Loss of tau delays the onset of motor symptoms after PFF inoculation
We previously showed that tau is required for synaptic and memory deficits caused by A53T mutant human α-synuclein (HuαSA53T) expression in primary neurons and in the TgA53T model [4, 5]. While tau expression did not impact αS-dependent motor abnormalities in presymptomatic mice, studies have proposed that a pathological relationship exists between tau and αS. Interestingly, recent studies show that while αS promotes pathological spreading of tau in brain [6], tau did not promote spreading of αS pathology [6, 11]. Therefore, we aimed to further evaluate the role of tau in the onset and progression of α-synucleinopathy derived from HuαS.
In order to induce αS pathology in a temporally regulated manner, we used intramuscular injections of wild-type (WT) HuαS PFF into our TgA53T mouse model [12], as well as TgA53T/mTau−/− and genotypic controls (mTau−/− and nTg; Fig. 1a). Following PFF inoculations, tissue samples for histology and biochemical analyses were collected at 40 and 70 days post inoculation (dpi), as well as when the mice reached the disease end stage (classified as ataxic deterioration to the point of complete hindlimb paralysis preventing ambulation; Fig. 1a). In addition, prior to 70 dpi, mice underwent a battery of behavioral tests: pole, rotarod, and open field testing.
All PFF-inoculated TgA53T animals in the wild-type background developed motor symptoms and reached end stage by ~ 100 dpi. Significantly, the average times to reach ataxic onset and end stage were significantly delayed in TgA53T/mTau−/− compared to TgA53T (P = 0.0122 and P = 0.0022, respectively; Fig. 1b, c). Moreover, the progression from initial onset of motor symptoms to end stage was also significantly prolonged in TgA53T/mTau−/− compared to TgA53T (P = 0.0247; Fig. 1d), indicating that the loss of tau delayed the progression of α-synucleinopathy-associated neurodegeneration.
Consistent with the delay in disease progression associated with the loss of endogenous mTau expression, behavioral analysis at 70 dpi also showed that loss of tau significantly attenuated HuαSA53T-mediated behavioral abnormalities in TgA53T mice. The rotarod test of motor coordination showed that while TgA53T mice had a significantly decreased latency to fall compared to all other groups on day 3 and to mTau−/− controls on day 4, the TgA53T/mTau−/− mice had a similar latency to fall as control mice (Fig. 1e). In the pole test, an additional test of motor coordination, the TgA53T mice required significantly more time to both descend the pole, as well as for total performance (cumulative time to re-orient on the pole and descend to the base) than the control mice. The mean times for descending the pole and for total performance of the TgA53T/mTau−/− mice was intermediate compared to TgA53T mice and control mice but was not significantly different from either group (Fig. 1f, g). As previously documented [4], the TgA53T and TgA53T/mTau−/− groups exhibited similar hyperactivity in the open field test (Additional file 1: Fig. S1a). Interestingly, while the TgA53T group presented increased anxiety-associated behavior as they spent less time in the center, such behavior was not observed in the TgA53T/mTau−/− mice (Additional file 1: Fig. S1b, c). Collectively, these results show that the loss of tau expression delays the onset and progression of overt disease as well as ameliorating behavioral deficits in the TgA53T model of α-synucleinopathy.
Loss of tau expression does not impact phospho-serine129 αS (pS129 αS) pathology in end-stage TgA53T mice but leads to modest reduction in pS129 αS accumulation in presymptomatic TgA53T mice
The delay in disease/ataxic onset, as well as behavioral deficits, in the TgA53T/mTau−/− mice compared with TgA53T mice seems to contradict prior studies using the mouse αS PFF inoculation model [6, 11]. Thus, we asked whether lack of tau expression impacts subcortical α-synucleinopathy in the TgA53T model.
To survey if the expression of tau was correlated with bulk changes in αS pathology, we performed biochemical analysis of lysates from spinal cord and brain stem regions that are prone to develop αS pathology in the TgA53T model [3, 15]. Immunoblot analysis of spinal cord from end-stage mice showed similar levels of full-length αS compared to age-matched controls (Fig. 2a) and a prominent increase in pS129 αS, a marker of pathological αS, in both TgA53T and TgA53T/mTau−/− mice (Fig. 2b). Further, consistent with a prior study [6], no difference in the levels of pS129 αS/total αS was observed between TgA53T and TgA53T/mTau−/− mice (Fig. 2c, d). To determine if there was a difference in the aggregation of αS that is not fully reflected by total pS129 αS levels, we examined Triton X-100 detergent-soluble and -insoluble fractions from the spinal cord (Additional file 1: Fig. S2). As expected from prior studies [4, 20], most of the pS129 αS was present in the detergent-insoluble fraction while the detergent-soluble fraction contained very little pS129 αS (Additional file 1: Fig. S2a). Our results showed that while PFF inoculation increased the amount of insoluble αS, expression of tau did not significantly impact the levels of insoluble αS, but there was slightly increased αS in the soluble fraction (Additional file 1: Fig. S2). Moreover, tau expression did not impact the levels of SDS-stable oligomers that resolve at ~ 25 kDa, ~ 37 kDa, and > 200 kDa (Additional file 1: Fig. S2d, g). Analysis of the brainstem region showed that PFF-inoculation led to similar increases in pS129 αS levels in both TgA53T and TgA53T/mTau−/− mice at the end stage (Additional file 1: Fig. S3a, b), similar to the patterns observed in the spinal cord.
We performed histological analysis for αS pathology (pS129 αS) to determine if the pattern of αS pathology was influenced or mediated by tau expression. The immunostained lumbar spinal cord sections were used to determine the percent (%) area covered by immunoreactivity. Consistent with the bulk biochemical analysis, our results showed that both TgA53T and TgA53T/mTau−/− mice developed similar levels of pS129 αS pathology at end stage while the control animals (nTg and mTau−/− alone) did not exhibit pS129 αS staining (Fig. 3a–e). In addition, we also examined activation of microglia and astrocytes by Iba1 and GFAP immunostaining, respectively. In end-stage animals, quantitative analysis of microglial activation (Iba1) showed significant activation with αS pathology but did not reveal any differences as a function of tau expression (Fig. 3f–j). Similarly, while the increase in GFAP staining occurred with αS pathology, there were no differences as a function of tau expression (Fig. 3k–o; see Additional file 1: Fig. S4 for representative low-magnification images). Importantly, pS129 αS histopathology and neuroinflammation (i.e., increased Iba1 and GFAP abundance) were only observed after injection of αS PFF to TgA53T overexpression mice but not in age-matched TgA53T without αS PFF injection, or age-matched nTg mice with or without αS PFF injection (Additional file 1: Fig. S5). In addition, pS129 αS, Iba1, and GFAP reactivity were elevated in end-stage brainstem, cerebellum, and motor cortex regions, without any obvious qualitative differences between TgA53T and TgA53T/mTau−/− mice (Additional file 1: Fig. S6).
Because end stage of disease is reached at a later dpi in TgA53T/mTau−/− mice, it is possible that αS pathology was normalized between groups at terminal stages. Thus, we evaluated tissues at the same presymptomatic stage following PFF inoculation (70 dpi and 40 dpi). Biochemical analysis showed that at 70 dpi, the relative levels of pS129 in the spinal cord were increased over nTg mice but at levels much lower than at the end stage. Further, the immunoblots of 70 dpi spinal cord (Fig. 4b) and brain stem lysates (Additional file 1: Fig. S3c, d) failed to show any differences in pS129 αS levels or insoluble αS as a function of tau expression (Additional file 1: Fig. S7e, g). Moreover, tau expression did not impact the levels of SDS-stable αS oligomers (Additional file 1: Fig. S7e, f, i, j).
We also performed histological analysis in 70 dpi mice for pS129 αS and neuroinflammation as was done for the end-stage subjects above. Significantly, despite the lack of differences in bulk immunoblot analysis of total pS129 αS levels (Fig. 4), quantitative analysis of the pS129 αS staining in the lumbar spinal cord revealed a modest decrease in pS129 αS pathology in TgA53T/mTau−/− compared to TgA53T mice (Fig. 5a-e). This suggested that while the overall abundance of pS129 αS along the spinal cord, measured biochemically (Fig. 4), was unchanged, there was a specific delay in the neuronal accumulation of pS129 αS in the grey matter of the spinal cord.
The analysis of microglial (Fig. 5f–j) and astrocytic (Fig. 5k–o) activation in presymptomatic animals at 70 dpi showed a significant increase in microglial activation in mice with αS pathology. Consistent with the reduced αS pathology in TgA53T/mTau−/− animals, the activation status of microglia and astrocytes was reduced in these mice compared to TgA53T mice (Fig. 5j, o; see Additional file 1: Fig. S8 for representative low-magnification images). In the brainstem, pS129 αS, Iba1, and GFAP were modestly increased at 70 dpi in TgA53T and TgA53T/mTau−/− mice. The overall abundance was qualitatively similar to the quantitative results seen with spinal cord sections (Additional file 1: Fig. S9).
Given that loss of tau expression is associated with reduced αS pathology at 70 dpi, we analyzed animals at 40 dpi to determine if tau expression affects the onset of αS pathology following intramuscular injection. A previous study showed that following intramuscular injection, αS pathology appears in spinal cord between 30 and 60 dpi [25]. Western blot analysis of 40 dpi mice showed no obvious increase in the levels of pS129 αS (Additional file 1: Fig. S10a, b). However, immunohistological analysis of 40 dpi mice showed that sparse pS129 αS pathology could be detected, confirming that the αS pathology was initiated between 30–60 dpi. Quantitative analysis showed that the amount of pS129 αS pathology at 40 dpi was not altered by tau loss (Additional file 1: Fig. S10c–f). Finally, analysis of neuroinflammation (Iba-1 and GFAP) showed that the low level of early αS pathology at 40 dpi was not associated with increased neuroinflammation (Additional file 1: Fig. S10g–n).
Loss of tau expression leads to reduced neurodegeneration in presymptomatic TgA53T mice but not in end-stage mice
Thus far, our results indicate that while tau is associated with modest increases in αS pathology at 70 dpi, tau seems not to impact the onset of αS pathology at 40 dpi or the extent of αS pathology at end stage. However, it is possible that tau might be acting downstream of αS abnormalities [4, 5]. Because loss of motor neurons is a robust neurodegenerative phenotype in the TgA53T model [16], including the intramuscular injection model [25], we examined if tau expression affects the loss of motor neurons in the spinal cord of TgA53T mice following the PFF injection.
We analyzed NeuN+ neurons in the ventral horn and dorsal horn of both 70-dpi and end-stage mice (see Additional file 1: Fig. S11 for representative low-magnification images and regions analyzed). Our analysis of end-stage mice following PFF inoculation showed that the presence of αS pathology and limb paralysis in the TgA53T mice were accompanied by severe loss of motor neurons in the ventral horn (Fig. 6a–e). Further, the average number of motor neurons per lumbar spinal cord section in TgA53T/mTau−/− mice was not different from TgA53T mice. This was further demonstrated with loss of total NeuN+ content of the ventral horn (Fig. 6f), while the dorsal horn neurons were left intact (Fig. 6g). Significantly, analysis of presymptomatic animals at 70 dpi showed that the presence of αS pathology in TgA53T mice was already associated with a significant loss of ventral horn motor neurons in the lumbar spinal cord, albeit less severe than in the end-stage animals (Fig. 6h–l). More importantly, the loss of motor neurons was attenuated in the TgA53T/mTau−/− mice, despite the significant presence of αS pathology, compared to the TgA53T mice (Fig. 6l; see Additional file 1: Fig. S11 for representative low-magnification images). In addition, this protection was also observed after quantification of ventral horn neurons via NeuN+ staining (Fig. 6m), while the dorsal horn neurons were not affected (Fig. 6n).
Loss of tau does not impact soluble α-synuclein oligomer formation or glycogen synthase kinase 3β (GSK3β) activity
Using a different HuαSA53T-expressing transgenic mouse line (M83), it has been previously demonstrated that TgA53T (M83) mice accumulate soluble tau oligomers with aging, and treatment with tau-oligomer-specific antibody can subsequently reduce αS oligomers and aggregates [10]. Therefore, while the loss of tau expression does not alter overt αS aggregation or SDS-stable oligomers (Fig. 2, Additional file 1: Fig. S2 and S7), we tested if the loss of tau expression reduces the levels of soluble αS oligomers in our TgA53T model using the Syn33 antibody [4, 10, 24] with non-denaturing dot blot analysis.
Dot blot analysis of buffer-soluble fractions from spinal cords of both end-stage and 70-dpi spinal cord lysates from TgA53T animals revealed higher levels of αS oligomers as a function of HuαSA53T expression compared to nTg and mTau−/− controls. Further, the accumulation of Syn33+ oligomeric species was comparable between TgA53T and TgA53T/mTau−/− groups (Fig. 7a, b). Consistent with the specificity of Syn33 to soluble oligomers, Syn33 did not react to the detergent-insoluble fractions (Fig. 7c, d). These results show that tau expression does not impact the levels of soluble oligomers recognized by Syn33 in our TgA53T model. Similar results were seen in our analysis of cortical and hippocampal tissues [4].
We also examined whether the αS pathology in the TgA53T mouse model was associated with obvious increases in pathological tau. However, our analysis for AT8, PHF1, and MC1 pathological tau showed that, even at the end stage, accumulation of hyperphosphorylated tau was not detected both biochemically and histologically (Additional file 1: Fig. S12g–l). Increased activation of GSK3β has been proposed as a mediator of αS-induced neuronal dysfunction [5, 7, 26, 27]. Thus, we also examined spinal cord lysates for the levels of active GSK3β, as measured by phosphorylated Tyr-216 (pY216) [28]. Our results showed that neither total GSK3β levels nor pY216-GSK3β activation was increased as a function of αS pathology, or was impacted by tau expression (Additional file 1: Fig. S12a–f).
Endoplasmic reticulum stress (ERS) and autophagy pathway protein clearance
We previously showed that α-synucleinopathy in the TgA53T model is associated with chronic ERS [14] and dysfunctions in the autophagy-lysosomal pathway (ALP) [15]. Because both ERS and ALP deficits follow the onset of αS pathology, we examined whether ERS and ALP in the TgA53T model are affected in a tau-dependent manner. We performed biochemical (Western blot) analysis in 70-dpi and end-stage spinal cord lysates for markers of ERS and ALP (Additional file 1: Fig. S13). While there was no obvious indication of ALP abnormalities at 70 dpi (Additional file 1: Fig. S13a), analysis of spinal cord lysates from end-stage TgA53T mice showed expected ALP abnormalities (Additional file 1: Fig. S13b). ALP markers LC3 II/I ratio, p62, and pAMPK/AMPK ratio were not different between TgA53T and TgA53T/mTau−/− mice.
Analysis of ERS markers Grp78 and p-eIF2α/eIF2α ratio showed expected signs of chronic ERS in end-stage TgA53T and TgA53T/mTau−/− mice (Additional file 1: Fig. S13b). While the level of Grp78 was similarly increased in both TgA53T and TgA53T/mTau−/− mice, TgA53T/mTau−/− had modest but significantly reduced p-eIF2α/eIF2α ratio compared to TgA53T. No signs of ERS were seen in lysates from 70-dpi subjects (Additional file 1: Fig. S13a). Similar results for ALP and ERS markers were observed in brainstem lysates from 70-dpi and end-stage animals (Additional file 1: Fig. S14). There were no signs of change in ERS or ALP in the brainstem of 70-dpi animals (Additional file 1: Fig. S14a). However, in the brainstem, there were elevated ALP markers at the end stage as previously described [15], but no sign of ERS in the end-stage subjects (Additional file 1: Fig. S14b).
Collectively, our results show that both ERS and ALP deficits associated with α-synucleinopathy in the TgA53T mouse model occur after the αS pathology is well developed. Further, unlike the loss of motor neurons (Fig. 6), both ERS and ALP deficits seem to be independent of tau expression.
Tau expression does not affect αS PFF uptake or processing in neurons but prevents PFF-induced neurotoxicity
Our in vivo studies using the TgA53T model show that the loss of tau expression leads to a delay in αS aggregation, inflammation, and neurodegeneration. However, it is unknown if the loss of tau expression affects early processes related to the development of αS pathology. Moreover, if loss of tau allows neurons to be more resistant to the toxic effects of αS pathology, neurons may be able to more efficiently clear abnormal αS. Thus, we determined whether tau modulates the initiation of αS pathology and αS-dependent neurodegeneration in a cell autonomous manner. To address this, we established primary neuronal cultures from wild-type (nTg) and mTau−/− mice and exposed the neurons to αS PFF.
To determine whether tau mediates the initiation and progression of intraneuronal αS pathology following PFF exposure, we examined the neuronal uptake of WT αS PFF in primary hippocampal neurons cultured from nTg and mTau−/− mice. Cultured neurons were exposed to αS PFF for 2 h. After this 2-h incubation, PFF-containing medium was removed, neurons were washed with PFF-free medium to remove any extracellular PFF, and then fresh PFF-free medium was added. Neuronal lysates were then collected at 0, 3, 6, 16, 24 and 48 h following the 2-h incubation with αS PFFs. As expected, neurons rapidly internalized αS PFF and the internalized αS accumulated as a truncated protein [29] (Additional file 1: Fig. S15a, b). Quantitative analysis of uptake, truncation, and clearance of αS PFF showed that the nTg and mTau−/− neurons metabolized exogenous PFF almost identically (Additional file 1: Fig. S15a, b).
We next investigated whether tau expression affects αS aggregation and neuronal survival following αS PFF exposure. Primary nTg mouse hippocampal neurons were exposed to αS PFF and evaluated for the presence of pathological pS129 αS and neurodegeneration 14 days post PFF exposure in vitro. αS PFF applied to primary hippocampal neurons in vitro led to a dose-dependent increase in pS129 αS accumulation at 14 days post-PFF in the absence of significant NeuN+ neuronal loss (Additional file 1: Fig. S15c–e). To determine if tau expression modulates PFF-induced αS aggregation in neurons, we examined the levels of pS129 αS in nTg and mTau−/− cultures exposed to PFF treatment. To account for possible differences in the density of neurons and/or neurites between the cultures, the area of pS129 αS staining was normalized to NeuN+ cells or the MAP2-stained area. When normalized to NeuN, Tau−/− neurons exhibited higher pS129αS staining than the nTg neurons and a significant difference existed between PFF-treated nTg and mTau−/− neurons (Fig. 8c). When normalized to MAP2-stained area, there was no difference in pS129 αS accumulation between nTg and mTau−/− neurons at 14 days post-PFF (Fig. 8a, b, d). The addition of αS PFF did not induce loss of NeuN+ neurons in culture (Additional file 1: Fig. S15c, d) but did induce progressive loss of MAP2-stained dendrites (Fig. 8e, g; Additional file 1: Fig. S16). Specifically, αS PFF-induced αS aggregation in nTg neurons led to simplification of dendritic arborization, as indicated by reduced MAP2-stained area per NeuN+ cell at 14 days post-PFF (Fig. 8e, g; Additional file 1: Fig. S16). Significantly, in mTau−/− neurons, the loss of MAP2 at 14 days post-PFF was prevented (Fig. 8f, g). These results show that the loss of mTau expression attenuates the PFF-induced neuronal toxicity without decreasing the neuronal accumulation of pathological αS.