Abstract
In Alzheimer’s disease, the distribution of neurofibrillary tangles, a histological hallmark comprised of phosphorylated forms of the protein tau, follows a distinct pattern through anatomically connected brain regions. The well-documented correlation between the severity of tau pathology and disease progression implies a prion-like seeding and spreading mechanism for tau. Experimentally, this has been addressed in transgenic mice by the injection of protein lysates isolated from brains of transgenic mice or patients with tauopathies, including AD, that were shown to behave like seeds, accelerating tau pathology and tangle formation in predisposed mice. More specifically, in vivo data suggest that brain lysates from mice harboring the P301S mutation of tau can seed protein aggregation when injected into the hippocampi of human wild-type tau transgenic ALZ17 mice. Here, we compared the seeding potential of lysates and extracellular vesicles enriched for exosomes (EVs) from wild-type and human P301L tau transgenic rTg4510 mouse brains. We show that transgenic EVs cause increased tau phosphorylation and soluble oligomer formation in a manner comparable to that of freely available proteins in brain lysates, a prerequisite for the formation of mature protein aggregates.
INTRODUCTION
The correlative relationship between the severity of Alzheimer’s disease (AD) and the degree of tau pathology has been well established both postmortem and in vivo, with the Braak staging, based on phosphorylation at the AT8 phospho-epitope of tau, becoming an integral diagnostic tool [1–3]. It has been proposed that this stereotypical spread of pathology reflects the ability of tau to transfer between anatomically connected neurons and corrupt proteins in previously healthy neurons in a ‘prion-like’ manner [4]. It has also been shown that whole brain lysates from mice containing the familial frontotemporal dementia mutation P301S can accelerate tau phosphorylation and seed neurofibrillary tangle (NFT) formation in ALZ17 mice, a human wild-type tau expressing strain that has not been reported to develop end-stage aggregates [5, 6]. Tau is known to be integral to this process, as immuno-depletion of tau from lysates decreases seeding efficacy [5, 8]. Furthermore, injection of synthetic tau fibrils alone is sufficient to induce NFT-like transmission [9, 10]. What remains to be clarified in vivo is whether tau seeding is specific to the uptake of freely available aggregates or whether it can also involve tau contained in extracellular vesicles.
The term extracellular vesicle encompasses various vesicular bodies ranging from exosomes with a diameter of 30–130 nm, up to apoptotic bodies with a diameter of 1,000–5,000 nm [11, 12]. Exosomes have been specifically implicated in AD based on the identification of phospho-tau species within exosomes derived from patient blood and cerebrospinal fluid [13, 14]. Furthermore, it has been shown that inhibition of exosome synthesis significantly reduces tau propagation [15]. However, it can be difficult to distinguish roles for secreted tau in vivo due to difficulties in distinguishing the origins of NFTs in the extracellular space [16]. Whether freely detected NFTs are representative of an active release, a passive release via cell death, or a post-secretion release from within vesicular bodies is difficult to determine [17].
To dissociate the role of freely available extracellular tau from that which is packaged into vesicles it is possible to isolate specific extracellular vesicle fractions from brain tissue [18]. Our previous work identified an exosome-enriched extracellular fraction of extracellular vesicles (EV) in P301L mutant tau transgenic rTg4510 mice that contained an increased level of tau compared to wild-type, and revealed specific hyperphosphorylation of tau at the AT270, Ser422, and Ser262 epitopes [19]. In a functional assay using FRET tau biosensor cells, we were able to show that rTg4510-derived EVs seeded tau aggregation in a threshold-dependent manner [19]. Furthermore, cells treated with transgenic EVs released EVs which were capable of seeding aggregation when applied to naïve biosensor cells, suggesting pathology may be propagated via exosomes when cells are undergoing aggregation [19]. To clarify the specific role of exosome-derived tau and to determine whether tau transgenic EVs are capable of seeding aggregation in vivo, EV fractions isolated from wild-type and rTg4510 mice were injected into 3 month old ALZ17 mice and their tau pathology was assessed 6 months later.
MATERIALS AND METHODS
Experimental design
Whole brain lysates and EVs were isolated from 4–6 month old wild-type and rTg4510 mice as described previously [19]. Each experimental group contained five ALZ17 mice of mixed gender. Three non-injected controls (NIC) were used for comparison (Fig. 1A). In the figures, ‘exosomes’ is used to indicate EVs. Samples were concentration-matched to 1μg/μl, and 2.5μl were injected bilaterally into the hippocampus (Bregma co-ordinates: anterior/posterior, –2.5 mm; medial/lateral±2.0 mm; dorsal/ventral, –1.8 mm) of 3 month old ALZ17 mice and aged for 6 months (Fig. 1A). Tau pathology was assessed using the phospho-specific AT8 antibody (reactive with tau phosphorylated at S202 and T205), which targets a pathological phosphorylation epitope. Phosphorylation was quantified as percentage area occupied in the hippocampus and as AT8-positive cell counts, and was performed blinded to the experimental condition. Early aggregation events were probed for using the T22 antibody, which is specific to an oligomeric confirmation [20], and mature NFT formation was investigated using the traditional silver staining techniques Gallyas and Bielschowsky. As positive controls, aged Tau58 mice carrying the FTD P301S mutation were used that have well characterized NFT formation [21, 22]. Western blotting of sucrose fractionations to isolate the EVs used in this study is consistent with previously published work [19], confirming that fraction 3 (F3) is the most enriched in both exosome markers and tau (Fig. 1B).
Preparation of the lysate and the extracellular vesicle (EV) fractions
Brains were dissected and the cerebellum and olfactory bulbs were removed. For whole brain lysate, brains were dounce homogenized in 10% w/v phosphate-buffered saline (PBS) containing 1x Complete protease inhibitors (Roche), 25 mM NaF and 20 mM NaVO4, and then briefly sonicated (6×10 s at 50% amplitude). Brain lysates were centrifuged at 3,000 g for 10 min at 4°C. Supernatants were collected and stored at –80°C until use.
For vesicle fractions, the tissue was processed following an established protocol [18] with minor modifications as described [19]. Briefly, to prepare the vesicle fractions, following dissection, the tissue was manually chopped and incubated in 7 ml of 20 units/ml papain (LS003119, Worthington) in Hibernate-A for 20 min at 37°C. To stop the reaction, 14 ml of ice-cold Hibernate-A with 1x Complete protease inhibitor (Roche), 50 mM NaF, 200 nM Na3VO4, and 10 nM E-64 inhibitor (E3132, Sigma) was added to the tissue, keeping the tissue at 4°C throughout the subsequent steps. The tissue was then disrupted by pipetting up and down with a 10 ml pipette. Cell contamination was cleared by centrifugation at 300 g for 10 min and discarding the pellet. The supernatant was then cleared of large membranous debris by centrifugation at 2,000 g for 10 min and discarding the pellet. For EV purification, the 10,000 g supernatant was syringe filtered at 0.22μm and centrifuged at 100,000 g for 70 min to pellet EVs. The 100,000 g pellet containing EVs plus contaminating proteins was resuspended in 2 ml of 0.95 M sucrose in 20 mM HEPES (15630-080, Life Technologies) before addition to a sucrose-step gradient column. For EV purification, the column consisted of 6×2 ml fraction running from the top 0.25 M, 0.6 M, 0.95 M, 1.3 M, 1.65 M, and 2.0 M at the bottom. Similarly, sucrose step gradients were centrifuged for 16 h at 200,000 g, after which the six fractions were collected. EVs settled typically at 0.95 M sucrose (fraction 3). The protein content for both lysates and vesicle fractions was quantified using the DC protein assay (Bio-Rad). To concentration match lysates and EVs, the lysates were diluted to equal EV fractions at 1μg/ml.
Western blotting
For western blotting, 20μg of protein from brain-derived EV fractions and whole cell extracts were separated by 10% SDS-PAGE electrophoresis and transferred onto PVDF membranes by the Trans-Blot Turbo transfer system (Bio-Rad). Membranes were blocked for 1 h at room temperature in Odyssey blocking buffer (Li-cor) and incubated overnight in primary antibodies. Antibodies used to probe the EV fractions included anti-total tau Tau5 (1:1,000, Millipore), AIP1/Alix (1:1,000, Millipore), and Flot1 (1:500, Santa Cruz). Membranes were washed 3 x in Tris-buffered saline/0.1% Tween-20 (TBST) followed by incubation with a HRP-coupled secondary antibody (1:10,000) for 1 h at room temperature. Membranes were washed a final 3 x in TBST and imaged on the Odyssey Fc imaging system (Li-Cor).
Stereotaxic surgery
Mice were anaesthetized with isoflurane vaporized in oxygen. The head was shaved and depilated before being placed securely in ear bars. Samples were concentration-matched to 1μg/μl, and 2.5μl of the sample was injected bilaterally into the hippocampus (Bregma co-ordinates: anterior/posterior, –2.5 mm; medial/lateral±2.0 mm; dorsal/ventral, –1.8 mm) of 3-month-old ALZ17 mice at an infusion rate of 0.25μl/min, after which mice were aged for 6 months. The same volume of the dye Evans Blue was injected in the same manner to confirm the injection site. For histology, mice were anaesthetized with a lethal dose of pentabarbitone before transcardial perfusion with 30 ml PBS followed by 30 ml 4% paraformaldehyde. Brains were removed from the skull and post-fixed overnight at 4°C.
Histology
Histology and neuropathological quantification of tau phosphorylation were carried out as previously described [23], and all quantification was conducted blinded to the treatment. Briefly, using ImageJ software, the hippocampus was selected as a region of interest and immunoreactivity was quantified as a percentage area. The threshold for positive labeling was determined using the normal distribution of immunoreactivity from the control group and set at three standard deviations from the mean. AT8-positive cell counts were performed in the CA1 region from 2 sections per animal, 14μm apart, at the level of the injection site. AT8 (Millipore) was used at 1:500 dilution, T22 (Millipore) was used at 1:1,000 for immunohistochemistry and 1:500 for fluorescence, Tau5 (Thermo Fisher Scientific) was used at 1:2,000 for immunohistochemistry and 1:1,000 for fluorescence, and fluorescence analyses with AT8 and human tau-specific HT7 (Thermo Fisher Scientific, 1:1,000) were performed as previously described [21]. Gallyas and Bielschowksy silver staining were performed following well-established methods with aged P301S mutant tau transgenic Tau58 mice being used as a positive control for NFT formation [21, 22].
RESULTS
Transgenic EVs increase pathological tau phosphorylation in the hippocampus
To establish whether EVs isolated from rTg4510 mice can accelerate tauopathy, we investigated hyperphosphorylation of tau at the pathogenic phospho-epitope AT8 (Fig. 2A). The percentage area occupied by AT8-immunoreactivity in the hippocampus was significantly increased following injection with transgenic lysate (Fig. 2Avi). This was also the case for rTg4510 EVs, although the response was more variable. In age-matched NIC, AT8-positivity was observed in both the dentate gyrus and the CA3 region of the hippocampus. To control for variation in these regions, the CA1 region as the target of the injection was assessed specifically by calculating the number of phospho-tau-positive cells within this cell layer (Fig. 2B). Both the rTg4510 lysate and rTg4510 EVs significantly increased the number of phospho-tau-positive neurons in the CA1 region (Fig. 2Bvi). No significant differences were observed in hippocampal area, CA1 area or number of CA1 nuclei between treatment groups, indicating comparable sample populations within the dataset with no significant toxicity in the CA1 region as a result of the injections (Fig. 2Bvii-ix). In addition, there was no change in total tau levels in the hippocampus, as probed for by Tau5, an antibody capable of picking up both mouse and human tau in a phosphorylation-independent manner (Fig. 3).
Transgenic EVs seed aggregation of endogenous tau into a soluble oligomeric state
To address whether rTg4510 EVs are able to increase pre-NFT tau species, we used the oligomeric antibody T22 to detect soluble tau aggregates [20]. Although no immunoreactivity was observed in the CA1 layer, punctate signals were observed in clusters throughout the stratum radiatum, a region of high synaptic connectivity harboring Schaffer collateral fibers from the CA3 to CA1 region (Fig. 4A). The number and size of these T22-positive clusters appeared to increase in both the transgenic lysate- and transgenic EV-injected groups (Fig. 4Aii,v). However, this signal co-localized with neither AT8 nor the human transgenic tau, as detected by the HT7 antibody (Fig. 4Aiii,vi). Interestingly, while HT7 did not show clustered immunoreactivity reminiscent of T22, Tau5, an antibody that detects both mouse and human tau, showed clustered immunoreactivity in a pattern similar to T22, that was restricted predominantly in the stratum radiatum (Fig. 4B). Co-labeling confirmed robust and consistent co-localization of T22 and Tau5 puncta (Fig. 4C), suggesting that the oligomeric tau aggregates seeded by rTg4510 lysate and EVs are composed of endogenous murine tau.
Mature NFTs are not seeded by the transgenic lysate or transgenic EVs
To further explore whether the observed hyperphosphorylation reflected mature NFT formation, we probed the lysate- and EV-injected mice with the Gallyas and Bielschowsky silver staining reagents. However, no NFT-like signal was observed across the treatment groups with either technique (Fig. 5A, B), whereas Tau58 mice that were included as a positive control displayed clear NFTs (Fig. 5Avi-vi, Fig. 2Bvi-vi). This suggests that neither the transgenic lysate nor transgenic EVs seeded mature NFT formation under our treatment conditions. As the injected volume was relatively large, we may have experienced a spill-over through the hippocampus and therefore seeding could potentially have occurred elsewhere in the brain. To investigate an exposure to the injected seeds in the hippocampus, an identical volume of the dye Evans blue was injected following the above experimental parameters. The dye was observed in the CA1 region of the hippocampus but also extended both anteriorly and posteriorly to the injection site (Fig. 5C). To be entirely confident that no seeding had taken place elsewhere, we probed for NFTs throughout the hippocampal structure in areas anterior and posterior to the injection site. However, we did not see any evidence of NFT formation as determined by the silver Gallyas impregnation in any of the treatment groups including the rTg4510 lysate (Fig. 5C).
DISCUSSION
It is widely accepted that specific tau species are able to transmit a pathology through nucleation-driven seeding events, thereby contributing to the spread of disease. Injection of (1) brain lysates taken from both transgenic mice and AD patients, (2) recombinant tau fibrils and (3) human-isolated soluble tau oligomers have all been shown to accelerate tau pathology and seed NFTs [5, 24]. However, these studies did not account for how infective tau species are transmitted in vivo as the injected material was freely available in the extracellular space.
Previously, we have shown that rTg4510 EVs contain tau seeds that are capable of inducing aggregation in vitro [19]. Here, we have demonstrated that rTg4510 EVs can accelerate tau hyperphosphorylation at the pathological epitope AT8 specifically in the CA1 region as the site of injection. As total tau levels were unchanged this would suggest an increased modification of existing tau rather than instigating tau accumulation. Furthermore, the observed increase was comparable to that of the transgenic lysate, suggesting that EVs containing pathological tau are similarly efficacious in accelerating phosphorylation. In addition, both the rTg4510 lysate and rTg4510 EVs increased the number of soluble oligomeric aggregates in the stratum radiatum of the hippocampus. Interestingly, T22 immunoreactivity did not co-localize with either AT8 or when an antibody for total human tau was used, suggesting that the predominant component of these aggregates is endogenous mouse tau. Indeed, an antibody capable of detecting both mouse and human tau demonstrated a similarly clustered staining and co-localization conclusively with T22, indicating that the oligomeric aggregates seeded by rTg4510 lysates and EVs are comprised of murine tau. This remains consistent with previous observations that both endogenous and human tau can be hyperphosphorylated and integrated into aggregates by transgenic tau seeds [25], and that T22 aggregates in regions neighboring the injection site can be comprised of endogenous tau alone [20].
In the current study, mice were aged for 6 months post-injection to 9 months of age, the earliest time point at which NFTs have previously been reported in response to the injection of transgenic lysates [5]. However, we were unable to recapitulate this effect with either the rTg4510 lysate preparation or rTg4510 EVs. In the current study, we chose a protein concentration that matched the total lysate to EVs in order to accurately draw comparisons between the two preparations. Considering that our previous work in the biosensor cell line strongly indicated that there is a concentration threshold necessary to initiate seeding [19], it is plausible that the preparations used in this study, although they were able to accelerate phosphorylation, were not of sufficient concentration to cause NFT formation within the time-frame of our experiment. However, small soluble oligomeric aggregates were observed, which could indicate that early NFT seeding events had been initiated. Previously, a substantial increase in NFT numbers was observed between 6 and 9 months [5]. Therefore, it is possible that both the rTg4510 lysate and rTg4510 EVs could be capable of seeding if the time-frame were extended.
Furthermore, the present study relied on lysate and EV preparations from a P301L transgenic strain, and previously a P301S tau transgenic strain was used, albeit that the age of the animals was similar. Both mutations are point missense mutations which do not effect exon splicing but reduce the ability of tau to bind to microtubules [26]. The specific P301S mutant used previously relies on expression through the common neuronal mouse Thy1 promotor, whereas the rTg4510 instead utilizes the CaMKII-α promotor under the transcriptional control of a tetracycline-regulatable element, resulting in transgene expression predominantly in the forebrain, which can be switched off by doxycycline [27]. Therefore, it is possible that differences in both promoter and mutation may contribute towards reduced seeding propensity of the tau species produced by the rTg4510 model.
In conclusion, we present the first evidence that tau contained within EVs can accelerate tau phosphorylation and soluble aggregate formation in vivo, in a manner comparable to that of freely available proteins in brain lysates. Our data show that rTg4510 EVs are capable of accelerating pathological markers which precede NFT formation, such as tau hyperphosphorylation and the formation of oligomeric aggregates, although no end-stage NFTs themselves were detected. However, this may be due to experimental constraints rather than inherent characteristics of the EV seeds as rTg4510 lysates also failed to seed mature NFTs under our experimental conditions, suggesting that the time-frame given was not sufficient for the development of NFTs. Nonetheless, our data verifies that extracellular vesicles are capable of seeding pathological changes in tau in vivo and therefore may play a role in the spreading of tau pathology in AD.
Footnotes
ACKNOWLEDGMENTS
This study was supported by the Estate of Dr. Clem Jones AO, the State Government of Queensland, and by grants from the Australian Research Council (DP13300101932) and the National Health and Medical Research Council of Australia (GNT1037746, GNT1003150) to J.G. We thank Tishila Palliyaguru, Sevannah Ellis, and Linda Cumner for expert technical assistance, and Trish Hitchcock and her team for animal care, and Rowan Tweedale for critically reading the manuscript.
