Histopathology studies demonstrate progressive tauopathy radiating from the EC
A detailed histopathological analysis was performed on relatively young (10–11 mon) and aged (~22 mon) neuropsin-tTA-tau (NT) transgenic mice using three different antibodies, MC1, CP27 and AT8. The MC1 monoclonal antibody detects human tau in an abnormal conformation 
that is associated with early stage NFT–tau in human AD patients 
. In young NT mice (), abnormal human tau recognized by MC1 was most abundant in the medial EC (MEC). Relatively dense staining was also seen in the lateral EC (LEC) and the para-(PaS) subiculum while the presubiculum (PrS) was less intensely stained. Dense staining was seen in superficial layers II and III of the EC, whereas deeper layers showed considerably less staining. Human tau was present in some cell bodies, but mostly in neurites within the superficial layers of the MEC and LEC (). Dense tau staining was seen in the middle third of the molecular layer of the DG and CA3 () but not the outer layer indicating tau in axon terminals of the perforant pathway (pp) that originate from layer II of the MEC 
. In the CA1 and subiculum, the outer molecular layer was labeled extensively (), indicating tau in perforant path terminals from layer III cells in both LEC and MEC 
. Mice expressing only the uninduced tau transgene (control) showed negligible (), or very limited immunoreactivity with the antibodies used, and it was usually restricted to the mossy fibers (for example, antibody CP27, ). Some non-specific staining in the fornix was seen in all mice, with all antibodies. By 22 months of age, the distribution of human tau in old NT mice had changed dramatically to resemble that seen in more affected AD brain tissue (Braak stages II–III). Intense MC1 immunoreactivity was readily detected not only in neurons in the superficial layers of the EC and throughout the subiculum (), but in pyramidal neurons in the hippocampus, especially in CA1, and also in dentate gyrus granule cells (DG GCs) (). Somatodendritic staining with MC1 was intense for cells in the MEC (). Scattered MC1 positive neuronal cell bodies could also be seen in the perirhinal and the parietal cortices (), and more extensively in the deeper layers of the EC.
Progressive spread of tauopathy in NT mice identified by antibody MC1.
Progressive spread of tauopathy in NT mice identified by antibody CP27.
The pattern of staining was reproduced in young and old NT mice using a human specific tau antibody (CP27) that recognizes all human tau, regardless of phosphorylation or conformation status (). Subtle differences in the relative intensity of staining in different areas were observed for different antibodies, especially in the DG GC layer where CP27 staining was more intense and extensive than MC1 ( and vs. ). This could either indicate differential sensitivity of the antibodies, differential synthesis or clearance of tau forms recognized by the two antibodies, or retarded development of the conformational change in tau recognized by MC1.
Progressive spread of tauopathy to monosynaptically connected regions of the hippocampus.
To assess whether tauopathy could spread across a synapse, we examined cells in the DG that are monosynaptically connected with cells in the EC (). Young NT mice () showed robust accumulation of CP27 immunoreactive human tau in the endzones of the perforant pathway that originate from neurons in the MEC and terminate in the middle third of the molecular layer of the DG (area 3). Low, but detectable levels of immunoreactivity were seen in the outer third of the molecular layer (area 4) which represents terminals from neurons originating in the LEC. Some human tau was seen in cells in the hilus (area 1), most likely in mossy cells. Notably, human tau did not accumulate in DG GCs (area 2) in young NT mice. Old NT mice however showed a very different distribution of human tau (). Robust accumulation of human tau was now seen in DG GCs (area 2) and increased accumulation of human tau was seen in layers 1, 2 and 4. The appearance of tau in DGGCs strongly supports the idea that tauopathy initiated in the EC can spread between cells that are connected, but physically separated by a synapse. Interestingly, the perforant pathway endzone in layer 3 was significantly depleted of tau which coincided with accumulation in originating cell bodies in the MEC (). This apparent relocalization of tau from axons to somatodendritic compartments is one of the earliest events in the pathological cascade of early Alzheimer's disease 
Tauopathy in AD is usually staged using the antibody AT8 
. This antibody recognizes phosphorylated epitopes S202/205 (in both mouse and human tau) that are abundant in tau from AD brain, but not normal brain 
. In young NT mice, () AT8 immunoreactive tau was mainly concentrated in the EC with no staining visible in the hippocampal subfields. Cell body staining was predominant with relatively less staining seen in neurites. In old NT mice (), there was considerably more neurite staining throughout the EC (), and in all fields of the hippocampus (), with cell body immunoreactivity being seen in scattered neurons that were most prominent in pyramidal cells in the CA1 and in DG GCs. As for MC1, in the old mice, additional cell body staining was apparent in the deeper layers of the EC, and in cells in the perirhinal and parietal cortices (). The control mouse () was essentially negative for this antibody. Overall, the pattern of staining, including extensive staining of cell bodies and neurites throughout the EC and hippocampus was reminiscent of that described for early Braak stages of AD 
Progressive spread of tauopathy in NT mice identified by antibody AT8.
Although the exact type of tau associated with functional impairment and degeneration is not known 
, the accumulation of insoluble, conformationally abnormal, hyperphosphorylated tau into mature neurofibrillary tangles in the somatodendritic cell compartment is generally associated with more severe pathology, degeneration and cell death. To test whether mature tangles had formed in the NT mice, we examined tissue sections stained with thioflavin S (thioS), a dye that binds to proteins in a β- sheet conformation, indicative of tau in mature tangles (). Special care was taken to mask lipofuscin fluorescence which is significant in old mice. A small number of neurons restricted to the MEC were positive for thioS in old NT mice (). Young NT () and old control mice () were essentially negative. Not all of the tau immunoreactive neurons in the MEC of old NT mice were thioS positive, and cells in the LEC, CA1 and DG GC layer were thioS negative, as were neurites and axonal terminals in the perforant pathway. As cells with the highest level of human tau occur in the MEC, the lack of staining in other areas is most likely explained by the lower tau levels rather than by regional sensitivity to tangle formation, but the latter interpretation cannot be ruled out in these studies.
Mature, filamentous neurofibrillary tangle formation in the EC of old NT mice.
Altered conformation of proteins can also be visualized by silver staining using one of several methods 
. Argyophilic plaques, tangles and neurites are abundant in the human AD brain. Abundant, argyrophilic cell body and neurite staining was also seen in the old (), but not the young NT mice (), and it was related to tauopathy development rather than aging as parallel-processed, old littermate control mice were negative (). The distribution of histopathology in the old NT mouse was extensive, with robust staining being seen in cells in the EC, as well as in the subiculum (). Staining was also extensive in the CA1, but to a lesser extent in the DG GC layer (). In general, the distribution of silver-staining matched that seen with the MC1 antibody more closely than that seen with the CP27 or AT8 antibodies, suggesting that it is the conformational change in tau recognized by MC1 that is recognized by the silver stain. Interestingly, MC1 immunoreactivity was robust in neurites in the young NT mice but these mice were negative for silver staining. Therefore the silver stain recognizes a more advanced conformational abnormality that lies between pre-tangle MC1 immunoreactivity seen in the young mice, and the overt conformational change recognized by thioS, which in the old mice, is restricted to cells in the MEC.
Trans-synaptic spread of pathology identified, but mechanism unknown
One of the most intriguing observations from our studies is the appearance of tauopathy in cells outside of the entorhinal cortex. As shown in , granule cells in the DG of old NT mice accumulate human tau protein, but it is unknown whether the human tau protein accumulated in the DG GCs derives solely from uptake or transfer of human tau from neurons in the EC, or if the human tau protein could be generated from endogenously produced human tau mRNA in the DG GCs resulting from “leaky” expression of the transgene. To test whether endogenously produced human tau in the DG GCs could be contributing to the tauopathy seen there, we collected approximately 1000 individual neurons by laser-capture microdissection (LCM) from the DG GC layer from old () and young () mice and assessed whether human tau mRNA was expressed in them. For this experiment, young mice at ~4 months of age were sampled to reduce the likelihood that incipient pathology had developed. Old NT, young NT and non transgenic mouse tissue sections were double immunolabeled with both CP27 and NeuN to ensure that neurons were isolated. shows total RNA extracted from LCM isolated cell populations from one mouse from each type. A gel image from the 2100 Bioanalyzer, which employs capillary gel electrophoretic methodology to measure RNA integrity (RIN) and abundance, demonstrated the quality of sample from the LCM isolated cells in young mice - Non Tg (RIN 6.3; 418 pg/µl; total yield 4.18 ng), tau protein negative (Tau−) GCs (RIN 8.1; 443 pg/µl; total yield 4.43 ng), and old mice - tau protein positive (Tau+) GCs (RIN 8.6; 907 pg/µl; total yield 9.07 ng) and Tau− GCs (RIN 7.8; 351 pg/µl; total yield 3.51 ng). The level of human tau mRNA was then measured by quantitative (q) RT-PCR. To confirm the specificity of our human tau primers, LCM was used to isolate DG GCs from a non Tg young mouse that had been processed in the same way as the young NT mouse. QRT-PCR identified no amplification from the non Tg sample (CT>40) demonstrating that the primers were completely human specific (). To assess whether amplification could result from residual DNA contamination even after DNAase treatment, RNA from DG GCs was used for qPCR with, or without the reverse-transcription step required to synthesize cDNA. Human tau amplicons in the sample that had not been reverse-transcribed were found to be at least twenty five fold (25X) lower than in the transcribed samples, suggesting that genomic DNA transgene priming resulted in only very low levels of PCR amplification compared to mRNA priming. shows human tau expression in the human tau protein negative DG GCs (Tau− GCs) from a young NT mouse (left side of panel), and in human tau protein positive (Tau+ GCs) and negative (Tau− GCs) DG GCs from an old NT mice (right side of panel). Surprisingly, in the young NT mouse where all of the DG GCs were tau protein negative, the Tau− GCs had low but detectable levels of human tau mRNA after normalization to β-actin. In old mice, human tau mRNA was again detected in DG GCs from the NT mouse, both in cells that were positive or negative for tau protein. Relative expression changes (2−ΔΔCT) between the two populations of DG GCs in the old NT mouse showed a 27% increase in expression levels between Tau− and Tau+ DG GCs. A second, old NT mouse also had quantifiable levels of human tau mRNA in all three cell populations, with mRNA levels in Tau+ GCs>Tau− GCs (data not shown). The reason for this apparent increase is unknown but if it is significant in a larger sample group, it could reflect a feedback mechanism whereby tau transcription is upregulated if tau is losing its ability to perform its' normal function as it accumulates in the cytosol in the Tau+ DG GCs. The apparent increase in tau levels in old compared to young samples could reflect an age-dependent increase in transcription, or it could simply represent inter-animal variability.
LCM-isolation of cell populations and assessment of mRNA levels by qRT-PCR.
Despite the absence of detectable human tau protein immunoreactivity in the tau negative DG GCs from young or old NT mice; human tau mRNA was identified in cells from three separate experiments suggesting that transgene expression was slightly leaky. To further examine whether ectopic expression in the DG GCs could explain the accumulation of protein at old age, we examined mice from two other crosses to the neuropsin tTA activator mouse. The first cross was to a reporter gene expressing LacZ with a nuclear localization signal to restrict the cellular distribution of the reporter to the nucleus of the cell in which it was produced 
. As shown in , positive staining for LacZ was restricted to very few cells in the DG GC layer. The second cross was to a mutant APP responder line. When crossed to the neuropsin-tTA activator, this mouse expresses APP predominantly in the EC as expected (data not shown), but due to secretion of Aβ, plaques accumulate in regions outside of the EC, including layers of the DG. Of note, the cells of the DG GC layer did not accumulate APP/Aβ supporting our observations that ectopic expression of responder genes in this cell layer is negligible. Therefore, despite our findings of some human tau mRNA in DG GCs, ectopic expression in these cells is very limited and unlikely to account for the extensive immunolabeling with human tau specific antibodies seen in the old mice. As we pooled many DG GCs and analysis of mRNA levels by qPCR is extremely sensitive, the actual number of cells contributing the human tau mRNA could be a very low percentage of the total that were human tau protein positive. How human tau protein accumulating in DG GCs that were likely to be human tau mRNA negative was derived is unknown, but it is possible that tau was released from cells originating in the EC, and internalized by DG GCs synapsing on to them. In support of this mechanism, a recent study has shown that tau can be released from cells via exosomes and tau positive exosomes have been identified in human CSF from AD patients 
In general, our NT mouse model replicates the spatial and temporal aspects of the earliest stages (I–III) of Braak staging of tauopathy in Alzheimer's disease. We have demonstrated that tau pathology initiating in the EC can spread to other synaptically connected brain areas as the mice age, supporting the idea that AD progresses via an anatomical cascade as opposed to individual events occurring in differentially vulnerable regions. Thus, our NT transgenic mouse provides a model in which the spatial and temporal propagation of the disease can be predicted, and correlative functional outcomes can now be tested. Given that the earliest Braak stages are not associated with cognitive decline, identifying an EC based “biomarker” for pathology or dysfunction and developing therapeutic strategies to prevent propagation are likely to be both possible, and beneficial.