I) and found that TDP-43 significantly decreased (66%) proteasome activity (p Ͻ 0.05, n ϭ 12), whereas Parkin co-expression significantly reversed proteasome activity to 74% of control or Parkin levels but remained significantly less (26%) than control. These data suggest that TDP-43 increases Parkin expression levels, whereas proteasomal inhibition leads to decreased degradation of proteins, including the rapidly degrading SIAH2.

Lentiviral Expression of TDP-43 in Rat Motor Cortex Results in Increased Protein Levels in Preganglionic Cervical Spinal
Cord Inter-neurons-We expressed wild type TDP-43 using lentiviral gene delivery into the motor cortex of 2-month-old Sprague-Dawley rats. We previously showed, using Western blot analysis, significantly increased levels of TDP-43 (41%) 2 weeks after lentiviral gene expression (43). Immunohistochemistry using rabbit polyclonal antibody that recognizes human and rat TDP-43 (ALS10, ProteinTech) showed increased TDP-43 protein levels and cytosolic accumulation 2 weeks post-injection (Fig. 2B) compared with the LacZ-injected contralateral ( Fig. 2A) hemisphere. To ascertain specificity of gene expression, we used human-specific (hTDP-43) mouse monoclonal antibody that recognizes amino acids 1-261 (Abcam) and observed positive human TDP-43 staining within a 4-mm radius in 38% (by stereology, n ϭ 8) of cortical neurons (Fig. 2D) compared with a LacZ-injected (Fig. 2C) hemisphere. Further examination of cervical spinal cord revealed a 13% increase in immunoreactivity to hTDP-43 (Fig. 2F) and increased reactivity to TDP-43 antibody (Fig. 2G) in preganglionic inter-neurons, which were morphologically identified in the contralateral side of TDP-43-injected motor cortex (Fig. 2E) compared with the contralateral spinal cord injected with LacZ (Fig. 2, H and I), suggesting that hTDP-43 expression in the motor cortex leads to increased protein levels in the contralateral spinal cord. Furthermore, stereological counting revealed a 46% (by stereology, n ϭ 8) increase in the levels of hTDP-43 (Fig. 2J) and increased immunoreactivity to TDP-43 antibody (Fig. 2K) in the dorsocortical spinal tract (DCST) of cervical spinal cord contralateral to cortical TDP-43 expression compared with the LacZ-injected side (Fig. 2, L and M). Toluidine blue staining and quantification by a blind investigator of centric axons within 10 random fields of each slide showed an increased number (18%, n ϭ 8) of axons (Fig. 2N, arrows) in enlarged circles, suggesting axonal degeneration compared with the contralateral DCST (Fig. 2O). Some centric axons were detected in all treatments.
Parkin Promotes Lys-48-and Lys-63-linked Ubiquitin to TDP-43-To demonstrate whether Parkin mediates TDP-43 ubiquitination, we performed immunoprecipitation to show ubiquitinated TDP-43 in the presence of Parkin expression. Western blot analysis of the input showed that increased exogenous Parkin (Fig. 4A, 1st blot, n ϭ 8, p Ͻ 0.05, 42%) in the rat motor cortex increases the levels of ubiquitinated proteins (Fig.  4A, 2nd blot). Densitometry analysis of TDP-43 blots (Fig. 4A,  3rd blot) showed a significant increase (48%, n ϭ 8) in TDP-43 levels in brains injected with lentiviral TDP-43 (consistent with our previous work (52)) compared with LacZ-or Parkin-injected brains. However, co-injection of TDP-43 and Parkin did not result in any significant changes in TDP-43 levels (p Ͻ 0.05, n ϭ 8), suggesting that Parkin mediates TPD-43 ubiquitination, which may not lead to protein degradation. We also used a nonfunctional Parkin mutant (T240R, threonine to arginine mutation), which was co-expressed with TDP-43 (Fig. 4A, top  blot) and detected no changes in ubiquitinated proteins (Fig.  4A, 2nd blot) or TDP-43 levels (Fig. 4A, 3rd blot). We then immunoprecipitated TDP-43 and probed with ubiquitin (Fig.  4A, 4th blot) to ascertain that high molecular weight species are ubiquitinated TDP-43 proteins and not some protein aggregates. An increase in protein smear was observed when TDP-43 was co-injected with Parkin, compared with TDP-43, Parkin, or LacZ alone, suggesting increased TDP-43 ubiquitination in the presence of wild type Parkin. However, no differences were observed in the levels of ubiquitinated proteins (Fig. 4A, 4th  blot) when TDP-43 was immunoprecipitated with or without expression of T240R mutant Parkin, suggesting that functional Parkin mediates TDP-43 ubiquitination.

JOURNAL OF BIOLOGICAL CHEMISTRY 4109
which contains all seven lysine residues. To determine whether Parkin activity is altered in the presence of TDP-43, we added both Parkin and TDP-43 to the enzyme mixture. As expected, no activity was detected with the lysine null ubiquitin (Lys-0), but Parkin activity was significantly increased compared with Parkin alone (Fig. 4C, p Ͻ 0.05, n ϭ 8), with Lys-48 (154%), and Lys-63 (156%) ubiquitin, indicating that Parkin activity is even higher in the presence of a substrate. Parkin also showed a significantly higher level of activity with wild type ubiquitin in the presence of TDP-43 (279%) compared with Parkin alone.
Parkin Forms a Multiprotein Complex with HDAC6 to Mediate TDP-43 Translocation from Nucleus to Cytosol-Lack of degradation of ubiquitinated TDP-43 and cytosolic accumulation of Parkin, TDP-43, and ubiquitin in gene transfer animals led us to examine possible mechanisms to translocate TDP-43 to the cytosol. It was reported that histone deacetylase 6 (HDAC6) directly binds to Parkin and mediates its transport in response to proteasome inhibition (53). Western blot analysis showed a significant increase (41%, p Ͻ 0.05) in HDAC6 levels when TDP-43 was expressed compared with LacZ-or Parkininjected animals (Fig. 4, G and H, 1st blot, p Ͻ 0.05, n ϭ 8). However, further increases in HDAC6 levels (Fig. 4, G and H,  112%, p Ͻ 0.05) were detected when Parkin was co-expressed with TDP-43, suggesting a possible interaction between these proteins. Examination of molecular markers of autophagy showed a significant increase in p62 (28%, p Ͻ 0.05) when Parkin was co-expressed with TDP43 (Fig. 4, G and H, 2nd blot) compared with all other treatments, suggesting accumulation of ubiquitinated proteins. We did not see any changes in other markers of autophagy (LC3, beclin, Atgs) or appearance of autophagic vacuoles by EM (data not shown). We immunoprecipitated human TDP-43 from transgenic mice and verified TDP-43 at 46 kDa using hTDP-43 antibody (Fig. 5A, 1st and  2nd blots). Stripping and re-probing with Parkin antibody showed a slightly higher band around 50 kDa, suggesting the presence of Parkin protein (Fig. 5A, 3rd blot). Further stripping and probing with HDAC6 antibody (Fig. 5A, 4th blot) showed a higher molecular mass band around 120 kDa, indicating a multiprotein complex between Parkin, TDP43, and HDAC6. We then performed a reserve experiment via Parkin immunoprecipitation and verification of human TDP-43 presence (Fig. 5B,  1st and 2nd blot). Stripping and probing with Parkin antibody showed a Parkin band in both transgenic and nontransgenic control mice (Fig. 5B, 3rd blot), indicating that Parkin was successfully immunoprecipitated. We also detected a higher molecular weight band representative of HDAC6 (Fig. 5B, 4th blot) in transgenic but not control mice, further suggesting multiprotein complex formation between TDP43, Parkin, and HDAC6.
To ascertain that both Parkin and HDAC6 are required for TDP-43 translocation, we expressed GFP-tagged TDP-43 in M17 neuroblastoma cells in the presence of wild type or lossof-function mutant (T240R) Parkin and treated with 5 M selective HDAC6 inhibitor for 24 h. GFP expression was predominantly observed within DAPI-stained nuclei in live M17 cells (Fig. 5C, inset is higher magnification); however, Parkin co-expression led to significant GFP fluorescence within the cytoplasm (Fig. 5, D and E) and neuronal processes (Fig. 5D, inset shows higher magnification of GFP fluorescence). Treatment with the HDAC6 inhibitor, tubacin, did not lead to GFP fluorescence in the cytosol in the presence (Fig. 5F) or absence (Fig. 5G) of Parkin. Loss of Parkin E3 ubiquitin ligase function (T240R) did not lead to TDP-43 accumulation in the cytosol (Fig. 5H), suggesting that the E3 ubiquitin ligase function of Parkin and HDAC6 activity are required to facilitate TDP-43 accumulation within the cytosol.
It was reported that TDP-43 depletion results in down-regulation of Park2 mRNA in stem cell-derived human neurons and in motor neurons containing TDP-43 inclusions in sporadic ALS (30). To verify whether TDP-43 expression increases Parkin mRNA levels, we performed qRT-PCR in samples isolated from rat cortex, human M17 cells, and TDP43-Tg mice. Park2 mRNA levels in M17 cells expressing Parkin was significantly higher (Fig. 5, I and J, 55%, p Ͻ 0.05, n ϭ 4) than LacZ but similar to TDP-43-injected brains (61%, p Ͻ 0.05). Parkin coexpression with TDP-43 showed significantly higher levels of park2 mRNA (Fig. 5J, 74%, p Ͻ 0.05, n ϭ 4) compared with Parkin alone. Similarly, Park2 mRNA levels in rat brains expressing Parkin was significantly higher (Fig. 5, K and L, 41%, p Ͻ 0.05, n ϭ 4) than LacZ animals, as well as TDP-43-injected brains (21%, p Ͻ 0.05). However, Parkin co-expression with TDP-43 showed significantly higher levels of park2 mRNA (Fig.  5J, 84%, p Ͻ 0.05, n ϭ 4) compared with all other treatments. The variation in the data may be due to differences in cDNA transfection in cell culture or efficiency of lentiviral infection in gene transfer animal models. Therefore, we compared park2 mRNA levels between TDP43-Tg and nontransgenic control littermates. A significant increase (Fig. 5, M and N, 114%, n ϭ 4, p Ͻ 0.05) in park2 mRNA was observed in TDP43-Tg brains injected compared with C57BL/6 controls, suggesting that Parkin is a transcriptional target for TDP-43 (30).

DISCUSSION
These studies contribute to our knowledge of the poorly understood physiological role of TDP-43 in health and disease. Accumulation of ubiquitinated TDP-43 within the nucleus was shown to cause profound pathological effects in the absence of cytosolic protein (44). Here, we show an important relationship between Parkin E3 ubiquitin ligase and TDP-43. TDP-43 may contribute to increased expression of Parkin mRNA and protein, consistent with other reports that TDP-43 regulates Parkin expression (29,30). Parkin expression was increased in transgenic TDP-43 mice harboring the A315T mutation as well as M17 neuroblastoma cells, where Parkin protein is difficult to detect, indicating that nuclear TDP-43 may control Parkin expression. The increase in soluble Parkin level failed to protect against TDP-43 pathogenicity in TDP43-Tg mice, perhaps due to Parkin's lack of ability to reduce TDP-43 levels via autophagy or the proteasome. Our data show that Parkin forms a complex with HDAC6 to translocate TDP-43, but it is unable to induce autophagic clearance of this protein (54). HDAC6 directly binds to Parkin and mediates its transport in response to proteasome inhibition, whereas Parkin transport is reversible when proteasome activity is restored (53). Therefore, lack of autophagic clearance and proteasomal inhibition may have led to Parkin accumulation and decreased solubility. We previously demonstrated that wild type and not mutant loss-of-function Parkin increases proteasome activity (40 -42), leading to degradation of ubiquitinated proteins, whereas amyloid protein stress and aging alter Parkin solubility (45). Decreased Parkin solubility is associated with alteration of its activity via increased phosphorylation by several kinase activities, including casein kinase 1, protein kinase A, protein kinase C (55), cyclin-dependent kinase 5 (56,57), c-Abl (58,59), and potentially PTEN-induced putative kinase 1 (PINK1) (60,61). It is important to realize that Parkin functions as part of a number of multiprotein complexes, including PINK1 (35)(36)(37), the Skp1-Cullin-Fbox (SCF)-like complex to facilitate proteasomal degradation (38), and the chaperone Hsp70, and the U-Box protein C terminus of Hsc70-interacting protein (39). The next phase of this work will be to inject TDP-43 into Parkin Ϫ/Ϫ mice and determine the role of endogenous Parkin.
It has been reported that accumulation of amyloidogenic proteins can alter Parkin solubility, perhaps leading to changes in its activity (45,58,62). It is unlikely that the appearance of Parkin protein is due to proteasome inhibition, leading to lack of degradation as is the case with SIAH2, because the significant reversal of proteasome activity in the presence of exogenous Parkin led to the disappearance of SIAH2 but not Parkin. TDP-43 expression leads to proteasome inhibition and accumulation of ubiquitinated proteins, including the rapidly degrading SIAH2 due to auto-ubiquitination and proteasomal clearance (46 -51). Conversely, activation of Parkin E3 ligase function leads to ubiquitination of TDP-43 via Lys-48-and Lys-63-linked ubiquitination, indicating Parkin involvement in TDP-43 metabolism, including its cytosolic accumulation. Parkin led to the cytosolic accumulation of ubiquitinated TDP-43, without any evidence of protein degradation or clearance, suggesting that Parkin-mediated TDP-43 ubiquitination perhaps facilitates nuclear translocation of TDP-43 to the cytosol. Parkin mediated Lys-48 and Lys-63 as well as wild type ubiquitin (which can include any of the five other ubiquitin linkages Lys-6, Lys-11, Lys-27, Lys-29, and Lys-33), indicating several possibilities about the functional role of these ubiquitin linkages (63). Mono-ubiquitination was reported to regulate DNA repair and receptor endocytosis, although Lys-48-linked ubiquitin chains are tags for proteasomal degradation (64). Lys-63linked ubiquitination was also reported to facilitate DNA repair and target proteins for degradation via interaction between ubiquitinated proteins and p62 (65). However, in a study using autophagy-deficient mice, all ubiquitin-ubiquitin linkages accumulated, in contradiction with the hypothesis that any particular ubiquitin linkage serves as a specific autophagy signal (66). Parkin-mediated TDP-43 ubiquitination led to p62 and cytosolic TDP-43 accumulation. It is therefore possible that the cell contains factors that prevent proteins linked to Lys-63ubiquitin from degradation (67). For example, Parkin-mediated Lys-63-linked polyubiquitination was reported to link misfolded proteins to the dynein motor complex via the adaptor protein HDAC6, promoting sequestration of misfolded proteins into aggresomes and subsequent clearance by autophagy (54). Other groups showed that HDAC6 directly binds to Parkin and mediates its transport (53). Others reported that Lys-48-and Lys-63-linked polyubiquitination, as well as monoubiquitination, contributes to inclusion formation; however, Lys-63-linked polyubiquitin enhances inclusion formation and selectively facilitates clearance via autophagy (68). Our data showed that Parkin forms a multiprotein complex with HDAC6 and TDP-43 without induction of autophagy or proteasomal degradation, suggesting that other factors may determine which type of polyubiquitin chains are specific labels for protein degradation. The multiprotein complex consisting of TDP-43, HDAC6, and Parkin may lead to translocation of nuclear TDP-43 into the cytosol via binding with dynein motors.
Increased detection of TDP-43 in the spinal cord following protein expression in the rat motor cortex is intriguing and may have relevance to human disease, mainly ALS, where affected cortical motor neurons are associated with degenerating spinal tracts (69 -72). It is unclear whether TDP-43 protein or mRNA is axonally transported in descending motor neuron fibers or whether increased TDP-43 protein levels in the cortex signals expression in the spinal cord causing axonal degeneration. Detection of TDP-43 in spinal cord and DCST alludes to the susceptibility of this subset of neurons within the descending spinal motor tracts to TDP-43 pathology. More importantly, these studies determine the early effects of TDP-43 accumulation in vivo. It is necessary to mention that no differences were observed in strength or RotaRod behavioral tests 2 weeks postinjection in gene transfer animals, perhaps due to unilateral gene expression (TDP-43 was injected into right hemisphere), but transgenic mice were symptomatic and died between 21 and 30 days. Cytosolic inclusions composed of misfolded accumulating proteins, including ubiquitinated TDP-43, are found in several neurodegenerative diseases (29,73,74), whereas TDP-43 knockdown impairs neuronal growth (75) perhaps due to involvement of TDP-43 in RNA/DNA processing of a large number of genes (29). Recent reports show that neurons bearing cytosolic TDP-43 aggregates have decreased Parkin levels (30), suggesting loss of nuclear TDP-43 function to regulate Parkin levels. Conversely, the A315T transgenic mice show an increased Parkin level with decreased solubility, suggesting altered function of de novo synthesized Parkin to facilitate cytosolic accumulation of ubiquitinated TDP-43. Parkin inactivation is described in the nigrostriatal region of patients with sporadic PD (58), and protein aggregates were previously shown to alter Parkin solubility (76). We previously showed that Parkin co-localizes with intraneuronal A␤(1-42) and has decreased solubility in AD (45). Therefore, TDP-43 stress may alter Parkin solubility and perhaps function. The loss of Parkin function is supported in early onset juvenile PD, where no cytosolic Lewy body inclusions are detected (77), suggesting that functional Parkin mediates cytosolic sequestration of misfolded proteins. Parkin may act alone to increase TDP-43 ubiquitination or collaborate with other ubiquitin ligases to enhance protein ubiquitination. Nuclear TDP-43 regulates a large amount of gene expression and is implicated in many steps of RNA expression and transport, including Parkin (29,78); therefore, translocation from the nucleus to the cytosol, perhaps via ubiquitination when Parkin is expressed, may be a sequestration strategy.
In conclusion, whether TDP-43 pathology involves loss of nuclear or gain of cytosolic function, it is important to have a functional proteasome to mediate protein clearance and prevent inclusion formation. The type of Lys-linked polyubiquitin chain seems to depend on many factors, perhaps autophagy and the proteasome, to determine the fate of ubiquitinated TDP-43. The activation of Parkin E3 ubiquitin ligase activity and TDP-43 ubiquitination result in cytosolic accumulation, despite the partial reversal of proteasome activity, leading to complex formation with HDAC6 and sequestration of TDP-43. Cytosolic accumulation of TDP-43 may be a coping mechanism to alleviate or temporarily delay the effects of TDP-43 on aberrant mRNA transcription.