Abstract
Amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration (FTLD-TDP) are two neurodegenerative disorders characterized by the accumulation of TDP-43. TDP-43 is proteolitically cleaved to generate two major C-terminal fragments of 35 and 25 kDa. The latter, known as TDP-25, is a consistent feature of FTLD-TDP and ALS; however, little is known about its role in disease pathogenesis. We have previously developed transgenic mice overexpressing low levels of TDP-25 (TgTDP-25+/0), which at 6 months of age show mild cognitive impairments and no motor deficits. To better understand the role of TDP-25 in the pathogenesis of ALS and FTLD-TDP, we generated TDP-25 homozygous mice (TgTDP-25+/+), thereby further increasing TDP-25 expression. We found a gene-dosage effect on cognitive and motor function at 15 months of age, as the TgTDP-25+/+ showed more severe spatial and working memory deficits as well as worse motor performance than TgTDP-25+/0 mice. These behavioral deficits were associated with increased soluble levels of TDP-25 in the nucleus and cytosol. Notably, high TDP-25 levels were also linked to reduced autophagy induction and proteasome function, two events that have been associated with both ALS and FTLD-TDP. In summary, we present strong in vivo evidence that high levels of TDP-25 are sufficient to cause behavioral deficits and reduce function of two of the major protein turnover systems: autophagy and proteasome. These mice represent a new tool to study the role of TDP-25 in the pathogenesis of ALS and FTLD-TDP.
Introduction
The accumulation of the transactive response DNA-binding protein 43 (TDP-43) is a common feature of several neurodegenerative disorders often referred to as TDP-43 proteinopathies. To this end, TDP-43 inclusions are present in Alzheimer's disease, frontotemporal lobar degeneration (FTLD), amyotrophic lateral sclerosis (ALS), Parkinson's disease and dementia with Lewy bodies (1,2). TDP-43 represents the major neuropathological signature of a subset of FTLD cases, known as FTLD-TDP, and in ALS (3). Genetic studies have shown conclusive evidence that TDP-43 plays a direct role in neurodegeneration. Indeed, several mutations in the gene encoding TDP-43 are linked to ALS, and more rare mutations have also being linked to motoneuron disease and FTLD (4,5).
TDP-43 is a DNA/RNA binding protein (6). While it is largely a nuclear protein, TDP-43 can be shuttled to the cytoplasm (7,8). Its primary function in the nucleus is linked to the regulation of RNA splicing and to the modulation of microRNA biogenesis (6,9). Pathological TDP-43 is aggregated, hyperphosphorylated, ubiquitinated and cleaved to generate C-terminal fragments of 35 and 25 kDa (3). The latter, known as TDP-25, is an invariable feature of FTLD-TDP and accumulates only in affected brain regions (10). TDP-43 inclusions are mainly found in the cytoplasm and are associated with clearing of nuclear TDP-43 (3). Despite this knowledge, little is known about the mechanisms by which TDP-43 leads to neurodegeneration and whether TDP-25 plays a role in this process.
The autophagic system is a conserved intracellular system designed for the degradation of long-lived proteins and organelles in lysosomes (11,12). During autophagy induction, an isolation membrane is generated surrounding cytosolic components, forming an autophagosome, which will eventually fuse with lysosomes for protein/organelle degradation (11). Several autophagy-related proteins (Atg) are involved in a series of ubiquitin-like reactions during the different stages of the autophagosome formation (13,14). Cumulative evidence suggests that an age-dependent decrease in the function of the autophagy/lysosome system may account for the accumulation of abnormal proteins during aging and in neurodegenerative disorders (12,15). Indeed, autophagy dysfunction has been linked to several neurodegenerative disorders, including TDP-43 proteinopathies (16–18). The proteasome is another intracellular system involved in protein turnover. The proteolytic core of the proteasome contains three proteolytic sites known as caspase-like, trypsin-like and chymotrypsin-like (19). Proteasome dysfunction has also been linked to TDP-43 proteinopathies (20,21). Indeed, we and others have shown that TDP-43 can be degraded by the proteasome or by the autophagic system (17,22–27).
Several animal models of TDP-43 proteinopathies have been generated. These animals develop a wide range of phenotypes (28). Overall, most lines show that overexpression of wild-type or mutant TDP-43 leads to neurodegeneration and behavioral deficits, which often occurs independently of cytosolic TDP-43 inclusions (28). This dissociation indicates that TDP-43 inclusions are not necessary to obtain motor and cognitive deficits, and raises the possibility that the inclusions may be a secondary event in the disease pathogenesis (29). We previously developed a transgenic mouse by overexpressing the 25 kDa C-terminal fragment of TDP-43 (herein referred to as TgTDP-25). At 6 months of age, these mice show misprocessing of endogenous TDP-43 and mild cognitive deficits in the absence of TDP-43 inclusions (30). To further study the role of TDP-25 in the pathogenesis of TDP-43 proteinopathies and determine whether aging plays a role in this process, we have generated aged hemizygous and homozygous TgTDP-25 mice. We report that overexpression of TDP-25 in old mice causes motor and cognitive deficits in a dose-dependent manner.
Results
We have previously described the development of transgenic mice expressing TDP-25 (TgTDP-25+/0) and characterized them behaviorally and pathologically at 2 and 6 months of age. Six-month-old TgTDP-25+/0 showed mild cognitive deficits but not motor deficits (30). To determine whether a further increase in TDP-25 levels would result in the development of more robust behavioral deficits, we bred TgTDP-25+/0 mice to each other and generated TDP-25 homozygous mice.
To assess cognitive function in mice expressing TDP-25, we first tested 15-month-old wild-type (WT), TgTDP-25 hemizygous (TgTDP-25+/0) and homozygous (TgTDP-25+/+) mice (n = 12/genotype) in the radial arm water maze (RAWM), a behavioral test that allows simultaneous testing of spatial and working memory (31). Mice were placed in an eight-arm pool and received 15 training trials per day for 2 days to find an escape platform located at the end of one arm. During the 15 trials on Day 1, the platform was alternated above the water (visible) and below the water to facilitate learning. On Day 2, the platform was kept submerged throughout the 15 trials. The total number of entries in arms without platform on Day 2 was considered spatial memory errors, while the number of re-entries in the same arm during the same trial was considered working memory errors. A two-way repeated-measures analysis of variance (ANOVA) for total errors revealed a significant effect for genotype [P < 0.0001; F(2, 33) = 33.31] and blocks [average of three trials; P < 0.0001; F(4, 33) = 7.46; Fig. 1A]. The effect of block indicated that one or more genotypes made a significantly different number of errors across the five blocks. The genotype effects indicated that across the five blocks, one or more genotypes performed significantly different from each other. A post hoc test with Bonferroni correction showed that compared with WT mice, Tg-TDP-25+/0 mice were significantly different at block 2 (P < 0.05; t = 2.61) and 4 (P < 0.05; t = 2.83) and TgTDP-25+/+ were significantly different at blocks 2 (P < 0.001; t = 4.61), 3 (P < 0.001; t = 3.94), 4 (P < 0.001; t = 4.05) and 5 (P < 0.001; t = 4.16). Further, TgTDP-25+/+ mice performed worse than TgTDP-25+/0 at block 3 (P < 0.05; t = 2.72) and 5 (P < 0.01; t = 3.39). Overall, these data indicate that spatial memory is significantly impaired in TgTDP-25+/+ mice and that such an impairment is less pronounced in TgTDP-25+/0 mice (Fig. 1A).
TDP-25 overexpression causes cognitive deficits in a dose-dependent manner. (A) Fifteen-month-old WT, TgTDP-25 hemizygous (TgTDP-25+/0) and homozygous (TgTDP-25+/+) mice (n = 12/genotype) were tested in the RAWM. The graph shows the total number of errors across blocks, which are an indication of spatial learning and memory. A block represents the average of three trials. TgTDP-25+/+ mice made the most errors, while WT mice made the fewest errors. TgTDP-25+/0 mice performed better than TgTDP-25+/+ but not as well as WT mice. (B) The graph shows the number of re-entries in the same arm during the same trial, considered as working memory errors. TgTDP-25+/+ mice performed significantly worse compared with the other two groups. (C) A different cohort of mice (WT, n = 16; TgTDP-25+/0, n = 12; TgTDP-25+/+, n = 14) were tested in the MWM. The graph shows the distance traveled to find the hidden platform across five consecutive days. Each day represents the average of four training trials. Starting on Day 2, WT mice performed significantly better compared with the two transgenic groups. On Days 4 and 5, TgTDP-25+/0 mice performed significantly better compared with TgTDP-25+/+. (D–F) The graphs show the number of platform location crosses, the time in the target quadrant, and the swim speed of mice during a single 60 s probe trial. WT mice showed a significantly higher number of crosses and time in the target quadrant compared with the other two transgenic groups. In contrast, swim speed was not statistically significant among the three groups. (G) Mice were trained in the rotarod over 3 days (six training trials per day). During the first day of training, all three groups were significantly different from each other. On Days 2 and 3, WT mice performed significantly better than the other two transgenic groups. (H) Twenty-four hours after the last training, mice were tested in the accelerating rotarod (six trials). All three groups were significantly different from each other. Data are presented as means ± SEM and were analyzed by two-way (A–C) or one-way (D–H) ANOVA, followed by Bonferroni's post hoc tests. Asterisk indicates a significant difference between WT and TgTDP-25+/+ mice; hash symbol indicates a significant difference between WT and TgTDP-25+/0 mice; double dagger indicates a significant difference between TgTDP-25+/0 and TgTDP-25+/+ mice.
Using a two-way repeated-measures ANOVA, we also analyzed the number of re-entries in the same arm during each single trial, which is an indication of working memory. We found a significant genotype effect [P < 0.0001; F(2, 33) = 16.01] but not block effect (P = 0.071; Fig. 1B). These data indicate that the number of re-entry remains constant across the five blocks but is different among genotypes. A post hoc test with Bonferroni correction showed that while TgTDP-25+/0 mice performed similarly to WT mice, TgTDP-25+/+ mice were significantly worse than WT mice at block 2 (P < 0.01; t = 3.67), block 3 (P < 0.01; t = 3.57), block 4 (P < 0.05; t = 2.64) and block 5 (P < 0.05; t = 2.77). TgTDP-25+/+ mice were also worse than TgTDP-25+/0 mice at blocks 1 and 2 (P < 0.05 and t = 2.64 for both blocks).
We next tested a different cohort of mice in the spatial version of the Morris water maze (MWM). Mice (WT, n = 16; TgTDP-25+/0, n = 12; TgTDP-25+/+, n = 14) received four training trials per day for five consecutive days to learn the location of a hidden platform using extra maze cues. The distance traveled to find the platform across the training trials is an indicator of mouse learning, with less distance interpreted as better performance (32). Using a two-way repeated-measures ANOVA, we found a significant effect for days [P < 0.0001; F(4, 195) = 66.03] and genotype [P < 0.0001; F(2, 195) = 49.61; Fig. 1C]. The effect of day indicated that mice learned the task across days whereas the effect of genotype indicated that one or more genotypes had a different pace of learning from each other. A post hoc test with Bonferroni correction showed that TgTDP-25+/0 mice performed worse than WT mice at Day 2 (P < 0.01; t = 3.21), Day 3 (P < 0.01; t = 3.75), Day 4 (P < 0.01, t = 3.26), while TgTDP-25+/+ mice performed worse than WT mice at Day 2 (P < 0.01; t = 3.30), Day 3 (P < 0.001; t = 5.86), Day 4 (P < 0.001; t = 6.47) and Day 5 (P < 0.001; t = 6.70). In addition, TgTDP-25+/+ mice performed significantly worse than TgTDP-25+/0 mice at Day 4 (P < 0.05; t = 2.85) and Day 5 (P < 0.001; t = 3.99). Twenty-four hours after the last training trial, mice were tested in the maze, without the platform, for a single 60 s trial, during which we measured the number of platform location crosses as well as the time mice spent in the target quadrant. One-way ANOVA indicated that the number of platform location crosses was significantly different among the three genotypes [P = 0.006; F(2, 39) = 5.84; Fig. 1D]. Post hoc Bonferroni's multiple comparison test indicated that the TgTDP-25+/+ mice were significantly different than WT mice (P < 0.05; t = 3.41), whereas no differences were found for the other pairwise comparisons. We also found that the time spent in the target quadrant was significantly different among the three groups [P < 0.0001; F(2, 39) = 13.98; Fig. 1E). Bonferroni's multiple comparison test showed that both TgTDP-25+/+ and TgTDP-25+/0 mice were significantly different than WT mice (P < 0.05; t = 2.56 and P < 0.05; t = 5.28, respectively). In contrast, the swim speed was not statistically significant among the three genotypes (Fig. 1F), indicating that the genotype effects on learning and memory were independent of swimming abilities. In summary, using two complementary tasks and two different cohorts of mice, we showed that TgTDP-25+/+ mice performed worse than TgTDP-25+/0 mice, which perform worse than WT mice, indicating a gene-dosage effect of TDP-25 on cognition.
ALS and FTLD-TDP have overlapping clinical and neuropathological features and both disorders present various degrees of motor dysfunction (33). To assess motor function in mice expressing TDP-25, we tested mice in the rotarod, which is routinely used as an indicator of motor coordination. For this test, we used the same mice (genotype and numbers) used for the MWM. During the learning phase, mice received six training trials per day during three consecutive days at a constant speed of 15 rpm. A two-way repeated-measures ANOVA indicated a significant day [P < 0.0001; F(2, 39) = 61.18] and genotype effect [P < 0.0001; F(2, 39) = 81.40; Fig. 1G]. A Bonferroni post hoc correction showed that both TgTDP-25+/+ and TgTDP-25+/0 mice performed significantly different than WT mice on Days 1–3 (P < 0.001 for all pairwise comparisons). In addition TgTDP-25+/+ mice performed worse than TgTDP-25+/0 on Day 1 (P < 0.05; t = 2.95). Twenty-four hours after the last training trail, mice received probe trials in the accelerated rotarod. In this case, the rod increased its speed by 1 rpm/s until the mice fell off. One-way ANOVA indicated that there was a significant difference among the three groups [P < 0.0001; F(2, 38) = 30.11; Fig. 1H]. Post hoc Bonferroni's multiple comparison test indicated that the TgTDP-25+/+ mice were significantly different than WT (P < 0.0001; t = 7.73) and TgTDP-25+/0 mice (P < 0.05; t = 2.613). Further, TgTDP-25+/0 mice were also significantly different compared with WT mice (P < 0.001; t = 4.73). Consistent with the cognitive data, these results show a dose-dependent effect of TDP-25 on motor function.
Pathological TDP-43 is more prone to aggregation (3). To assess TDP-43 and TDP-25 solubility, we extracted proteins from brains of 15-month-old WT, TgTDP-25+/0 and TgTDP-25+/+ mice (n = 6/genotype) using buffers of increasing stringency (see Materials and Methods). In the low-salt fraction, which contains proteins with high solubility, we found that full-length TDP-43 levels (gray arrow in Fig. 2A) were significantly different among the three groups as indicated by one-way ANOVA [P < 0.0001; F(2, 15) = 23.35; Fig. 2A and B]. A Bonferroni's multiple comparison test revealed that the levels of TDP-43 in TgTDP-25+/+ mice were significantly lower than WT (P < 0.05; t = 5.33) and TgTDP-25+/0 mice (P < 0.05; t = 6.37). TDP-25 levels (black arrow in Fig. 2A) were also significantly different among the three groups [P < 0.001; F(2, 15) = 168.1; Fig. 2A and C]. A Bonferroni's multiple comparison test revealed that the levels of TDP-25 in WT mice were significantly lower than TgTDP-25+/0 mice (P < 0.05; t = 9.50) and TgTDP-25+/+ mice (P < 0.05; t = 18.33). TDP-25 levels were also significantly different between TgTDP-25+/0 and TgTDP-25+/+ mice (P < 0.05; t = 8.83). The decrease in soluble TDP-43 levels was surprising and further studies are needed to assess the mechanisms behind this change. In contrast, full-length TDP-43 levels (gray arrow in Fig. 2D) were not statistically significant among the three groups in the high salt fraction (Fig. 2D and E). Consistent with the low-salt fraction and the presence of the transgene, TDP-25 levels (black arrow in Fig. 2D) were significantly different among the three genotypes [P < 0.0001; F(2, 15) = 35.34; Fig. 2D and F]. Bonferroni's multiple comparison test indicated that the levels of TDP-25 were higher in TgTDP-25+/+ mice compared with TgTDP-25+/0 (P < 0.05; t = 4.44) and WT mice (P < 0.05; t = 8.40). Further, TgTDP-25+/0 mice had higher levels of TDP-25 in the high salt fraction than WT mice (P < 0.05; t = 3.96). We then used sarkosyl to extract proteins that were not soluble in low salt or in high salt buffer. While we found that TDP-43 levels (gray arrow in Fig. 2G) in the sarkosyl fraction were not different among the three genotypes (Fig. 2G and H), TDP-25 levels (black arrow in Fig. 2G) were significantly different [P = 0.0002; F(2, 15) = 15.52; Fig. 2G and I]. Bonferroni's multiple comparison test indicated that the levels of TDP-25 were higher in TgTDP-25+/+ mice compared with TgTDP-25+/0 (P < 0.05; t = 5.27) and WT mice (P < 0.05; t = 4.19). No statistically significant difference was found between WT and TDP-25+/0 mice. Taken together, these data indicate that the transgene is present in different aggregation forms, similarly to FTLD-TDP and ALS cases, in which TDP-25 is present in fractions with different solubility.
Accumulation of soluble and insoluble TDP-25 in TgTDP-25 transgenic mice. (A) Representative western blots of proteins extracted in a low-salt buffer from the brains of WT, TgTDP-25+/0, and TgTDP-25+/+ mice (n = 6/genotype). Shorter and longer exposures of blots probed with an anti-TDP-43 antibody are shown. (B) Quantitative analysis of full-length TDP-43 levels indicated a significant decrease in TgTDP-25+/+ mice. (C) Quantitative analysis of TDP-25 levels indicated that all three groups were significantly different from each other. (D) Representative western blots of proteins extracted in a high-salt buffer. Shorter and longer exposures of blots probed with an anti-TDP-43 antibody are shown. (E) Quantitative analysis of full-length TDP-43 levels indicated no significant changes among the three groups (F) Quantitative analysis of TDP-25 levels indicated that all three groups were significantly different from each other. (G) Representative western blots of proteins extracted in sarkosyl buffer. Shorter and longer exposures of blots probed with an anti-TDP-43 antibody are shown. (H) Quantitative analysis of full-length TDP-43 levels indicated no significant changes among the three groups. (I) Quantitative analysis of TDP-25 levels indicated that TgTDP-25+/+ mice were significantly different than the other three groups. Quantitative analyses were obtained by normalizing the protein of interest to β-Actin, used as loading control. Data are presented as means ± SEM and were analyzed by one-way ANOVA, followed by Bonferroni's post hoc tests. Asterisk indicates a significant difference between WT and TgTDP-25+/+ mice; hash symbol indicates a significant difference between WT and TgTDP-25+/0 mice; double dagger indicates a significant difference between TgTDP-25+/0 and TgTDP-25+/+ mice. Gray arrows point to full-length TDP-43. Black arrows point to TDP-25.
Pathological TDP-43 is mislocalized from its nuclear location to the cytosol, where it accumulates to form ubiquitinated inclusions (29). To assess the subcellular distribution of TDP-43 and TDP-25 in TgTDP-25 transgenic mice, we extracted proteins from the nuclear and cytosolic fractions of 15-month-old WT, TgTDP-25+/0 and TgTDP-25+/+ mice (n = 6/genotype). Although we found a clear trend for lower TDP-43 levels in the cytosolic fraction of Tg-TDP-25+/+ mice (gray arrow in Fig. 3A), these changes did not reach statistical significance [P = 0.08; F(2, 15) = 2.97; Fig. 3A and B]. In contrast, we found that cytosolic TDP-25 levels (black arrow in Fig. 3A) were significantly different among the three genotypes [P < 0.0001; F(2, 15) = 37.08; Fig. 3A and C]. A Bonferroni's multiple comparison test indicated that cytosolic TDP-25 levels were higher in TgTDP-25+/+ mice compared with TgTDP-25+/0 (P < 0.001; t = 5.68) and WT mice (P < 0.001; t = 8.44). TDP-25 levels were also statistically significant between WT and Tg-TDP-25+/0 mice (P < 0.05; t = 2.86). As in the cytosolic fraction, TDP-43 levels were similar among the three groups in the nuclear fraction (Fig. 3A and D). We found, however, that TDP-25 levels were different in the nuclear fraction among the three genotypes [P < 0.0001; F(2, 15) = 55.99; Fig 3A and E]. Specifically, Tg-TDP-25+/+ mice had higher levels of nuclear TDP-25 than Tg-TDP-25+/0 (P < 0.05; t = 5.10) and WT mice (P < 0.05; t = 10.58). Further, nuclear TDP-25 levels were higher in Tg-TDP-25+/0 mice compared with WT mice (P < 0.05; t = 5.48). Together, these data indicate that the transgene is equally distributed in the nucleus and cytosol. To assess whether TgTDP-25 mice accumulate cytosolic TDP-43 inclusions, we stained sections from TgTDP-25+/0 and TgTDP-25+/+ mice. We found that the vast majority of TDP-43 immunoreactivity was nuclear (Fig. 3F). Rare cytosolic TDP-43 deposits were also evident in TgTDP-25+/+ mice (arrow in Fig. 3F).
Increased cytosolic and nuclear accumulation of TDP-25 in TgTDP-25 mice. (A) Representative western blots of brain proteins extracted from the cytosolic and nuclear fraction of 15-month-old WT, TgTDP-25+/0 and TgTDP-25+/+ mice (n = 6/genotype). Blots were probed with an anti-TDP-43 antibody. (B and C) Quantitative analyses of the blots for the cytosolic fraction revealed no change in full-length TDP-43 levels. In contrast, TDP-25 levels were significantly higher in TgTDP-25+/+ mice compared with the other two groups. TDP-25 were also significantly higher in TgTDP-25+/0 mice compared with WT mice. (D and E) Quantitative analyses of the blots for the nuclear fraction indicated that TDP-43 levels were not statistically significant among the three groups. In contrast, TDP-25 levels were statistically significant for all thee pairwise comparisons. Quantitative analyses were obtained by normalizing the protein of interest to the loading controls, β-actin for the cytosolic fraction, nucleolin for the nuclear fraction. (F) Representative microphotographs of brain sections stained with a TDP-43-specific antibody and counterstained with a nuclear dye. The arrow points to cytosolic TDP-43 accumulation. Data are presented as means ± SEM and were analyzed by one-way ANOVA, followed by Bonferroni's post hoc tests. Asterisk indicates a significant difference between WT and TgTDP-25+/+ mice; hash symbol indicates a significant difference between WT and TgTDP-25+/0 mice; double dagger indicates a significant difference between TgTDP-25+/0 and TgTDP-25+/+ mice. Gray arrows point to full-length TDP-43. Black arrows point to TDP-25.
Evidence from several laboratories has shown that deficits in protein turnover could contribute to the progression of TDP-43 proteinopathies. Autophagy and ubiquitin proteasome systems represent the two major cellular protein degradation systems that have been linked to TDP-43 proteinopathies (17,34,35). However, further studies are required to understand whether TDP-25 is directly involved in these processes. To determine the in vivo effects of high TDP-25 levels on proteasome function, we utilized the fluorogenic substrates Bz-VGR-AMC, Suc-LLVY-AMC and Z-LLE-AMC to measure chymotrypsin-like, trypsin-like and caspase-like activities of the proteasome in the brains of 15-month-old WT, TgTDP-25+/0 and TgTDP-25+/+ mice (n = 4/genotype). For the trypsin-like activity, using a two-way repeated-measures ANOVA, we found a significant effect for time [P < 0.0001; F(40, 360) = 293.6] and genotype [P = 0.001; F(2, 360) = 8.10; Fig. 4A]. A post hoc test with Bonferroni correction showed that WT mice were significantly different compared with TgTDP-25+/0 and TgTDP-25+/+ mice after a reaction time greater than 48 and 35 min, respectively. We found no difference between TgTDP-25+/0 and TgTDP-25+/+ mice. We then analyzed the average trypsin-like activity during the last 5 min of the reaction time by one-way ANOVA. We found a significant difference among the three genotypes [P < 0.0001; F(2, 12) = 173.2; Fig. 4B]. Post hoc Bonferroni's multiple comparison test indicated that WT mice were significantly different compared with TgTDP-25+/0 (P < 0.001; t = 11.80) and TgTDP-25+/+ mice (P < 0.001; t = 15.97). Furthermore, a significant difference was also evident between TgTDP-25+/0 and TgTDP-25+/+ mice (P < 0.01; t = 4.17). For the caspase-like activity, using a two-way repeated-measures ANOVA, we found a significant effect for time [P < 0.0001; F(40, 360) = 77.89] and genotype [P = 0.002; F(2, 360) = 13.21; Fig. 4C]. A post hoc test with Bonferroni correction showed that WT mice were significantly different compared with TgTDP-25+/0 and TgTDP-25+/+ mice after a reaction time >21 and 36 min, respectively. We found no difference between TgTDP-25+/0 and TgTDP-25+/+ mice. We then analyzed the average caspase-like activity during the last 5 min of the reaction time by one-way ANOVA. We found a significant difference among the three genotypes [P < 0.0001; F(2, 12) = 689.5; Fig. 4D]. Post hoc Bonferroni's multiple comparison test indicated that WT mice were significantly different compared with TgTDP-25+/0 (P < 0.001; t = 30.22) and TgTDP-25+/+ mice (P < 0.001; t = 30.80). Furthermore, a significant difference was also evident between TgTDP-25+/0 and TgTDP-25+/+ mice (P < 0.05; t = 3.58). For the chymotrypsin-like activity, using a two-way repeated-measures ANOVA, we found a significant effect for time [P < 0.0001; F(40, 360) = 249.9], but not genotype [P = 0.223; F(2, 360) = 1.78; Fig. 4E]. A post hoc test with Bonferroni correction did not reveal any significant pairwise comparison. We then analyzed the average chymotrypsin-like activity during the last 5 min of the reaction time by one-way ANOVA. We found a significant difference among the three genotypes [P < 0.0001; F(2, 12) = 109.7; Fig. 4F]. Post hoc Bonferroni's multiple comparison test indicated that WT mice were significantly different compared with TgTDP-25+/0 (P < 0.001; t = 13.10) and TgTDP-25+/+ mice (P < 0.001; t = 12.54). No significant difference was evident between TgTDP-25+/0 and TgTDP-25+/+ mice (P > 0.05; t = 0.57). Together, these data suggest that accumulation of TDP-25 decreases proteasome function in a dose-dependent manner.
TDP-25 accumulation reduces proteasome activity. We measured chymotrypsin-like, trypsin-like and caspase-like activities of the proteasome in the brains of 15-month-old WT, TgTDP-25+/0 and TgTDP-25+/+ mice (n = 4/genotype). (A) Trypsin-like activity during the 1 h reaction. It was statistically different in WT mice compared with TgTDP-25+/0 and TgTDP-25+/+ mice after a reaction time >48 and 35 min, respectively. The activity was not statistically significant between TgTDP-25+/0 and TgTDP-25+/+ mice. (B) Trypsin-like activity during the last 5 min of the reaction. All three groups were statistically significant from each other. (C) Caspase-like activity during the 1 h reaction. It was statistically different in WT mice compared withTgTDP-25+/0 and TgTDP-25+/+ mice after a reaction time >21 and 36 min, respectively. The activity was not statistically significant between TgTDP-25+/0 and TgTDP-25+/+ mice. (D) Caspase-like activity during the last 5 min of the reaction. All three groups were statistically significant from each other. (E) Chymotrypsin-like activity during the 1 h reaction. It was not statistically different among the three different groups (F) Chymotrypsin-like activity during the last 5 min of the reaction. It was significantly higher in WT mice compared with both transgenic groups. No statistically significant difference was obtained between TgTDP-25+/0 and TgTDP-25+/+ mice. Data are presented as means ± SEM and were analyzed by two-way (A, C and E) or one-way (B, D and F) ANOVA, followed by Bonferroni's post hoc tests. Asterisk indicates a significant difference between WT and TgTDP-25+/+ mice; hash symbol indicates a significant difference between WT and TgTDP-25+/0 mice; double dagger indicates a significant difference between TgTDP-25+/0 and TgTDP-25+/+ mice.
Next, we assessed autophagy induction by measuring the levels of Atg3, Atg5, Atg7 and LC3, p62 and Beclin 1 (n = 8/genotype). These are key proteins in autophagy and their levels serve as good indicators of autophagy induction (13,14). We found that the steady-state levels of Atg3 and Atg5 were significantly different among the three groups [Atg3: P < 0.0001; F(2, 21) = 21.41. Atg5: P = 0.0035; F(2, 21) = 7.5. Fig. 5A–C]. Bonferroni post doc test indicated that Atg3 levels in WT mice were significantly different compared with TgTDP-25+/0 mice (P < 0.001, t = 4.31) and TgTDP-25+/+ mice (P < 0.001, t = 6.42). Atg5 levels in WT were also significantly different compared with TgTDP-25+/0 mice (P < 0.05, t = 2.85) and TgTDP-25+/+ mice (P < 0.01, t = 3.68). The levels of both autophagy-related proteins were similar between TgTDP-25+/0 and TgTDP-25+/+ mice. Quantification of the Atg7 levels revealed no significant difference among the three groups (Fig. 5A and D). LC3-II is an autophagy-related protein derived from LC3-I during autophagy induction and is incorporated into the growing membrane of autophagosomes. We quantified the ratio of LC3-II over LC3-I and found that it was significantly different among the three genotypes [P < 0.0001, F(2, 21) = 24.60; Fig. 5A and E). Bonferroni post doc test indicated that the ratio LC3-II over LC3-I in WT mice was significantly different compared with TgTDP-25+/0 mice (P < 0.001, t = 4.97) and TgTDP-25+/+ mice (P < 0.001, t = 6.77). No difference was found between TgTDP-25+/0 and TgTDP-25+/+ mice. To further analyze autophagy induction, we measured the steady-state of p62, a protein that binds ubiquitin and LC3, and facilitates clearance of ubiquitinated proteins. We found that p62 levels were different among the three groups [P < 0.0001; F(2, 21) = 20.38; Fig. 5A and F]. Bonferroni post hoc test indicated that p62 levels in WT mice were significantly different compared withTgTDP-25+/0 mice (P < 0.05, t = 2.67) and TgTDP-25+/+ mice (P < 0.001, t = 6.36). Further, a significant difference was evident between TgTDP-25+/0 and TgTDP-25+/+ mice (P < 0.01; t = 3.687). We also found that Beclin 1 levels were significantly different among the three groups (P = 0.0003; F(2, 21) = 12.27; Fig. 5A and G]. Bonferroni post hoc test indicated that Beclin 1 levels in WT mice were significantly different compared with TgTDP-25+/0 mice (P < 0.01, t = 3.84) and TgTDP-25+/+ mice (P < 0.001, t = 4.63). No difference was found between TgTDP-25+/0 and TgTDP-25+/+ mice. Together, these data suggest that accumulation of TDP-25 impacts autophagy induction but fail to identify a dose-dependent effect, which was only evident for the changes in p62 levels.
Reduced markers of autophagy induction in TgTDP-25 mice. (A) Representative western blots of protein extracted from the brains of WT, TgTDP-25+/0 and TgTDP-25+/+ mice (n = 8/genotype) and probed with the indicated antibodies. (B and C) Quantitative analyses of Atg3 and Atg5 levels. In both cases, WT mice were significantly higher compared with the two transgenic groups. No statistically significant difference was observed between TgTDP-25+/0 and TgTDP-25+/+ mice. (D) Quantitative analyses of Atg7 levels indicated no statistically significant difference among the three groups. (E) Quantitative analyses of the ratio LC3-II over LC3-I levels. WT mice were significantly higher compared with the two transgenic groups. No statistically significant difference was observed between TgTDP-25+/0 and TgTDP-25+/+ mice. (F) Quantitative analyses of p62 levels. All three groups were statistically significant from each other. (G) Quantitative analyses of Beclin 1 levels. WT mice were significantly higher compared with the two transgenic groups. No statistically significant difference was observed between TgTDP-25+/0 and TgTDP-25+/+ mice. Quantitative analyses were obtained by normalizing the protein of interest to β-actin, used as loading control. Data are presented as means ± SEM and were analyzed by one-way ANOVA, followed by Bonferroni's post hoc tests. Asterisk indicates a significant difference between WT and TgTDP-25+/+ mice; hash symbol indicates a significant difference between WT and TgTDP-25+/0 mice; double dagger indicates a significant difference between TgTDP-25+/0 and TgTDP-25+/+ mice.
Discussion
Accumulation of TDP-43 is the major event in ALS and FTLD-TDP, two highly prevalent neurodegenerative disorders (5,36). Additionally, TDP-43 inclusions are also present in 30% of Alzheimer's disease cases (2). The C-terminal 25 kDa fragment of TDP-43 (TDP-25), generated by proteolytic cleavage of TDP-43, is consistently found in these inclusions. Therefore, understanding the mechanisms underlying TDP-43-mediated neurodegeneration and the role of TDP-25 in this process will have a wide spread impact. Here we report for the first time the behavioral and biochemical phenotype of old transgenic mice expressing TDP-25. We showed that accumulation of TDP-25 causes motor and cognitive dysfunctions in a dose-dependent manner. Given that TDP-25 is an invariable feature of FTLD-TDP and accumulates only in affected brain regions (3), these mice may represent an invaluable tool to further dissect the pathways linking TDP-25 accumulation to motor and cognitive deficits.
Pathological TDP-43 is mislocalized, from its nuclear location to the cytoplasm where it forms ubiquitin-positive inclusions (14). This type of evidence has led to the hypotheses that TDP-43 toxicity may arise from the toxic gain of function of cytosolic TDP-43 inclusions or from the nuclear loss of function of TDP-43 (37–39). The results presented here suggest that TDP-25-mediated behavioral deficits occur via a toxic gain of function mechanisms. To this end, we showed that soluble and insoluble TDP-25 accumulates in both the cytosol and nucleus of TgTDP-25 mice. In contrast, nuclear full-length TDP-43 levels were unaffected. These findings are consistent with previous reports in a Drosophila model expressing TDP-25. These animals showed reduced survival and a rough eye phonotype; this phenotype was suppressed by antagonizing TDP-25 aggregation (40). Likewise, adeno-associated virus gene transfer of TDP-25 induces toxicity in rats (41), which further strengthens the hypothesis that TDP-25 contributes to the pathogenesis of TDP-43 proteinopathies. We have previously reported that 6-month-old TgTDP-25+/0 mice developed mild deficits in behavioral tasks mainly controlled by cortical regions, without overt motor deficits (30). Here, we show that TDP-25-mediated behavioral deficits occur in a dose- and age-dependent manner. Indeed, we report that 15-month-old TgTDP-25+/0 mice show a more robust behavioral phenotype as the cognitive deficits now extend to hippocampal-dependent tasks, and there is a strong motor deficit. Such deficits were further exacerbated in TgTDP-25+/+ mice. Consistent with these findings, we have previously reported that pharmacologically increasing TDP-25 levels exacerbates behavioral deficits in TgTDP-25 mice (42). While our results support a toxic-gain of function mechanisms for TDP-25-mediated behavioral deficits, they do not exclude the loss-of-function hypothesis, which is supported by the development of TDP-43 conditional knockout mice. These mice develope neurodegeneration and motor deficits (12), clearly indicating that the loss of TDP-43 is sufficient to cause neurodegeneration. Overall, there is growing appreciation that both gain and loss of function mechanisms might contribute to the pathogenesis of TDP-43 proteinopathies (37–39).
Protein homeostasis, which is critical for normal brain function, is regulated by a balance between protein production and degradation. To this end, alteration in autophagy or proteasome function, the two major cellular systems involved in protein turnover, have been linked to several neurodegenerative disorders, including TDP-43 proteinopathies (34,35). Early in vitro work showed that modulating proteasome or autophagy activity has a direct effect on TDP-43 levels in a variety of cells (20,22,26,43). For example, we showed that autophagy inhibition increased accumulation of TDP-25, whereas inducing autophagy decreased TDP-25 levels (17). Here, we report the novel finding that expression of TDP-25 in vivo is sufficient to reduce both proteasome and autophagy function. These observations are consistent with previous reports showing that TDP-25-mediated toxicity in cells was dependent on proteasome activity (44). While the behavioral and biochemical alterations are more severe in TgTDP25+/+ mice compared with TgTDP25+/0 mice, such dose-dependent effects are less apparent for proteasome and autophagy function. For example, we measured six autophagy-related proteins but only the levels of p62 were decreased in a dose-dependent manner. While further studies are needed to delineate the mechanisms underlying these changes, it is tempting to speculate that protein homeostasis in neurons can be maintained when the levels of a toxic protein are low (e.g. like in TgTDP-25+/0 mice). However, with further accumulation of toxic proteins (e.g. like in TgTDP-25+/+ mice) the homeostasis system goes into disarray, leading to alterations in autophagy and proteasome activity.
Taken together, these data highlight a possible crosstalk between TDP-25 and these protein quality control systems: high TDP-25 levels impair autophagy and proteasome activity; in contrast reduced autophagy and proteasome activity further contribute to TDP-25 accumulation. Overall, these findings suggest that facilitating autophagy induction or proteasome activity may be a valid therapeutic approach to mitigate TDP-43 proteinopathies. To this end, proof of concept studies in mice have shown that activating autophagy ameliorates TDP-43-mediated neuronal pathogenesis in mice (27).
Materials and Methods
Mice
The generation of the TgTDP-25 mice has been described elsewhere (30). TgTDP-25+/0 mice were crossed with each other to generate TgTDP-25+/+ mice. All mice were housed 4–5 to cage at 23°C, kept on a 12 h light/dark cycle and were given ad libitum access to food and water. All animal procedures were approved by The Institutional Animal Care and Use Committee of the Banner Sun Health Research Institute. Mice were assigned to a specific group based on their genotype after birth and there were no other factors that determined group selection.
Radial arm water maze
We used a 2-day RAWM protocol as detailed in ref. (45). To summarize, the maze had eight arms and an open central area. An escape platform was only located at the end of the target arm, which changed between mice. The maze was located in a room with extramaze visual cues, which served as a reference point for the mice. The location of the cues and platform were kept constant throughout the testing period. During the first day of training, mice received 15 test trials; between each trial the platform was alternated between visible and submerged. Each trial lasted 60 s or until the mouse found the escape platform. The intertrial interval was 5 min, during which mice were placed in warm cages. On Day 2, mice received 15 additional trials. This time, all trials were conducted with the hidden platform. Every entry into an arm other than the target arm was counted as an error. Entering the same arm, other than the target arm, more than once during a trial was called a same-arm error. A video camera mounted on the ceiling recorded each performance and was scored using the EthoVision tracking system.
Morris water maze
Mice were tested in the spatial reference version of the MWM as detailed in ref. (46). The test was performed in the same room used for the RAWM, which contained extramaze cues that served as reference points for mice during the learning and probe trials. The pool was circular, 1.5 m in diameter, and filled with opaque water. The hidden platform was 14 cm in diameter and its location was kept constant throughout the testing period. Mice were tested for five consecutive days, four training trials per day. The intertrial interval was 20 s. During each day, mice were introduced into the maze from four pseudorandom locations. During the probe trials, the platform was removed from the water and mice were allowed to freely swim in the tank for 60 s. The entire test was recorded by a video camera mounted on the ceiling. Data were obtained using specialized tracking software (EthoVision XT, Noldus).
Rotarod
Mice were tested as detailed in ref. (47). Briefly, each mouse was trained for three consecutive days (six trials/day, at least 30 min apart) where the speed of the rotor was accelerated from 0 to 15 rpm in 15 s and then maintained at a constant speed of 15 rpm for 75 additional seconds, for a total of 90 s. Twenty-four hours after the last training session, the mice were tested in a probe trial consisting of six consecutive trials on a constantly accelerating rod (1 rpm/s). The latency to fall was recorded, which is indicative of motor abilities and coordination.
Protein extractions and western blots
Protein extractions were performed as described in (42). Briefly, frozen hemibrains were homogenized with a power homogenizer in 1 ml of low-salt buffer (10 mm Tris pH 7.5, 5 mm EDTA, 1 mm DTT, 10% Sucrose) containing protease inhibitors and centrifuged at 14 400 rpm for 30 min at 4°C. The supernatant was stored at −80°C as the low-salt fraction while the pellet was re-homogenized in high-salt solution (low-salt buffer plus 1% Triton X-100 and 0.5 m NaCl), in the presence of protease inhibitors and centrifuged at 20 000 rpm for 1 h at 4°C. The supernatant was stored at −80°C as the high-salt fraction. The pellet underwent further extraction in sarkosyl buffer (low salt + 1% sarkosyl + 0.5 M NaCl) and after another centrifugation at 20 000 rpm for 1 h at 4°C, the pellet was stored at −80°C as the sarkosyl fraction.
To obtain the cytosolic and nuclear fractions, brains were washed in PBS and then homogenized in a dounce homogenizer with 2 ml of solution A (10 mm HEPES pH 7.9, 10 mm KCL, 0.1 mm EDTA, 0.1 mm EGTA, 1 mm DTT) in the presence of protease inhibitors and 0.5% NP40 for a total of 10 strokes. The solution was then kept on ice for 10 min and centrifuged 1 min at 11 000 rpm. The supernatant was removed and stored at −80°C as cytosolic fraction. The pellet was re-suspended in 1 ml of solution B (20 mm Hepes pH 7.9, 400 mm NaCl, 1 mm EDTA, 1 mm EGTA, 1 mm DTT) in the presence of protease inhibitors and placed on ice for 15 min. Samples were then centrifuged 5 min at 11 000 rpm and the supernatant was stored at −80°C as nuclear fraction. Western blot experiments were run as previously described in ref. (48).
Proteasome activity
Proteasome activity was measured as detailed in ref. (49). Briefly, brain homogenates were incubated with 75 mm proteasomal substrates Suc-LLVY-AMC, Bz-VGR-AMC and Z-LLE-AMC, which probe for chymotrypsin-like, trypsin-, and caspase-like activities, respectively. Reactions were carried out in assay buffer [25 mm 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), pH 7.5, 0.5 mm ethylenediaminetetraacetic acid (EDTA), 0.05% NP-40] in a total of 200 μl in black 96-well plates. Substrates were added immediately prior to readings. Kinetic readings were taken at 37°C every 1.5 min for 60 min (excitation 360 nm, emission 460 nm) using the Synergy HT multimode microplate reader using the Gen5 software (BioTek, Winooski, VT, USA). Readings were normalized to protein concentration.
Statistical analyses
All data were analyzed using GraphPad Prism, San Diego, CA, USA, www.graphpad.com. Data were analyzed by one- or two-way ANOVA followed by Bonferroni's post hoc analysis.
Funding
This work was supported by start-up funds provided to S.O. by the Department of Basic Medical Sciences at the University of Arizona, College of Medicine – Phoenix.
Conflict of Interest statement. None declared.
