Neuronal MSUT2 level is associated with the presence of tau pathology in human brains

The human brain contains multiple cell populations, which are differentially affected in human tauopathies [12]. We used double immunofluorescence to co-stain human fronto-cortical sections with antibodies to MSUT2 and neuronal (NeuN), astrocytic (GFAP), microglial (IBA1), or oligodendroglial (SOX10) markers to determine MSUT2 expression in different cell types in the central nervous system. Generally, MSUT2 was found to be expressed in all types of brain cells (Fig. 1a) with a higher level of co-localization in neurons compared to the other cell types (Fig. 1b), emphasizing the potential importance of MSUT2 in neuronal populations. A previous study showed a biphasic pattern of MSUT2 expression in AD patient brains [82], in which some of the AD cases contained higher levels of MSUT2 than non-AD control brains while other AD cases contained very low levels of MSUT2. We hypothesize that this was due to variability in neuron loss in these cases. Thus, to determine if tau pathologies affects MSUT2 levels in patient brains, we measured neuronal MSUT2 in both non-tauopathy (non-tau pathology control: Parkinson’s disease, Multiple System Atrophy and non-pathologic cases) and tauopathy patients (Table 1), and determined whether there was a correlation with the amount of tau pathology in the fronto-cortical region (Fig. 1c). Our results showed that MSUT2 (in green) only occasionally co-localized with PHF-1-positive neurofibrillary tangles (NFTs, in red) in AD and CBD brains, and in general we did not observe a high level of colocalization between MSUT2-positive neurons and the presence of NFTs in diseased brains. However, the quantification of total MSUT2 and tau pathology showed that neuronal MSUT2 levels were significantly elevated in AD and corticobasal degeneration (CBD) cases (Fig. 1d, e), with a non-significant trend toward increase in PSP cases (Fig. 1f) that showed higher variability and a lower level of tau pathology than AD or CBD cases. In addition, neuronal MSUT2 positively and weakly correlated with the total amount of tau pathology (Fig. 1g) but not with patient age, disease duration, or Aβ plaque load (Fig. S1a–c).

Fig. 1

MSUT2 expression is associated with human tau pathology. a Immunofluorescence co-staining of frontal cortical human brain sections with antibodies against neurons (NeuN), astrocytes (GFAP), microglia (IBA1), oligodendrocytes (SOX10) and MSUT2. Scale bar = 50 µm. b Measurement of MSUT2 and cellular fluorescent signal co-localization in a. *P < 0.05 by one-way ANOVA followed by Tukey’s post hoc test, n = 3. The error bars represent the standard deviation. c Representative images of non-tauopathy control (CTR), Alzheimer’s disease (AD), Corticobasal degeneration (CBD), and Progressive supranuclear palsy (PSP) patient brain sections (frontal cortex) co-stained with MSUT2 and p-tau (PHF1) antibodies to reveal MSUT2 protein and tau pathology. Scale bar = 5 µm. d–f Quantification of MSUT2 immunoreactivity (MFI = mean fluorescence intensity) in non-tauopathy control (CTR) vs. diseased brain sections as exemplified in c. MSUT2 area is normalized to the neuron counts. Four random images of the cortical region were used for the quantification of each case. *P < 0.05, **P < 0.01, n.s., not significant, by t-test when comparing CTR (n = 9) vs. AD (n = 17), CBD (n = 6), or PSP (n = 5). The error bars represent the standard deviation. g Correlation of MSUT2 expression and tau pathology (PHF1 immunoreactivity) quantified from sections as exemplified in c. R2 = 0.2256, P = 0.0030 by normal linear regression. Four random images of the cortical region/case were used for the quantification of both MSUT2 and PHF1. Each dot represents the mean value from one individual case. Both MSUT2 and tau pathology level was presented by the integrated mean fluorescence intensity of PHF1 staining. h Representative immunoblots of MSUT2 and p-tau levels in human brains and cellular homogenates probed with MSUT2, PHF1, and GAPDH antibodies. Samples include CTR, MCI (mild cognitive impairment), AD, CBD, and PSP brains. The CTR, MCI, and AD cases were separated and run on separate blots due to lane limitations. In addition, for the gels involving CTR cases in CBD and PSP, the control (CTR) cases were randomly selected from within the same CTR group. Three isoforms of MSUT2 are indicated using QBI HEK-293 cell lysate samples. i–l Quantification of MSUT2 immunoreactivity in immunoblot samples as depicted in h. All samples were normalized to GAPDH to gain the relative quantity (RQ). All three isoforms were included for quantification. n.s., not significant, *P < 0.05, **P < 0.01 by t-test, n = 14 CTR vs. 12 MCI, n = 14 CTR vs. 12 AD, n = 7 CTR vs. 6 CBD, n = 8 CTR vs. 5 PSP. The error bars represent the standard deviation

Table 1 Demographic features of used human patient cases

To extend these immunohistochemical observations, we measured MSUT2 expression levels in human tauopathy cases using immunoblots (Fig. 1 h). The results are consistent with the immunofluorescent experiments and verified that the MSUT2 protein level was increased in patient brains with tau pathology when compared with the non-tauopathy controls (CTR). Notably, increased levels of MSUT2 could be demonstrated in different tauopathies and at different disease stages, as patients with mild cognitive impairment (MCI), AD, and CBD all showed significantly greater levels of MSUT2 (Fig. 1h–k). However, we observed no statistical difference in MSUT2 levels between PSP and CTR brains in the immunoblots, in keeping with the non-significant trend observed by immunofluorescence staining (Fig. 1f and l). In summary, these observations indicate that neuronal MSUT2 level is increased in MCI, AD and CBD brain, and there is a weak but positive correlation between MSUT2 levels and the amount of tau pathology.

MSUT2 mediates tau spreading in vivo

Most human tauopathy cases are sporadic and independent of known genetic causes. Although it was previously reported that loss of MSUT2 mitigates tau pathology in tau transgenic models [30, 82], its relevance to sporadic human tau pathology is unclear and the specific molecular mechanism(s) of MSUT2 regulation of tau pathology have not been elucidated. Here, we undertook studies to determine if MSUT2 is involved in pathways controlling cell-to-cell spreading of tau. Due to transgene-driven overexpression of tau in transgenic models, it is difficult to tease apart the spatiotemporal development of tau pathology and the extent to which tau spreading contributes to the development of tau pathology. Therefore, we leveraged an established tau seeding and spreading model [28, 60], comparing MSUT2 knockout (KO) and wild-type mice [83] to quantitatively evaluate the effect of MSUT2 on the development and spreading of tau pathology independent of tau overexpression. Before initiating seeding studies, we first examined the expression levels of a series of key proteins that might be associated with tau pathology and overall neurobiology in brains of adult MSUT2 KO and wild-type mice. We did not observe any significant changes in the expression of the tested proteins, other than MSUT2 (Fig. S2). Given that MSUT2 was functionally knocked out at one of the zinc finger domains [83] in the MSUT2 KO mice, we did detect faint bands in the MSUT2 blots that likely result from incomplete degradation of the non-functional MSUT2 protein. These results indicate that levels of proteins involved in neuron viability (tau, APP, Fox-3), synaptic plasticity (synapsin, synaptophysin, PSD95), proteasomal activity (K48-polyUb, ubiquitin), and glial function (GFAP, Iba1) are not significantly altered in the brains of MSUT2 KO mice compared to their wild-type littermates. These findings are in keeping with the observation that MSUT2 KO mice appear to be free of significant developmental or central nervous system abnormalities [83].

To investigate whether MSUT2 KO affects tau spreading, we enriched pathogenic tau seeds from AD postmortem brains (AD-tau) and injected them into the hippocampi of MSUT2 KO and wild-type mice at 3 months of age. We measured the turnover of tau seeds and the spreading of tau pathology at different time points post-injection (Fig. 2a). Since prior studies [2, 28] have already documented that human tau seeds can be degraded within 7–14 days post-injection, we first evaluated if the elimination of injected human tau is affected by MSUT2 genotype. To this end, we probed for AD-tau seeds at 3- and 7-days post-injection (d.p.i.) using an anti-phospho-tau (p-tau) antibody (AT8) (Fig. S3a and b) or anti-human tau antibody (HT7) (Fig. S3c and d). After quantification of the immunoreactivity of both antibodies, we found no difference between human tau injected into MSUT2 KO mice or their wild-type littermates, suggesting that the global removal of pathogenic tau seeds was not substantially affected by the loss of MSUT2 in the mouse brain.

Fig. 2

MSUT2 modulates the spatiotemporal spreading of tau pathology induced by human-derived AD-tau seeds. a Schematic picture of the experimental paradigm: AD-tau seeds were enriched from postmortem AD brains and stereotactically injected into the hippocampi of MSUT2 KO mice and wild-type (WT) littermates at 3 months of age. Mouse brains were collected at different time points (from 3 days to 12 months post-injection time; m.p.i.) and analyzed by immunohistochemistry and biochemistry. b Representative images of AD-tau-injected mouse hippocampi at 1 to 12 m.p.i. AD-tau-seeded MSUT2 KO and WT mouse tau pathology was revealed with a S199/T205-p-tau antibody (AT8). Scale bar = 60 and 15 (inset) µm. c Quantification of neurofibrillary tangles (NFTs) in the ipsilateral hippocampal region of the AD-tau-injected MSUT2 KO and WT mouse brains. **P < 0.01, ***P < 0.001 by two-way ANOVA followed by Bonferroni’s post hoc test, n = 6 per group. The error bars represent the standard deviation. d MSUT2 KO and WT mouse brains were injected with AD-tau and analyzed at 3 m.p.i. Proteins were extracted and fractionated from the ipsilateral hippocampi. Soluble fractions were probed with a mouse tau (m-tau) antibody T49 and GAPDH antibody as the loading control. Insoluble fractions were probed with T49 antibody. e, f Quantification of T49-positive optical density from immunoblots in d. Relative Quantities (RQ) were determined by normalizing T49 to GAPDH signals in the soluble fraction. n.s., not significant, ***P < 0.001 by t-test, WT vs. MSUT2 KO n = 4 per group. The error bars represent the standard deviation. g Representative images showing T231-p-tau pathology in the AD-tau-injected (12 m.p.i.) WT and MSUT2 KO mouse brains. Mouse brain sections were stained with a p-tau antibody (AT180). Scale bar = 60 and 15 (inset) µm. h Quantification of AT180-positive area in the ipsilateral hippocampi of the mice as depicted in g. *P < 0.05 by t-test, WT vs. MSUT2 KO, n = 6 per group. The error bars represent the standard deviation. i Representative images showing S396/S404-p-tau pathology in the AD-tau-injected (12 m.p.i.) WT and MSUT2 KO mouse brains. Mouse brain sections were stained with a p-tau antibody (PHF1). Scale bar = 60 and 15 (inset) µm. j Quantification of PHF1-positive area in the ipsilateral hippocampi of the mice as depicted in i. *P < 0.05 by t-test, WT vs. MSUT2 KO, n = 6 per group. The error bars represent the standard deviation. k Heatmap of tau pathology at 12 m.p.i. in different brain regions of MSUT2 KO and WT mice aligned in high (top) to low (bottom) anterograde connectivity strength to the injection site. Color hue indicates the abundance of tau pathology (Tau). n = 4 per group. l Heatmap of tau pathology at 12 m.p.i. in different brain regions of MSUT2 KO and WT mice aligned in high (top) to low (bottom) retrograde connectivity strength to the injection site. Color hue indicates the abundance of tau pathology (Tau). n = 4 per group. m, n Fold-change (WT/KO) in tau pathology based on different levels of anterograde and retrograde connectivity strength. High = 1st–25th ranked regions, Medium = 26th–50th ranked regions, Low = 89th–113th ranked regions (lowest 25 regions). n.s., not significant, *P < 0.05 by one-way ANOVA followed by Tukey’s post hoc test, high vs. medium vs. low connectivity region, n = 25 per connectivity category. The error bars represent the standard deviation

Acute neuroinflammation may impact the viability of neurons in various types of neurodegenerative diseases [31, 49]. To evaluate if KO of MSUT2 alters markers of astrocyte or microglia activation after AD-tau seeding, we used immunohistology to assess astrocytes (GFAP) and microglia (Iba1) in the area of the injection site where glial activation can be observed in 7 d.p.i. mouse brains. Notably, no differences were observed in the area occupied by these glial cells, or in overall glial morphology, between MSUT2 KO mice and wild-type mice (Fig. S3e–h). The results suggest that MSUT2 does not affect the acute neuroinflammatory response caused by the injection of AD-tau in mouse brains. As expected, we did not observe any mouse tau pathology at 3 and 7 d.p.i. in AD-tau injected mouse brains (Fig. S3 a and b). However, the AD-tau injected mice began to develop mouse tau pathology at 1-month post-injection (m.p.i.), which could be revealed with the anti-pTau AT8 antibody (Fig. 2b). As in human AD patient brains, the induced-AD-tau pathology in the mouse brain is present as both neurofibrillary tangles and neuritic tau pathology, which likely represent different stages of tau spreading [26]. We compared both forms of tau pathology by quantifying the number of AT8-positive NFTs, the area occupancy of AT8-positive staining, and the ratio of NFTs to total area of tau pathology after AD-tau injection in MSUT2 KO mice and their wild-type littermates. Our results show that NFTs are significantly reduced in MSUT2 KO mice (Fig. 2c) up to 12 m.p.i. Quantification of total tau pathology (combined NFT and neuritic tau pathology) shows a similar trend of significant decrease in the MSUT2 mouse brains (Fig. S4a).To further confirm alteration in pathologic tau in the MSUT2 KO mice, we analyzed the insoluble mouse tau aggregates in 3 m.p.i brains and found that the amount of sarkosyl-insoluble tau was decreased in the MSUT2 KO mouse brain (Fig. 2d and f), while the soluble tau remained unchanged (Fig. 2d and e). Moreover, we confirmed the decrease in tau pathology in the MSUT2 KO mouse brains with two additional anti-p-tau antibodies, AT180 (Fig. 2g and h) and PHF1 (Fig. 2i and j). We also evaluated the astrocytic and microglial response during the progression of tau spreading and found no difference between MSUT2 KO and wild-type littermate brains (Fig. S4b and c), which further implicates a neuronal role of MSUT2 in tau pathology in these models. Our data suggest that MSUT2 regulates temporal tau seeding and/or spread in vivo.

We next determined how MSUT2 expression affects the spatial distribution of tau pathology at the neuronal connectome level. Assuming the initial seeding of tau pathology started near the injection sites of the mouse brains [15, 19, 57], we compared the distribution of AT8-positive tau pathology in 12 m.p.i mice and correlated the severity of tau pathology burden with neuronal connectivity strength to the injection sites [62]. At this time point, tau pathology was present in most brain regions. We quantitatively measured tau pathology in more than 100 regions of the brain and analyzed the abundance of tau pathology based on anterograde and retrograde connectivity of neurons with the injection sites. Initially, we noticed that tau pathology in some brain regions is differentially affected by KO of MSUT2, with certain brain regions having higher (e.g., dentate gyrus) or lower (e.g., midbrain reticular nucleus) pathology relative to wild-type mice (Fig. S5a). When we aligned brain regions containing tau pathology with the connectivity strength, we found the distribution of tau pathology was well correlated with the anterograde connectivity strength (Fig. 2k) and to a lesser extent, with retrograde connectivity strength (Fig. 2l, Data 1) in both MSUT2 KO and wild-type mice. Furthermore, when comparing wild-type mice with MSUT2 KO mice, we observed a global reduction in tau pathology in the latter, which was again more significant in regions anterogradely connected to the injection site compared to regions retrogradely connected to the injection site. The fold-change in tau pathology (WT/KO) was significantly higher in the anterogradely connected regions with high connectivity strength than those ones with low connectivity strength (Fig. 2m). This trend was not observed in the retrogradely connected regions, perhaps due to there being less overall tau pathology in retrogradely than in anterogradely connected neurons (Fig. 2n). A distribution heat-map of tau pathology using a semi-quantitative scale (Fig. S5b) shows that the brain regions containing tau pathology had lesser pathologic burden in the MSUT2 KO mice at earlier time points (3, 6, 9 m.p.i.). Altogether, our observations suggest that loss of MSUT2 reduces the neuron connectome-dependent spreading of tau pathology throughout the brain, perhaps by inhibiting the seeding of tau pathology.

MSUT2 specifically modulates tau pathology but not other proteinopathies

Human tau pathology exhibits strain-like properties, including neuronal and non-neuronal forms of tau pathology, and differing isoform composition and bioactivity of tau aggregates [34, 59, 85]. We thus wanted to evaluate whether the modulatory effects of MSUT2 KO on tau pathogenesis that was observed upon AD-tau seeding would also be seen after seeding with pathologic tau isolated from other subtypes of tauopathies. We previously established models of FTLD-tau pathology after seeding mouse brains with tau preparations enriched from human brains with CBD (CBD-tau) or PSP (PSP-tau). These mouse models recapitulated the tau isoform composition and cell-type distribution of tau pathology observed in their human counterparts [34, 85]. Here, we injected CBD-tau (Fig. 3a–c, Fig. S6a–c) and PSP-tau (Fig. 3d–f, Fig. S6d–f) into the MSUT2 KO mice and wild-type littermates and measured the amount of tau pathology that formed over time using the AT8 antibody. The results show that NFTs were reduced in multiple brain regions of MSUT2 KO mice compared to wild-type mice after injection of either CBD-tau or PSP-tau (Fig. 3b, e). To determine if loss of MSUT2 affects non-neuronal forms of tau pathology, we quantified CBD-specific astrocytic plaque tau (APs), PSP-specific tufted astrocyte tau (TAs) (Fig. 3c, f) and oligodendroglia tau pathology (Fig. S6g–j) in the mouse brains and found no differences between MSUT2 KO mice and wild-type littermates. In summary, our data show that MSUT2 regulates the seeding and/or spreading of different subtypes of neuronal tau pathology, while it has little effect on glial forms of tau pathology, further emphasizing the neuronal role of MSUT2 on tau pathogenesis.

Fig. 3

MSUT2 modulates tau pathology but not Aβ or α-synuclein pathology. a Representative images of AT8-positive tau pathology in CBD-tau-injected MSUT2 KO (KO) and wild-type (WT) mouse brains. Mice were injected with CBD-tau at 3 months of age and brains were collected, sectioned, and stained with AT8 antibody for tau pathology at 6 m.p.i. iHP = ipsilateral hippocampus; NFTs = neurofibrillary tangles; APs = astrocytic plaques (arrows). Scale bar = 50 μm. b, c Quantification of NFTs and APs (count) in sections stained as in a. *P < 0.05, n.s., not significant by t-test, MSUT2 KO (KO) vs. wild-type (WT) mouse, n = 6 mice per group. Error bars represent the standard deviation. d Representative images of AT8-positive tau pathology in PSP-tau-injected MSUT2 KO (KO) and wild-type (WT) mouse brains. Mice were injected with PSP-tau at 3 months of age and brains were collected, sectioned, and stained with AT8 antibody for tau pathology at 6 m.p.i. iHP = ipsilateral hippocampus; NFTs = neurofibrillary tangles; TAs = tufted astrocytes (arrowheads). Scale bar = 50 μm. e, f Quantification of NFTs and TAs (count) in sections stained as in d. *P < 0.05, n.s., not significant by t-test, MSUT2 KO (KO) vs. wild-type (WT) mouse, n = 6 mice per group. Error bars represent the standard deviation. g Representative images of Aβ plaque pathology in the brains of 5xFAD/MSUT2 KO (5xKO) and 5xFAD/MSUT2 wild-type (5xWT) mice. Mice were sacrificed at 8 months of age. Mouse brain sections were stained with H31L21 (Aβ42) antibody to reveal the Aβ plaques. Scale bar = 100 µm. h Quantification of Aβ plaque-positive area (H31L21 immunoreactivity) in 5xKO and 5xWT mouse brains in sections stained as in g. n.s., not significant by t-test, 5xFAD vs. 5xKO, n = 6 mice per group. Error bars represent the standard deviation. i Counts of Aβ plaque numbers in 5xKO and 5xWT mice in sections stained as in g. n.s., not significant by t-test, 5xFAD vs. 5xKO, n = 6 mice per group. Error bars represent the standard deviation. j Representative images of ipsilateral hippocampal regions of MSUT2 KO (KO) mice and wild-type (WT) littermates injected with mouse-α-synuclein preformed fibrils (mSyn-pffs) at 1 m.p.i. Mouse brain sections were stained with EP1536Y antibody to reveal α-synuclein pathology. Scale bar = 200 μm. k Quantification of α-synuclein pathology area (EP1536Y immunoreactivity) in the ipsilateral hippocampal regions of mSyn-pffs-injected MSUT2 KO (KO) mice and wild-type (WT) littermates in sections stained as in j. n.s., not significant by t-test, MSUT2 KO (KO) vs. wild-type (WT) mouse, n = 4 mice per group. Error bars represent the standard deviation. l 5xKO and 5xWT were injected with AD-tau at 7 months of age. Mouse brains were collected, sectioned and stained for neuritic plaque (NP) tau pathology with AT8 antibody at 1 m.p.i. representative images show ipsilateral hippocampal regions of injected mouse brains. Scale bar = 500 µm. m Quantification of NP tau pathology area in AD-tau-injected 5xKO and 5xWT mice at 1 m.p.i. in sections stained as in l. *P < 0.05 by t-test, 5xFAD vs. 5xKO, n = 6 mice per group. Error bars represent the standard deviation. n Brain sections from aged (12–15-month-old) AD-tau-injected MSUT2 KO (KO) and wild-type (WT) littermates at 3 m.p.i. were probed with AT8 antibody for tau pathology. Representative images show AT8-positive staining in the caudal hilus regions of the injected mouse brains. Scale bar = 500 µm. o Neurofibrillary tangles were counted in the ipsilateral hippocampal regions of AD-tau-injected aged MSUT2 and wild-type littermates at 3 m.p.i. in sections stained as in n. *P < 0.05, ***P < 0.001 by one-way ANOVA followed by Newman-Keuls’ post hoc test, MSUT2 KO (KO) vs. wild-type (WT) mouse, n = 6 mice per group. Error bars represent the standard deviation. p AT8-positive tau pathology area was quantified in the ipsilateral hippocampal regions of AD-tau-injected aged MSUT2 KO (KO) and wild-type (WT) littermates in sections stained as in n. n.s., not significant, *P < 0.05 by one-way ANOVA followed by Newman-Keuls’ post hoc test, WT vs. KO, n = 6 mice per group. Error bars represent the standard deviation

Human neurodegenerative diseases usually present with multiple co-pathologies, and besides the defining tau and amyloid β (Aβ) pathologies, α-synuclein (α-syn) and TDP-43 inclusions are the most common types of co-pathologies in AD. To test if MSUT2 can regulate other forms of proteinopathy besides tau pathology, we crossbred MSUT2 KO mice with 5xFAD mice, which develop human-like Aβ plaques due to overexpression of mutant APP and PS1 genes [61]. Aβ plaque load was quantified in the 5xFAD/MSUT2 KO (5xKO) mice and 5xFAD mice with normal MSUT2 expression mice (5xWT) by quantifying the area occupancy of Aβ pathology and the number of plaques in the mouse brains (Fig. 3g). No significant difference was found in either measure between the 5xKO and 5xWT mice at 8 months of age (Fig. 3h, i). To test if MSUT2 is involved in α-syn-related pathways, we injected mouse α-synuclein (α-syn) preformed fibrils (mSyn-pffs) into the MSUT2 KO mice and wild-type littermates and evaluated α-syn pathology immunohistochemically using an anti-phospho-α-syn (EP1536Y) antibody at 1 m.p.i. Again, we found no significant changes in α-syn pathology in MSUT2 KO mice compared to wild-type littermates (Fig. 3j, k, Fig. S6 k, l). Taken together, these data support the idea that MSUT2 is specifically involved in tau pathogenesis and not in other forms of neurodegenerative proteinopathies.

Although MSUT2 has been previously shown to reduce tau pathology in transgenic mice, and here in human tau seeding mouse models, it is not known whether MSUT2 KO would still suppress tau pathology in the presence of the concurrent Aβ plaques found in AD brain, particularly the neuritic plaque (NP) tau pathology found in the vicinity of Aβ deposits. To investigate this, we injected AD-tau into the 5xKO and 5xWT mouse brains and quantified the amount of tau pathology at 1 m.p.i. Consistent with previous studies, AD-tau was found to induce NP tau pathology instead of NFTs in mice at this time point [33, 85] (Fig. 3l). Interestingly, we found that NP tau was significantly reduced in the AD-tau-injected 5xKO mice compared to the injected 5xWT mice (Fig. 3m), suggesting MSUT2-dependent tau seeding and/or spreading is observed both in the absence and presence of Aβ pathogenesis, with MSUT2 regulating NP, NFT and NT tau pathology.

As aging increases the risk of sporadic tauopathies, we examined whether aging could modulate the effect of MSUT2 KO on tau pathogenesis. MSUT2 KO mice and age-matched wild-type littermates were injected with AD-tau at 12—15 months of age, and the amount of tau pathology was analyzed at 3 m.p.i. (Fig. 3n). Our results reveal that the number of NFTs (Fig. 3o) was significantly reduced in the aged MSUT2 KO mice, and the total amount of tau pathology (NFTs + neuritic tau pathology) showed a non-significant trend toward reduction in aged MSUT2 KO mice (Fig. 3p). Altogether, our data suggest that MSUT2 regulates pathways in tau seeding and/or spreading independent of other proteinopathies, and that the effect of MSUT2 on tau pathology is still observed with advanced age, although it may be somewhat reduced.

MSUT2 regulates pathogenic tau internalization in neurons

To determine the mechanism(s) of how MSUT2 modulates the seeding and spreading of tau pathology, we used antisense oligonucleotides (ASOs) against MSUT2 to knock down (KD) MSUT2 expression in primary mouse neurons. We first validated the effects of six different ASO sequences on MSUT2 protein expression (Fig. S7a) and found that all 6 ASOs can significantly suppress MSUT2 protein expression in wild-type mouse primary neurons (Fig. S7b). When we co-treated the neurons with MSUT2 ASOs and pathogenic protein seeds from AD, CBD, or PSP brains, or mSyn-pffs, we found that tau inclusions, but not α-syn aggregates, were reduced in the ASO-treated neurons compared to PBS- or scrambled sequence control (SCR)-treated neurons (Fig. 4a, b). We also confirmed that the amount of insoluble tau was reduced in the ASO-treated neurons using immunoblots (Fig. S8a, b). These data are consistent with the aforementioned in vivo observations in which loss of MSUT2 expression reduces tau pathology.

Fig. 4

MSUT2 modulates the internalization of pathogenic tau seeds in neurons. a Representative images showing wild-type mouse primary neurons that were immunocytochemically stained with R2295M antibody (mouse tau, green) and DAPI (blue) to assess the amount of mouse tau pathology induced by human tau seeds. Wild-type mouse primary neurons were pretreated with antisense oligonucleotides (ASOs) against MSUT2 or scrambled control ASO (SCR) at DIV2 and human-derived tau seeds or mSyn-pffs were added at DIV7. Cells were extracted with detergent to remove soluble proteins and fixed at DIV21. Scale bar = 250 µm. b Quantification of R2295M immunoreactivity of each condition represented in a. Data were quantified using the fluorescent density x area of occupancy/DAPI count and normalized to the PBS-treated samples (as 100%). *P < 0.05, **P < 0.01, ***P < 0.001, n.s., not significant by two-way ANOVA followed by Tukey’s multiple comparisons test, n = 3 biologic repeats per group. Data were normalized to PBS-treated neurons in each group. Error bars represent the standard deviation. c Representative images showing neurons treated with MSUT2 ASOs (ASOs) or scrambled control ASO (SCR) at DIV2 and treated with pHrodo red dye (PhR)-labeled amplified AD-tau seeds (ADT40P1) at DIV7. PhR dye was detected at 2, 8, 18, and 48 h after the addition of ADT40P1. Scale bar = 5 µm. d Quantification of pHrodo red signal intensities from neurons treated as in c. Data are present as density x area/neuron count (DxA/Neuron count). The dashed line delineates the region corresponding to the neuron’s outline, as indicated by the brightfield channel. **P < 0.01, ***P < 0.001 by two-way ANOVA followed by Tukey’s multiple comparisons test, n = 3 biologic repeats per group. Error bars represent the standard deviation. e Representative images show MSUT2 KO and wild-type mouse primary neurons treated with pHrodo red (PhR)-labeled ADT40P1 at DIV7 and live imaged at 24 h post-treatment time. Scale bar = 25 µm. f Quantification of pHrodo red signal in neurons treated as in e. Data are present as density x area coverage/neuron count and normalized to WT neurons. **P < 0.01 by t-test, wild-type (WT) vs. MSUT2 KO (KO), n = 5 biologic repeats per group. Error bars represent the standard deviation. g Representative images show the uptake of fluorescently labeled tau seeds in MSUT2 KO and wild-type mouse brains in vivo. Mice were stereotactically-injected with pHrodo red-labeled ADT40P1 at 3 months of age. Mouse brains were quickly dissected at 2 d.p.i., sectioned, and imaged using live imaging microscopy. Representative images show the dorsal hippocampal regions of injected mice. Scale bar = 500 µm (overview) and 100 µm (insets). h Quantification of pHrodo red signal density x area of occupancy (DxA) in the hippocampal regions of the ADT40P1-injected MSUT2 KO and wild-type mice in sections stained as in g. **P < 0.01 by t-test, wild-type (WT) vs. MSUT2 KO (KO), n = 5 mice per group. Error bars represent the standard deviation. i Representative images showing wild-type primary neurons that were treated with MSUT2 ASOs (ASOs) or scramble controls (SCR) at DIV2 and treated with bodipy (BDY)-labeled mSyn-pffs (mSyn-pffs), Alexa594 (A594)-labeled transferrin, or fluorescein (FL)-labeled dextran (500 kDa) at DIV7, with imaging 24 h later. Internalized proteins were revealed using live cell microscopy. Extracellular bodipy and fluorescein fluorescence were quenched by 500 mM trypan blue before imaging. Representative images show fluorescent signals for bodipy, Alexa594, and fluorescein as well as fluorescence images merged with bright field images. Scale bar = 5 µm. j–l Quantification of fluorescent signal intensities (DxA) in each condition in neurons stained as in i. n.s., not significant, *P < 0.05 by t-test, MSUT2 ASOs (ASOs) vs. scramble controls (SCR), n = 4–5 biologic repeats per group. Error bars represent the standard deviation

To further examine the effect of MSUT2 on tau seeding and spreading, we adapted a recently developed cell-free amplification method for amplifying AD-tau that allowed for the generation of fluorescent-labeled recombinant tau seeds (Fig. S9a). Neuron-based activity tests showed that AD-tau seeds that were amplified with recombinant T40 tau containing the pH-insensitive hylite488 fluorescent label (ADT40P1-h488) could induce mouse tau pathology like the parent AD-tau seeds (Fig. S9b). Using a previously described protocol [41], we specifically quenched the fluorescent signal coming from extracellular seeds (green) using trypan blue (Fig. S9c), which allows for the visualization of internalized intracellular fluorophore-tagged tau seeds during live imaging. The amount of internalized tau seeds in neurons was quantified based on the integrated intracellular fluorescence as a function of time after seeds addition (Fig. S9d, e), and we observed that the half-life of internalized ADT40P1-h488 was twice as long (111.6 h) as T40 monomers (51.5 h) (Fig. S9e). This indicates that the assembled fibrillar tau has significantly greater resistance to degradation than monomeric tau.

The internalization of ADT40P1-h488 tau seeds was examined in neurons treated with or without ASOs against MSUT2 (Fig. S10a). Intracellular fluorescent signal was significantly decreased in MSUT2 ASO-treated neurons compared to PBS or SCR-treated neurons after 24 h post-treatment time (Fig. S10b), suggesting a global reduction of internalized tau seeds in the MSUT2 KD neurons. Using an ADT40P1 preparation made with a pH-sensitive dye (pHrodo™ red; PhR), we used time-lapse live imaging microscopy to measure the progression of tau seeds into acidic compartments such as late endosomes and lysosomes, and internalized tau seeds could be observed in acidic compartments as early as 8 h post-treatment time. Notably, the amount of internalized tau seeds that progressed to acidic compartments was reduced in the MSUT2 ASOs-treated neurons compared to controls (PBS and SCR), in agreement with the global reduction of pathogenic tau seed uptake (Fig. 4c, d). These phenotypes could also be reproduced using MSUT2 KO mouse primary neurons (Fig. 4e, f) and importantly, in vivo after injection of ADT40P1-PhR into the adult MSUT2 KO mouse brains (Fig. 4g, h). Interestingly, in the latter in vivo studies, the majority of pathogenic tau seeds were taken up by cells located in the CA and dentate gyrus regions (Fig. 4g), confirming the central role of neurons in tau uptake. To reveal the endocytic pathways that MSUT2 regulates, we employed different macromolecules that enter neurons via differential endocytic routes in primary neurons (Fig. 4i), including mSyn-pffs (receptor-mediated endocytosis [11, 54] and macropinocytosis [4]), transferrin (clathrin-mediated endocytosis [77]), and dextran (100 kDa; macropinocytosis [50]) and measured their uptakes in the primary neurons treated with MSUT2 or control ASOs. We did not observe any significant change in the internalization of these other protein species (Fig. 4j, k) except for dextran (Fig.