Mouse model
A mutation in the MAPT gene that causes a P301S substitution in tau results in the early onset of familial frontotemporal dementia with parkinsonism linked to chromosome 17 (33, 34). The PS19 line expresses human P301S-tau under the mouse prion protein (Prnp) promoter and shows progressively decreased synaptophysin immunoreactivity accompanied by microgliosis in the hippocampus starting at 3 months of age, followed by reduced LTP in the hippocampal CA1 region at 6 months of age. At this age, the mice develop NFT-like inclusions in the brainstem and spinal cord, whereas neuronal loss typically starts at 9 months of age (28). Behaviorally, PS19 mice display progressive, age-associated motor deficits, including clasping and limb retraction when lifted by the tail, and cognitive deficits including increased hyperactivity and decreased anxiety-like behavior (disinhibition) (35).
Experimental design and safety assessment of CLR01 treatment
Previously, CLR01 was used to treat different animal models of proteinopathy using intracerebroventribular (32, 36) or subcutaneous administration. In the latter route, the compound was administered either via osmotic minipumps (20, 25, 36, 37), as was done here, or by daily injection, 2–7 days a week (38–41). Here, we chose to use the less labor-intensive osmotic-minipump subcutaneous administration to answer the specific question whether CLR01 treatment affected tau directly.
In all of the previous studies, in which CLR01 was administered at doses ranging from 0.04 to 5.0 mg/Kg per day, the treatment was found to be safe, and treatment-related adverse effects were not observed. Moreover, safety studies in wild-type mice showed that when administered as a single intraperitoneal injection at 100 mg/Kg, CLR01 caused liver injury but did not affect other organs and did not reach lethal levels (42). Chronic administration of the compound to wild-type mice for 30 days at 10 mg/Kg caused a ~ 40% reduction in blood cholesterol as the only statistically meaningful finding (42), suggesting the CLR01 had a high safety margin. Nonetheless, because safety and toxicity may vary in transgenic mouse lines, we assessed the P301S-tau mice for mortality, morbidity, behavioral changes, and weight changes that could signal potential toxicity at the doses used here—0.3 and 1.0 mg/Kg per day.
Most of the mice included in the study appeared and behaved normally. One P301S-tau female mouse receiving 1.0 mg/Kg per day CLR01 showed limited movement, appeared dehydrated, and died shortly after the beginning of the treatment from what appeared to be a particularly aggressive course of the disease, not related to the treatment. Two male wild-type mice receiving 1.0 mg/Kg per day CLR01 also died prematurely for unknown causes, one before completing the treatment and the other after completing the treatment, just before it was to be euthanized. As these were wild-type mice, their deaths were not related to an interaction of the compound with the transgene. The transgenic mice receiving 1.0 mg/Kg CLR01 did not show a change in behavior, morbidity, or mortality for the duration of the treatment.
On average, the weight of the P301S-tau mice was ~ 10% lower than that of their wild-type littermates. The differences were more pronounced in the males compared to the females at all time points (Supplementary Figure S1) but there were no significant differences among the treatment groups, supporting the safety of the compound.
CLR01 treatment protects P301S-tau mice from deterioration of muscle strength
Muscle-strength deficits are early signs of disease in the PS19 mouse model starting at ~ 7 months of age (43). To test whether CLR01 treatment protected the mice from such deficits, we measured the muscle strength of the mice in the beginning, middle, and end of the treatment period using the grip-strength test (30). After 17/18 days of treatment when the mice were 6.5–7-months old, the P301S-tau mice receiving vehicle showed a small decrease in their latency to fall from 81 ± 56 to 68 ± 41 s (Fig. 2A, p = 0.242, repeated-measure one-way ANOVA with post hoc Tukey test). During the second half of the treatment period, the mice deteriorated substantially and their latency to fall at the end of the treatment was 30 ± 21 (p = 0.006 compared to the start of the treatment). The deterioration was stronger in female mice (Fig. 2B) than in the males (Fig. 2C), possibly because the males’ latency to fall was substantially shorter than that of the females already at the beginning of the treatment due to their heavier weight.
CLR01 treatment prevented the deterioration in muscle strength in both treatment groups (Fig. 2D). Although in both groups the latency to fall decreased gradually from the beginning to the end of the treatment period, the differences were much smaller than in the vehicle-treated mice and the combination of their small magnitude and associated p-values suggested that these differences were not statistically meaningful (Supplementary Figure S2A, D). The differences between male and female mice were similar to those in the vehicle-treated group yet in both sexes and both treatment groups, the mice appeared to be protected from the decline in muscle strength (Supplementary Figure S2B, C, E, F). There were also no meaningful changes during the treatment in the muscle strength of wild-type mice receiving 1.0 mg/Kg CLR01 per day compared to the group receiving vehicle (Fig. 2D and Supplementary Figure S3), further demonstrating that the treatment did not adversely affect the mice.
CLR01 protects P301S-tau mice from disinhibition-like behavior
Disinhibition is a prominent behavioral feature of the behavioral variant of FTD (bvFTD) and often presents also in patients with AD as increased aggression, hyperactivity, and socially intrusive behavior (44–46). An early-onset, progressive disinhibition-like behavior has been reported in the P301S-tau mice together with early NFT pathology in the CA3 region, resembling clinical presentations and neuropathology of bvFTD (47).
The open-field test is a widely used method for assessing rodent behavior. Rodents show aversion to large, brightly lit, open, and unknown environments and perceive these types of environments as potentially dangerous. To determine whether the P301S-tau mice displayed genotype-related deficits, particularly disinhibition-like behavior, and whether CLR01 treatment could protect them from such deficits, we tested the mice in the open field-arena before the beginning of the treatment and on the last day of the treatment, and recorded their exploration pattern and locomotion activity continuously over 5 min.
As the mice used in our study were relatively young and the open-field test is not physically or mentally challenging, some analyses did not show differences between the P301S-tau and wild-type mice, either before or at the end of the treatment, including overt anxiety-like behavior manifested as the time in the center or the ratio of time spent in the center relative to the periphery. Several measurements, including average speed, latency to enter the center zone, number of entries into the center, time spent in the center, number of line crossings, and path efficiency increased or decreased substantially between the two measurements, but these changes were consistent across all the groups, regardless of genotype or treatment (data not shown).
The time spent mobile (before treatment – 205 ± 36 s, after treatment – 156 ± 49 s, Supplementary Figure S4A) and total distance traveled (before treatment – 12.1 ± 0.6 m, after treatment – 8.4 ± 1.5 m, Supplementary Figure S4B) decreased significantly between the two time points, likely reflecting the aging of the mice. This decrease was smaller in the P301S-tau mice than in the wild-type mice, so at the end of the treatment, the transgenic mice were ~ 35% more active in both measurements (Supplementary Figure S4, p = 0.001 for time mobile, p = 0.002 for distance traveled), suggesting that they were hyperactive (43). CLR01 treatment had little effect on this phenotype (Supplementary Figure S4).
Freezing episodes during the open-field test may represent difficulty with movement and/or anxiety experienced by the mice. Because the mobility of P301S-tau mice at the end of the treatment increased relative to their wild-type littermates, we ruled out increased freezing due to movement difficulties and interpreted such observations as indicating anxiety. Comparing the number of freezing episodes at the beginning and end of the treatment, the wild-type mice treated with vehicle or 1.0 mg/Kg per day exhibited a substantial increase from 3.4 ± 3.0 to 10.2 ± 6.8, p = 0.001, and from 3.3 ± 3.1 to 11.9 ± 6.2, p = 0.0002, respectively. In contrast, the P301S-tau mice receiving vehicle showed little change in freezing episodes, from 6.0 ± 8.1 to 6.9 ± 3.8, p = 0.991 (Fig. 3). This lack of change was interpreted as a disinhibition-like behavior akin to the symptoms of bvFTD (48). Notably, one mouse in this group was an outlier, freezing 34 times during the test before the beginning of the treatment. When this outlier point was removed, the average number of freezing episodes in this group was 4.1 ± 3.2 and the p-value for the difference between the measurements before and after the treatment decreased to 0.419. The change in freezing episodes without the outlier was still substantially smaller in the vehicle-treated transgenic mice than in the wild-type mice. CLR01 treatment protected the mice from the disinhibition-like phenotype. The mice treated with 0.3 or 1.0 mg/Kg per day showed increases in freezing episodes between the beginning and end of the treatment from 3.8 ± 2.3 to 9.8 ± 7.7, p = 0.009 and from 2.9 ± 1.5 to 9.1 ± 5.7, p = 0.004, respectively (Fig. 3).
The wild-type mice also showed a mild increase in grooming episodes between testing in the beginning and end of the treatment, a behavior suggesting discomfort. This change was not observed in P301S-tau mice treated with vehicle, but the P301S-tau mice treated with CLR01 showed a dose-dependent trend toward the difference seen in the wild-type mice (Supplementary Figure S5).
Assessment of total tau in CNS-derived exosomes as a biomarker for treatment effect
Analysis of biomarkers in CNS-derived exosomes isolated from the blood is a new approach that offers a minimally invasive, unique window into biochemical changes in the brain (49, 50). This methodology has been applied to the measurement of tau in neuronal exosomes from patients with AD, FTD, and other diseases (51–54). However, how drugs targeting tau aggregation, such as CLR01, affect the concentration of tau in such exosomes is not known. To address this question, we collected the blood of the mice at sacrifice, separated the serum, and enriched neuronal exosomes from the serum by immunoprecipitation using magnetic beads conjugated to a monoclonal antibody against the neuronal marker L1CAM (50). After lysis of the exosomes, the concentration of total tau was measured using ELISA. The small volume of blood that could be collected from each mouse (100–150 µL) did not allow measurement of additional biomarkers. The measurement showed a dose-dependent increase from 1.8 ± 2.0 pg/mL in the vehicle-treated mice to 3.2 ± 2.9 and 5.9 ± 6.7 pg/mL in the low- and high-dose CLR01 groups, respectively (Fig. 4). The high variability observed precluded making definite conclusions regarding the utility of this assay for assessment of treatment effect, yet the data suggested that measurement of tau in CNS-derived exosomes could be useful in human studies, where larger volumes of blood are easily attainable, and possibly using higher-sensitivity methods for biomarker measurement than the ELISA we used here.
CLR01 reduces hyperphosphorylated tau in the hippocampus
After completing the behavioral tests and sacrificing the animals, we analyzed their brain pathology using several histological and immunohistochemical stains. Tau hyperphosphorylation in the hippocampus was analyzed using monoclonal antibody AT8, which detects specifically phosphorylation at Ser202 and Thr205, two major sites of tau phosphorylation in paired helical filaments in the AD brain (55, 56). The vehicle-treated transgenic mice showed abundant AT8 staining in the hippocampus, primarily in the CA3 region, covering 11.2 ± 6.7% of the area (Fig. 5A,D). CLR01 treatment reduced the staining to 5.1 ± 3.7% (p = 0.0156) and 5.0 ± 3.8% (p = 0.0127) in the 0.3- and 1.0-mg/Kg per day groups (Figs. 4B-D), suggesting that a maximal effect was reached already at the low dose. Analysis by sex showed that the results were driven mainly by the male mice, which showed substantially higher AT8 reactivity in the hippocampus compared to the female mice (Supplementary Figure S6).
CLR01 treatment reduces tau aggregates in the CA3 region
In view of the predominant hyperphosphorylated tau pathology observed in the CA3 region, subsequent histological analyses focused on this region. Gallyas silver-staining is a classical histologic method for identification of pathological tau deposits (57). Black or dark-brown cells indicate a large amount of protein deposition, whereas yellow/light brown cells are healthy and devoid of abnormal protein deposition. Our analysis showed abundant black deposits in the CA3 region of vehicle-treated P301S-tau mice, whose morphology suggested that they were dead cells filled with tau deposits or “ghost” protein deposits left behind after the cells died (Fig. 6A). In contrast, in the CA3 region of mice treated with 0.3 (Fig. 6B) or 1.0 (Fig. 6C) mg/Kg per day CLR01, most of the cells appeared to be healthy and tau deposition was observed only in a few cells (Fig. 6B, C, arrows). We quantified the data as the number of cells containing aggregated tau per mm2. The vehicle-treated group had 477 ± 244 of these cells per mm2 (Fig. 6D), whereas the low-dose group had 126 ± 62 cells per mm2 (p = 0.0001 compared to the vehicle-treated mice), and the high-dose group had 171 ± 68 aggregated-tau-containing cells per mm2 (p = 0.0005 compared to the vehicle-treated mice), demonstrating again that the full effect of the treatment on aggregated tau deposition was achieved already at 0.3 mg/Kg per day CLR01. Analysis of male and female mice separately showed again that the main difference was in the males. This was because the vehicle-treated males had substantially a higher number of aggregated-tau-containing cells in the CA3 (651 ± 176) compared to the females (303 ± 80, p < 0.0001, Supplementary Figure S7).
In view of the silver-stain data presented above, we asked if overt neurodegeneration might be observed in the mice. Previously, neurodegeneration in the PS19 mouse model has been reported at 9 months of age (58) and the mice used in our study were sacrificed at age 7–7.5 months. Nonetheless, we reasoned that if early signs of neurodegeneration could be observed, they would allow assessing a potential protective effect of CLR01 treatment against neurodegeneration. For this assessment, brain sections were stained using a monoclonal antibody against the neuronal nuclear marker NeuN. The analysis showed neurodegeneration in the CA3 region of two mice in the vehicle-treated group, which was accompanied by apparent astrogliosis (Supplementary Figure S8A), yet other mice in this group did not show neurodegeneration, likely due to the relatively young age of the mice. Neurodegeneration was not observed in either of the treatment groups (Supplementary Figure S8b, C), yet the lack of an overt phenotype in the vehicle-treated group precluded meaningful analysis of treatment effect (Supplementary Figure S9D.)
We also analyzed synapse density in the CA3 region using monoclonal antibodies against the presynaptic marker Bassoon and postsynaptic marker Homer. The analysis showed that the P301S-tau mice had ~ 50% reduction in synapse density in the CA3, which was only marginally affected by the treatment (Supplementary Figure S9). Impaired synaptic function and hippocampal synapse loss have been detected in the P301S mouse as early as 3 months of age (28), 3–3.5 months before the beginning of the treatment, at which point substantial synapse loss was already present and apparently could not be reversed by the treatment.
CLR01 treatment effect on hyperphosphorylated tau in different brain regions of P301S-tau mice
To test whether the CLR01 treatment affected the levels of total and hyperphosphorylated tau in different brain regions of the treated P301S-tau mice, we compared the fraction of total tau that was phosphorylated at pT231, each measured by ELISA, among the treatment groups. Total tau was normalized to the total protein in each sample. Tau contains 85 potential phosphorylation sites but not all of them become phosphorylated in disease and contribute to the pathology. We measured pT231-tau as phosphorylation at this site is associated with most tauopathies and decreases the ability of tau to stabilize microtubules (59).
Measurement of total tau in the buffer-soluble fraction showed that the concentration was slightly higher in the CLR01-treated groups compared to the vehicle group, yet the differences were small and in almost all cases statistically insignificant (Fig. 7A). Similar small, insignificant changes were observed in the membrane-bound (Fig. 7C) and insoluble (Fig. 7E) fractions, in which the concentration of tau was substantially lower than in the buffer-soluble fractions.
The fraction of pT231-tau in total tau in the soluble fraction decreased in all the brain regions analyzed in a dose-dependent manner, except in the midbrain/brainstem extract where the decrease in the low-dose group was more pronounced than in the high-dose group (Fig. 7B), reflecting high experimental variability in these measurements. Interestingly, in the membrane-bound fraction of the cortex, hippocampus, and cerebellum, but not the midbrain/brainstem, we found a dose-dependent increase in the fraction of pT231-tau in total tau (Fig. 7D). A similar increase was found in the insoluble fraction of the hippocampus, though it appeared to be driven by a few samples in which the pT231-tau fraction in total tau was 1–2-orders of magnitude higher than in most other samples (Fig. 7F) and is therefore difficult to interpret.
CLR01 reduces tau oligomers in the hippocampus dose-dependently
Tau oligomers often are considered the most toxic form of the protein (60, 61). Mouse studies have shown that memory loss and synapse loss correlated better with oligomer concentration levels than with the amount or spreading of NFTs (62, 63). In view of these data, we tested whether CLR01 treatment affected tau-oligomer levels using dot-blots and native-PAGE/western-blot analyses in the soluble fraction of the brain extracts. The other two fractions were not used because the detergent or chaotrope in their buffers likely alter the oligomer composition. We probed the dot blots using monoclonal antibody tau oligomeric complex-1 (TOC-1) (64). We did not detect any TOC-1 signal in the cortex, cerebellum, or midbrain/brainstem regions even when the total protein amount loaded was 9 µg, three times higher than in the hippocampus (data not shown). In contrast, a strong TOC-1 signal was detected in the hippocampus extracts when 3 µg of total protein were loaded, suggesting that TOC-1-positive tau oligomers were present primarily or only in the hippocampus. The analysis showed a dose-dependent decrease in the TOC-1 signal (Fig. 8A). Densitometric quantitation of the dot blots showed a relatively high degree of variability within each treatment group. The densitometric signal in the vehicle-treated mice was 10,871 ± 3,664 and was reduced to 8,475 ± 4,863 (p = 0.375 compared to the vehicle-treated group) and 6,182 ± 4,082 (p = 0.005) in the 0.3 and 1.0 mg/Kg per day CLR01 groups, respectively (Fig. 8B). Analysis of male and female mice separately showed similar trends in both sexes (Supplementary Figure S10A).
We also attempted to quantify individual tau oligomers by native-PAGE/western blots probed using the anti-human tau monoclonal antibody HT7. However, high variability among the mice yielded little differences among the groups and the results were not statistically meaningful (p-value range 0.107–0.996). An example of the analysis in the hippocampus is shown in Supplementary Figure S10B. Analyses in other brain regions yielded similar results (not shown).
CLR01 reduces the concentration of tau seeds in the hippocampal soluble fraction
Intercellular propagation of proteopathic protein seeds is thought to be a major mechanism by which tau pathology propagates in the brain in a prion-like manner (65–67). Therefore, we asked if the reduction of aggregated, hyperphosphorylated, and oligomeric tau observed in the P301S-tau mice treated with CLR01 correlated with a reduction in the concentration of tau seeds. We addressed this question by measuring seeding activity in the soluble and insoluble fractions using a HEK293 biosensor cell line that expresses stably the 4R-tau repeat domain containing a P301S substitution and conjugated to CFP or YFP (68). The detergent present in the membrane-bound fraction is toxic to the cells, preventing the use of this fraction in the assay. In each experiment, the cells were imaged by fluorescence microscopy using the fluorescence of YFP to obtain a qualitative assessment of the seeding response and the number of seeds was quantified by measuring the FRET signal between CFP and YFP using flow cytometry, as described previously (26).
Examples of the fluorescence-microscopy images are shown for the soluble fraction of the hippocampus. In the absence of seeds, only diffuse fluorescence was observed (Fig. 9A), whereas in the presence of an extract from a vehicle-treated mouse, abundant bright puncta were apparent, indicating the formation of intracellular tau aggregates (Fig. 9B). The aggregates were substantially decreased in the presence of extracts from mice treated with 0.3- or 1.0-mg/Kg per day CLR01 (Figs. 8C, D, respectively).
The FRET-based flow-cytometry analysis showed a marked decrease in the seeding activity (expressed as integrated FRET density) of hippocampal extracts from 133 ± 114 in the vehicle-treated P301S-tau mice to 59 ± 90 (p = 0.053) and 45 ± 48 (p = 0.016) in the low- and high-dose groups, respectively (Fig. 9E). Both the male and the female mice contributed to these differences (Supplementary Figure S11A). To further explore whether the presence of tau seeds in the brain of the mice had a deleterious effect, we asked whether the seeding correlated with the ratio between the grip-strength at the end versus the beginning of the treatment. Spearman correlation showed an increase in the seeding response of the hippocampal extract correlated with a stronger decline in grip-strength (Fig. 9F, r = -0.258, p = 0.074). Stronger correlations were found in extracts of the cerebellum (Supplementary Fig. 11F) and midbrain/brainstem (Supplementary Fig. 11I), which are important brain regions for control of movement, but not in the cortex (Supplementary Fig. 11I). The correlation was strong also for all the brain regions combined (Fig. 9G, r = -0.314, p = 0.028). In contrast to the impact of the treatment on the seeds in the soluble fraction, the treatment had no apparent effect on the amount of seeds in the insoluble fraction (Fig. 9H).
Despite the contribution of other brain regions to the increased correlation between the seeds and the grip-strength deterioration (Fig. 9G), the seeding response of the soluble fraction of the cortex (Supplementary Figure S11B), cerebellum (Supplementary Figure S11E), and midbrain/brainstem (Supplementary Figure S11H), which was substantially lower than that of the hippocampus, was largely unaffected by the treatment. The seeding responses of the insoluble fractions of the same regions were higher than those of the soluble fractions but were not affected by the treatment (Supplementary Figure S11D, G, J).
CLR01 reduces microgliosis in P301S-tau mice
Glial cells, including microglia and astrocytes, are major contributors to the neuropathological processes, neuroinflammation, and spreading of pathology in tauopathies (69, 70). To assess the impact of CLR01 treatment on glial involvement in the developing neuropathology in the P301S-tau mouse brain, we stained the brains using antibodies against the microglial and astrocytic markers ionized calcium-binding adaptor molecule 1 (Iba-1) and glial fibrillary acidic protein (GFAP), respectively. Compared to wild-type mice treated with vehicle (2.7 ± 2.1% of the CA3 area, Fig. 10A, E) or 1.0 mg/Kg per day CLR01 (2.8 ± 1.3%, Fig. 10B, E), microglia were found to be densely populated in CA3 region of the vehicle-treated P301S-tau mice (16.3 ± 9.9% Fig. 10C, E, p = 0.0006 compared to vehicle-treated wild-type mice). The density was substantially reduced in the low- (10.3 ± 4.7%, p = 0.054 compared to vehicle-treated P301S-tau mice, Fig. 10C, E) and high-dose (7.9 ± 3.8%, p = 0.002, Fig. 10D, E) CLR01 treatment groups. The treatment effect was driven primarily by the male mice, whereas in the females, a similar trend was observed but the differences among the groups were much smaller (Supplementary Figure S12A).
A peculiar phenomenon we observed in the cortex of two female mice in the vehicle-treated P301S-tau group while analyzing the microglia was clustering of the cells at apparently random spots (Supplementary Figure S12B). The cells appeared much larger than neighboring microglia and their focal point did not appear to be hyperphosphorylated tau.
In contrast to the substantial differences in microglia density, the differences in astrocyte density between the wild-type and transgenic mice and the impact of CLR01 on the astrocyte density showed a similar trend but were of smaller magnitude and statistically insignificant (Supplementary Figure S13).