Spread of tau down neural circuits precedes synapse and neuronal loss in the rTgTauEC mouse model of early Alzheimer's disease

Abstract Synaptic dysfunction and loss is the strongest pathological correlate of cognitive decline in Alzheimer's disease (AD) with increasing evidence implicating neuropathological tau protein in this process. Despite the knowledge that tau spreads through defined synaptic circuits, it is currently unknown whether synapse loss occurs before the accumulation of tau or as a consequence. To address this, we have used array tomography to examine an rTgTauEC mouse model expressing a P301L human tau transgene and a transgene labeling cytoplasm red (tdTomato) and presynaptic terminals green (Synaptophysin‐EGFP). All transgenes are restricted primarily to the entorhinal cortex using the neuropsin promotor to drive tTA expression. It has previously been shown that rTgTauEC mice exhibit neuronal loss in the entorhinal cortex and synapse density loss in the middle molecular layer (MML) of the dentate gyrus at 24 months of age. Here, we observed the density of tau‐expressing and total presynapses, and the spread of tau into the postsynapse in the MML of 3–6, 9, and 18 month old red–green‐rTgTauEC mice. We observe no loss of synapse density in the MML up to 18 months even in axons expressing tau. Despite the maintenance of synapse density, we see spread of human tau from presynaptic terminals to postsynaptic compartments in the MML at very early ages, indicating that the spread of tau through neural circuits is not due to the degeneration of axon terminals and is an early feature of the disease process.


| I N T R O D U C T I O N
The observation that neurons of the entorhinal cortex (EC) are affected very early by neurofibrillary tangle pathology in Alzheimer's disease (AD) has been recognized for >30 years (Braak & Braak, 1991;Hyman, Van Hoesen, Damasio, & Barnes, 1984). These layer II neurons link the cerebral cortex with the hippocampus via the perforant pathway (Hyman, Kromer, & Van Hoesen, 1987), a critical projection for memory function.
Disruption of this neural circuit through selective loss of neurons, synapses, and the accumulation of tau lesions, is thought to contribute to the early memory impairments observed in AD (Hyman, Van Hoesen, & Damasio, 1990). As the disease progresses, tau pathology propagates from the EC in a well-characterized anatomical pattern extending to limbic and association cortices (Braak & Braak, 1991). The mechanism of this spreading has yet to be determined, however, mounting evidence suggests that tau spreads trans-synaptically and that synaptic activity increases the spread of tau through synapses (de Calignon et al., 2012;Harris et al., 2012;Liu et al., 2012;Pooler, Phillips, Lau, Noble, & Hanger, 2013;Walsh & Selkoe, 2016;Wu et al., 2016). One potential route of tau moving from presynapses to postsynapses is the degradation of This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. presynaptic terminals due to the presence of toxic tau. This would release pathological tau from the degenerating presynaptic terminal which could then be taken up by the postsynapse; a possibility supported by many studies showing that cells in culture can take up extracellular tau (Frost, Jacks, & Diamond, 2009;Guo & Lee, 2011;Kfoury, Holmes, Jiang, Holtzman, & Diamond, 2012;Lewis & Dickson, 2016).
In the present study we directly address the question of whether degeneration of tau expressing presynaptic terminals is necessary for the spread of tau to postsynaptic compartments in a defined neural circuit. To do this, we examine the red-green-rTgTauEC transgenic mouse model, which reversibly express human mutant P301L tau under the control of the neuropsin promoter restricting expression primarily to the medial entorhinal cortex (de Calignon et al., 2012;Polydoro et al., 2013;Pooler, Polydoro, et al., 2013). These neurons overexpressing mutant tau also express Myc-tagged tdTomato and full-length synaptophysin/mut4EGFP fusion protein (EC-tdTomato/Syp-GFP) (Li et al., 2010;Miyamichi et al., 2011). tdTomato expression is cytoplasmic, while GFP expression is directed to the presynapse. This mouse line allows visualization of the synaptic terminals of human tau expressing neurons. Here, we utilized array tomography to determine whether the density of tau and GFP-expressing presynapses is altered prior to global synapse loss. Employing this high resolution imaging technique, we further characterized the spread of human tau protein within this model and demonstrate that the propagation of tau through anatomically connected brain regions is not due to the degeneration of presynaptic terminals and is an early feature of the disease process.

| Array tomography
Fresh brain tissue samples were collected from rTgTauEC 1 EC-tdTomato/Syp-GFP and EC-tdTomato/Syp-GFP control transgenic mice as outlined previously Koffie et al., 2009). Briefly, small tissue blocks containing the dentate gyrus and entorhinal cortex were fixed in 4% paraformaldehyde and 2.5% sucrose in 20 mM phosphate buffered saline pH7.4 (PBS) for 3 hr. Samples were then dehydrated through ascending cold graded ethanol and embedded into LR White resin (EMS) which was allowed to polymerize overnight at 538C. Resin embedded tissue blocks were cut into array ribbons of 70 nm thick sections using an ultracut microtome (Leica) equipped with a Jumbo Histo Diamond Knife (Diatome, Hatfield, PA) and collected onto gelatin coated coverslips.
For immunolabeling of synaptic density, array ribbons were immunostained with primary antibodies against total presynapses (synaptophysin) and presynapses from neurons overexpressing human tau (GFP) and fluorescently labeled secondary antibodies (Table 1). For immunolabeling of pathological tau spread, array ribbons were immunostained with primary antibodies against postsynapses (PSD95) and human tau (Tau13) and fluorescently labeled secondary antibodies (Table 1). Sections were counterstained with .01 mg/ml 4 0 -6diamidino-2-phenylindole (DAPI). In each experiment, a short extra ribbon was used as a no primary negative control. For each area of interest (middle molecular layer of the dentate gyrus), images were obtained on serial sections using a Zeiss axio Imager Z2 epifluorescent microscope at 63X 1.4 NA Plan Apochromat objective with equipped CoolSnap digital camera and AxioImager software with array tomography macros (Carl Zeiss, Ltd, Cambridge, UK).  images were processed and analysed in MATLAB to remove background noise and to calculate the colocalization of Tau13 with postsynapses (a minimum of 50% overlap between Tau13 and PSD95 puncta was required to classify colocalization).

| Statistical analyses 2.3.1 | Array tomography-synaptic density
Total presynaptic density detected by synaptophysin labeling and GFPpositive presynaptic density were normally distributed across crops for each mouse, therefore, the mean total presynapses and mean GFPpositive presynapses were taken for individual mice. Group total presynaptic density and GFP-positive presynaptic density were normally distributed so a two-way ANOVA was performed (SPSS). All values are reported as mean and SEM.

| Array tomography-spread of tau pathology
The % postsynapses colocalising with Tau13 from each mouse was cal- mice at all ages tested, starting at 3 months, which is 21 months prior to the loss of synapses that occur in this line ( Figure 4A,B). Quantification within the middle molecular layer of the dentate gyrus revealed 10% of postsynapses at 3-6 months of age, 8% of postsynapse at 9 months of age and 9% of postsynapses at 18 months of age colocalising with Tau13 ( Figure 4G). Further analysis of putative synaptic pairs (composed of presynapses derived from layer II EC neurons (GFP 1 ve), and postsynapses of the MML within 0.5 lm proximity), revealed a subset of synaptic partners in which both the presynaptic and postsynaptic compartments were tau positive at all ages studied ( Figure 4H).
This suggests that tau protein can spread from intact presynapses to neighboring postsynapses within this model without requiring presynaptic terminal degeneration.

| DISCUSSION
The spread of tau inclusions through the brain occurs in a wellcharacterized hierarchal pattern in AD and plays a role in the disease pathogenesis. Despite accumulating evidence supporting trans-  Harris et al., 2012;Liu et al., 2012), the mechanism underpinning this pathological progression is not fully understood and remains a matter of debate. It has been proposed that degeneration of presynaptic terminals may result in the leakage of tau and subsequent spread to neighboring postsynapses (Wang & Mandelkow, 2016). Our data suggest that in the rTgTauEC 1 EC-tdTomato/Syp-GFP mouse model, the propagation of tau is not solely a consequence of axon terminal degeneration since tau is detected at postsynapses in the MML of the dentate gyrus prior to the loss of presynapses and in postsynapses directly opposed to presynaptic terminals. Our results are consistent with data from cultured primary rodent neurons and human iPSC derived neurons showing that tau is released from healthy neurons in the absence of cell death (Kanmert et al., 2015); and expand this concept demonstrating that in vivo, tau can spread from presynaptic to postsynaptic elements without substantial loss of presynaptic terminals.
Several key questions remain to understand tau propagation through the brain including, which types of tau spread and what are the mechanisms of tau release from presynapses and uptake from postsynapses? Many forms of tau have been observed to be secreted in vitro including phosphorylated tau (Pooler, Phillips, et al., 2013;Saman et al., 2012) and C-terminally truncated tau (Kanmert et al., 2015). Recent data from AD cerebrospinal fluid and brain samples indicates that high molecular weight tau species may be released from human neurons and are competent to induce tau seeding in cultured cells (Takeda et al., 2016). Several mechanisms of tau release have been proposed (reviewed by (Wang & Mandelkow, 2016)) including exocytosis of tau, vesicle mediated release such as in exosomes (Saman et al., 2012), or synaptic vesicle release (Polanco, Scicluna, Hill, & Gotz, 2016). Once released from the presynaptic cell, multiple mechanisms could also regulate uptake in recipient neurons including endocytosis or fusing of exosomes with recipient cells. In addition to neuronal release and uptake of tau, there is evidence for a role of glia in modulating tau spreading. Microglial activation has been reported to precede tau propagation (Maphis et al., 2015) and it has been proposed that microglia act to phagocytise and release tau in microglial-derived exosomes (Asai et al., 2015). Despite the many questions that remain surrounding the propagation of tau through the brain, it is clear that tau spreads through the brain, likely via synaptic circuits. A promising therapeutic avenue is to prevent the spread of tau using immunotherapies (Gerson & Kayed, 2016). Our results indicate that this spread begins very early in the disease process, and shows the utility of the array tomography technique for detecting tau protein at the level of individual synapses. This will be useful in preclinical studies of treatments to determine whether they prevent the trans-synaptic spread of tau.

ACKNOWLEDGMENTS
Funding provided by Alzheimer's Society, the European Research Council, Alzheimer's Research UK and the Scottish Government, and a University of Edinburgh Wellcome Trust ISSF.