Interaction between a MAPT variant causing frontotemporal dementia and mutant APP affects axonal transport

In Alzheimer's disease, many indicators point to a central role for poor axonal transport, but the potential for stimulating axonal transport to alleviate the disease remains largely untested. Previously, we reported enhanced anterograde axonal transport of mitochondria in 8- to 11-month-old MAPTP301L knockin mice, a genetic model of frontotemporal dementia with parkinsonism-17T. In this study, we further characterized the axonal transport of mitochondria in younger MAPTP301L mice crossed with the familial Alzheimer's disease model, TgCRND8, aiming to test whether boosting axonal transport in young TgCRND8 mice can alleviate axonal swelling. We successfully replicated the enhancement of anterograde axonal transport in young MAPTP301L/P301L knockin animals. Surprisingly, we found that in the presence of the amyloid precursor protein mutations, MAPTP301L/P3101L impaired anterograde axonal transport. The numbers of plaque-associated axonal swellings or amyloid plaques in TgCRND8 brains were unaltered. These findings suggest that amyloid-β promotes an action of mutant tau that impairs axonal transport. As amyloid-β levels increase with age even without amyloid precursor protein mutation, we suggest that this rise could contribute to age-related decline in frontotemporal dementia.


Introduction
Alzheimer's disease (AD) is the most common form of dementia and is characterized by the presence of amyloid plaques and neurofibrillary tangles in brain tissue (Masters et al., 2015). Many studies indicate that axonal transport defects play an important role in the pathogenesis of AD, and overexpression of tau, whose mutation causes some types of frontotemporal dementia, has also been reported to impair axonal transport (Bertrand et al., 2013;Ittner et al., 2008;Majid et al., 2014). Axonal transport is impaired in several amyloid and tau mouse models, and further impairment of transport increases pathology (Adalbert and Coleman, 2013;De Vos et al., 2008;Millecamps and Julien, 2013). These include transgenic mouse models overexpressing amyloid precursor protein (APP) or familial Alzheimer's disease (FAD) mutant APP (Salehi et al., 2006;Stokin et al., 2005), mutant presenilin-1 (Lazarov et al., 2007), and wild-type and frontotemporal dementia with parkinsonism (FTDP) mutant tau (Ishihara et al., 1999;Zhang et al., 2004). The mechanisms by which the pathologic forms of APP, tau, PS1, and amyloid-b (Ab) disrupt axonal transport are not properly understood. Ab itself and overexpression of either wild-type and FTDP tau disrupt axonal transport of a variety of cargoes including mitochondria (Hiruma et al., 2003;Rui et al., 2006;Stamer et al., 2002;Zhang et al., 2004). Live-imaging studies of axonal transport in nervous system tissue have largely focused on mitochondria (Gilley et al., 2012;Mar et al., 2014;Marinkovic et al., 2012;Milde et al., 2015;Misgeld et al., 2007). We have reported live imaging of peripheral nerve explants of mitochondrial transport in a MAPT P301L/P301L knockin mutant mouse and in normal aging (Gilley et al., 2012;Milde et al., 2015).
Axonal swellings are a pathologic feature common to human AD and to mice expressing pathogenic forms of APP or tau and mice expressing ApoE4 (Lewis et al., 2000;Tesseur et al., 2000;Wirths et al., 2006). Our earlier study showed that large axonal swellings in one FAD model, TgCRND8 mice, precede axon loss by many months and that axonal transport may be disrupted specifically at these swellings (Adalbert et al., 2009). TgCRND8 mice inherit a highly aggressive amyloidopathy caused by the expression of double mutant amyloid precursor protein (APP; K670N/M671L plus V717F) (Chishti et al., 2001), leading to early dystrophic neurites and some cell death (Bellucci et al., 2007). These findings raise the question of whether boosting axonal transport can alleviate the pathology. Until now, very few methods have been available to enhance axonal transport in vivo, and no studies have addressed whether this type of intervention can alleviate amyloid pathology or the associated axonal damage. There have been encouraging results using microtubule (MT)-stabilizing agents, such as those used in the treatment of cancer, to improve axonal transport in tauopathy models (Brunden et al., 2012;Zhang et al., 2005). These drugs, however, elicit severe side effects, especially as they have to be administered on a chronic basis, including axonal damage in peripheral neuropathy (Boyette-Davis et al., 2013). In previous work, we have enhanced axonal transport of mitochondria in young mice without adverse effects by knocking in the P301L mutation of MAPT (Gilley et al., 2012), a mutation causing FTDP-17T in humans (Wolfe, 2009). The effect on axonal transport was reversed as animals aged, suggesting other changes during aging alter how axons react to mutant MAPT. The mice remained healthy during the normal life span, with no overt tau pathology, motor, behavioral, or memory deficits. The initial aim of this study, therefore, was to test whether boosting axonal transport in young TgCRND8 mice by crossing them to MAPT P301L knockin mice could alleviate axonal swelling.
We found no alterations in the number of plaque-associated axonal swellings or the number of amyloid plaques in TgCRND8 brains when MAPT mutation was introduced. However, while we successfully confirmed the enhancement of axonal transport in young MAPT P301L/P301L mice and observed a trend in MAPT P301L/þ mice more representative of FTDP-17T patient genetics, the mitochondrial transport in the peripheral nerve axons of TgCRND8 mice showed a significant decrease in the presence of MAPT P301L/P301L , similar to that found in older MAPT P301L/P301L mice. This suggests increasing amyloid beta peptide, or other APP processing products reverses an action of mutant tau on axonal transport and may contribute to a similar switch observed during normal aging.

Animals
All animal work was approved by the Babraham Institute Animal Welfare and Ethical Review Body and UK Home Office and carried out in accordance with the Animals (Scientific Procedures) Act, 1986, under Project Licenses 80/2254 and70/7620. First to confirm whether the enhanced mitochondrial transport is found in 3-month-old homozygous MAPT P301L knockin mice (MAPT P301L/P301L ), knockin mice were crossed with Mito-P (Thy-1-mitoCFP-P) mice as previously described (Gilley et al., 2012;Misgeld et al., 2007). Mito-P mice were kindly provided by Prof. Martin Kerschensteiner (University Munich) and Prof. Thomas Misgeld (TU Munich). Heterozygous male TgCRND8 mice (Chishti et al., 2001) on a 50:50 C57BL/6:129Sv background were bred to double homozygous MAPT P301L knockin/Mito-P or MAPT P301L knockin/YFP-H (Thy1.2-YFP-H) (Feng et al., 2000) female mice to generate triple heterozygotes TgCRND8/MAPT P301L /Mito-P or TgCRND8/MAPT P301L /YFP-H mice. These triple heterozygotes were backcrossed to MAPT P301L homozygotes to generate offspring that were heterozygous for mutant APP and Mito-P (or YFP-H) and homozygous for MAPT P301L . Those mice which were heterozygous for Mito-P were used for transport studies, whereas those with YFP-H were used to assess histopathology. Controls without MAPT P301L knockin were generated by crossing to non-MAPT P301L mice on the same genetic background. Animals were kept on a 12:12 hours light:dark cycle at a constant temperature of 19 C in a pathogenfree environment with up to 5 animals per cage.

Live imaging of axonal transport and image analysis
Imaging of mitochondrial transport and analysis was performed as described previously (Gilley et al., 2012;Milde et al., 2015). Briefly, mitochondrial movements were imaged in sciatic nerve explants using an Olympus Cell R imaging system (IX81 microscope, Hamamatsu ORCA ER camera, Â100 1.45 NA apochromat objective). Peripheral nerves were used as a surrogate for central nervous system (CNS) tissue because in our hands quantitative analysis of axonal transport here is more reproducible than in CNS (Milde et al., 2015), but the presence of Ab in peripheral nerves was confirmed (Fig. 2). During imaging, tissues were maintained in oxygenated Neurobasal-A medium at 37 C in an environment chamber (Solent Scientific Ltd). Images were captured using fixed light intensity and camera exposure time settings at a rate of 2 frames per second for 5 minutes. Five to 10 individual movies (often containing multiple axons) were captured for each tissue explant. Individual axons were straightened using the Straighten plugin in ImageJ software, version 1.44 (Rasband, W.S., ImageJ, US National Institutes of Health, Bethesda, MD, USA; http://imagej.nih.gov/ij/, 1997e2012). Axonal transport parameters were determined for individual axons using the Difference Tracker set of ImageJ plugins (Andrews et al., 2010). The principal output of these plugins is the number of moving particles identified in each frame of the image, normalized to axon length (presented as particle count per second per 100 mm axon length).

Histology and staining
Mice were perfused transcardially with 4% phosphate-buffered paraformaldehyde, and brains and spinal cords processed as previously described (Adalbert et al., 2009). Sagittal brain sections were cut at 50 mm, and serial transverse spinal cord sections were cut at 25 mm through the L3eL5 segments using a Leica CM 1850 cryostat. To stain fibrillar amyloid deposits, brain sections were incubated in 0.02% Thioflavine S in tris-buffered saline (TBS) for 10 minutes and then rinsed in 50% ethanol in TBS and 100% TBS before mounting in Vectashield Mounting Medium. For motor neuron analysis, the spinal cord sections were stained with 0.5% cresyl violet (Adalbert et al., 2006). The excised YFP-H sciatic nerves were incubated in 1% Triton X-100 (Sigma) in 0.1M PBS for 10 minutes at room temperature, washed extensively in 0.1M PBS, and then mounted in Vectashield.

Microscopy and quantitative/qualitative analysis
The stained sections were imaged with an Olympus FV1000 confocal microscope imaging system using a 40Â NA 1.3 or 63 Â NA 1.4 oil immersion objective. Neuronal structures and amyloid plaques were scanned using a multitrack configuration with laser excitation (ex.) lines and emission (em.) filters as follows: Thioflavine S 405 nm ex. 475e525 nm em. and YFP 488 nm ex. 505e550 nm em. Single confocal slices or z-stacks at 0.4e0.5 mm steps were acquired to generate the data images. Conventional fluorescence imaging was carried out using an Olympus IX 81 inverted microscope coupled to an Olympus U-TV0.5XC digital camera system. Bright-field images of Nissl-stained sections were acquired using an Olympus BX41 upright microscope (40Â objective) and MicroPublisher 3.3 camera (QImaging).
The number of amyloid plaques stained with Thioflavin S was determined in 6 sagittal sections (3 sections/hemisphere) as previously described (Adalbert et al., 2009). Amyloid plaques that had at least 1 YFP-containing axonal dystrophy (cutoff size of the dystrophy 5 mm) in the immediate vicinity were counted as plaques associated with axonal swellings. Cresyl violetestained motor neurons with centrally placed nuclei and cytoplasm rich in rough endoplasmic reticulum were considered normal morphology. Every second section from L3eL5 spinal cord segments was qualitatively analyzed. YFP positive axons from sciatic and tibial nerves with no fragmentation or swellings along their length were classified as intact (Beirowski et al., 2004).

Ab ELISA
To determine the levels of human Ab 1e42 in brain and sciatic nerve, samples were analyzed using a commercially available ELISA kit (Life Tech: KHB3441). Brain and sciatic nerves were . For all graphs, each data point represents the mean value obtained for 1 animal (5 fields of view and, on average, 15 axons per animal). Horizontal bar indicates mean and error bars indicate standard error of the mean. Statistically significant differences between genotypes are indicated (*p < 0.05, **p < 0.01; 1-way analysis of variance with Tukey multiple comparisons posttest). Abbreviation: n.s., not significant.

Fig. 2.
Human Ab 1e42 is present in peripheral nerves of TgCRND8 mice, and its levels in brain are independent of the MAPT P301L knockin mutation. (A) Concentration of human Ab 1e42 (ng/g) in sciatic nerves of 3-month-old wild-type (MitoP) and TgCRND8 mice. (B) Brain concentration of human Ab 1e42 (ng/g) in TgCRND8 and TgCRND8 mice heterozygotes for the MAPT P301L knockin mutation. Each data point represents 1 animal (mean and standard error of the mean, unpaired t test). Abbreviation: n.s., not significant. homogenized in 5 M guanidine hydrochloride (8 mL per gram of tissue) supplemented with 1Â Protease Inhibitor Cocktail (Roche) for 3e4 hours at room temperature. The sample was then frozen at À20 C until use. Before running in the ELISA, the homogenate was diluted 1:50 in ice cold reaction buffer (Dulbecco's PBS þ0.03 % Tween þ5 % BSA supplemented with 1x Protease Inhibitor Cocktail) and centrifuged for 20 minutes at 4 C at 16,000 Â g. The supernatant was then diluted (in the commercial kit dilution buffer) before undergoing ELISA. Briefly, samples were incubated with Ab detection antibody for 3 hours at RT then thoroughly washed. HRP-conjugated antibody was added to sample wells for 30 minutes. After another wash step, samples were incubated with stabilized chromogen for 30 minutes. The reaction was stopped using an acid-based stop solution and absorbance read at 450 nm using a PheraStar FS plate reader. Samples were run with a standard curve (4-parameter fit) to obtain a concentration readout. Final concentration is given as ng of Ab per gram of tissue (Harwell and Coleman, 2016).

Immunohistochemistry and Gallyas staining
Mouse brains were fixed in 10% formalin for 24 hours before embedding in paraffin. Then, 7-mm-thick tissue sections were stained with the Gallyas silver staining method to identify neurofibrillary tangles as reported previously (Gilley et al., 2012;Kuninaka et al., 2015). They were examined with a Zeiss Axioplan microscope, and digital images were acquired using an AxioCam HRc camera.
The immunohistochemical labeling was performed using the ABC method. Sections were treated with 0.3% H 2 O 2 to inhibit endogenous peroxidase and then blocked in 10% vol/vol normal horse serum in TBS (0.01 M Tris, 0.15 M NaCl, pH 7.4). After overnight incubation with diluted PHF-1 or AT8 antibodies (Innogenetics), sections were incubated with biotin-conjugated horse anti-mouse antibodies followed by ABC complex (Vector Laboratories, Burlingame, CA, USA). Peroxidase activity was developed using diaminobenzidine as chromogen.

Statistical analysis
Statistical tests, as described in the figure legends, were performed using Prism software (GraphPad Software Inc, La Jolla, CA, USA). A p value of >0.05 was considered not significant (ns), and *p < 0.05, **p < 0.01, and ***p < 0.001 was significant.

Enhanced mitochondrial transport in 3-month-old MAPT P301L/P301L knockin mouse peripheral nerves
First, we wanted to confirm whether the enhanced mitochondrial transport found previously in 8e11 months homozygous MAPT P301L tau knockin mice (Gilley et al., 2012) was also present in younger, 3-month-old homozygous tau knockin mice, an age when TgCRND8 shows substantial axonal swellings and an increasing plaque density (Adalbert et al., 2009). To test this, we crossed MAPT P301L knockin mice to Thy-1-mitoCFP-P (Mito-P) mice, which have CFP-labeled mitochondria in a subset of their neurons (Misgeld et al., 2007). The transport of CFP-labeled mitochondria was assessed in axons of explanted sciatic nerves from 3-month-old MAPT P301L/P301L , MAPT P301L/þ , and MAPT þ/þ mice. Peripheral nerves were used because we find quantitative analysis of axonal transport in nerves is more reproducible than that in CNS (Milde et al., 2015). We found that the number of CFP-labeled mitochondria moving in an anterograde direction was significantly higher in MAPT P301L/P301L mice relative to wildtypes (Fig 1A). In MAPT P301L/þ mice, there was a trend toward increased mitochondrial transport, but the difference was not statistically significant. No significant differences were found for retrograde transport (Fig 1B), and changes in anterograde flux were not accompanied by significant alterations to maximum speed or average speed (not shown). Thus, axons from young 3month-old MAPT P301L/P301L knockin mice show an increase in anterograde mitochondrial flux, similar to older (8e11 months) mice, prior to the reversal of this effect we have previously observed in old age (Gilley et al., 2012).

3.2.
In the presence of the TgCRND8 mutation, MAPT P301L/P301L reduces the axonal transport of mitochondria Next, we assessed whether the presence of MAPT P301L knockin mutation, alters the mitochondrial transport in TgCRND8 peripheral axons. Instead of the expected enhancement, we observed a significant decrease in anterograde mitochondrial flux in 3-month-old TgCRND8 mice homozygous for the MAPT P301L knockin mutation (TgCRND8/MAPT P301L/P301L ) (Fig 1C), reminiscent of old MAPT P301L/P301L mice without mutant APP (Gilley et al., 2012). Retrograde mitochondrial transport also declined, but this was not statistically significant in our sample (Fig 1D), and average speed of transport was unaltered (not shown). No significant differences in any transport parameters were identified for TgCRND8 mice heterozygous for the tau knockin mutation (TgCRND8/MAPT P301L/þ ). These data indicate that in the presence of the TgCRND8 mutation, the effect of MAPT P301L/P301L on axonal transport of mitochondria in young animals is reversed, suggesting an interaction between Ab, or other APP processing products, and mutant tau that resembles the effect of normal aging.

Ab 1e42 is present in peripheral nerves of TgCRND8 mice and Ab 1e42 level in brain is not altered by MAPT P301L/þ
Using ELISA, we quantified the Ab 1e42 in sciatic nerves from young (3 months) TgCRND8 mice and found that Ab 1e42 is present in peripheral nerves, at the age when the axonal transport measurements were conducted ( Fig. 2A). However, the levels were less than 10% of those in brain of the same animals (Fig. 2B). This suggests that the use of peripheral nerves as a surrogate for CNS tissue in this study could underestimate the effect of Ab, or other APP processing products, in combination with tau in the CNS. Brain levels of Ab were not altered by the presence of the MAPT P301L/þ knockin mutation (Fig. 2B).

MAPT P301L/P301L knockin mutation does not reduce the number of plaques and the associated axonal swellings in TgCRND8 brains
To test whether the presence of MAPT P301L knockin mutation ameliorates amyloid-associated pathology, we quantified the number of amyloid plaques and the number of plaque-associated axonal swellings in the brains of TgCRND8/MAPT þ/þ /YFP-H and TgCRND8/MAPT P301L/P301L /YFP-H mice at 3 months of age. We found that the MAPT P301L/P301L mutation did not significantly reduce the number of amyloid plaques (Fig 3A, C) or the associated axonal swellings in the brains of TgCRND8 mice (Fig 3B, D). Although as axonal transport was not enhanced in these animals as expected, it remains undetermined whether boosting transport can alleviate axonal swellings.

No morphologic alterations are present in TgCRND8/ MAPT P301L/P301L spinal cord motor neurons and peripheral nerves
Histologic analysis of motor neurons from lumbar spinal cord segments L3-L5 and YFP positive axons from sciatic and tibial nerves showed no obvious alterations in 3-month-old TgCRND8/MAPT P301L/ P301L mice. Cresyl violet staining revealed motor neurons with normal morphology with centrally placed nuclei and a cytoplasm rich in rough endoplasmic reticulum (Fig 4A). Also, YFP positive axons from sciatic and tibial nerves were intact with no fragmentation or swellings along their length (Fig 4B, C). These observations show that the reduced anterograde axonal transport of mitochondria in peripheral nerves of TgCRND8/MAPT P301L/P301L mice occurs in morphologically normal axons and spinal cord motor neurons.

Absence of tau pathology
We previously showed that pathologic aggregation of hyperphosphorylated tau does not occur in MAPT P301L/P301L tau knockin mice (Gilley et al., 2012). Similarly, in this study, no Gallyaspositive neurons or neuronal PHF-1 immunostaining for phosphorylated tau was detected in 6-month-old brains even after combining TgCRND8 transgene with MAPT P301L/P301L or MAPT P301L/þ knockin mutation (Fig 5). Only a puncta-like staining of dystrophic neurites around the congo red positive plaques was found ( Fig 5A). Hence, the presence of FAD-APP does not induce tau pathology in MAPT P301L tau knockin mice, at least up to the age of 6 months.

Discussion
In this study, we confirmed and extended our previous observation of increased anterograde mitochondrial transport in 8-to 11month-old homozygous MAPT P301L knockin mice (Gilley et al., 2012) to 3-month-old homozygous knockin mice (Fig 1A) and to a trend in the same direction in heterozygotes. Retrograde transport (Fig 1B), maximum speed or average speed in either direction (not shown) were unaltered, and after crossing to TgCRND8, the numbers of plaque-associated axonal swellings or amyloid plaques in TgCRND8 brains were neither reduced nor increased by MAPT P301L/P301L . By crossing the MAPT P301L knockin mice with an FAD model, TgCRND8 mice, our initial aim was to test whether boosting axonal transport can alleviate axonal swellings. We have selected TgCRND8 mice for our study as they develop robust axonal swellings at an early stage in amyloid pathology that we have extensively characterized, preceding axon and cell death, and major loss of synapses (Adalbert et al., 2009). Although these features arise in young animals, they closely resemble those in an aging human AD brain. Therefore, it is reasonable to use this model to test whether they can be alleviated by boosting axonal transport. These axonal swellings contain accumulations of fragmented mitochondria and other organelles, which suggest impairment of axonal transport. Therefore, "unblocking" these swellings could alleviate AD-like histopathology and symptoms in TgCRND8 mice. Surprisingly, the presence of the homozygous MAPT P301L mutation significantly decreased the anterograde mitochondrial flux in 3-month-old TgCRND8 mice instead of enhancing it as we initially hypothesized, so it remains untested at present whether increasing axonal transport could alleviate plaques and/or axonal swellings.
Previous studies indicated that the physiological function of tau is required for aspects of FAD-APP pathogenesis. The block of axonal transport by Ab 1e42 oligomers in hippocampal primary cultures requires the presence of tau (Vossel et al., 2010), and FAD-APPeinduced behavioral deficits in mice are alleviated by removing tau (Roberson et al., 2007). Recently, it was also reported that mutant tau with markedly reduced MT-binding capacity still enabled Ab-induced axonal transport deficits (Vossel et al., 2015). We now show that co-expression of MAPT P301L , even at physiological levels, synergistically impairs axonal transport with FAD-APP. All these studies suggest that different tau genotypes, in the presence of Ab, can lead to transport impairment.
Our findings support the hypothesis that Ab and tau converge in axons to influence axonal transport or that tau itself is a downstream target of Ab on the same pathway. Previously, it was suggested that tau may lie downstream of APP/Ab or on a necessary pathway that converges onto a common, downstream target (Karran et al., 2011;Morris et al., 2011). In different neurodegenerative diseases, both in humans and in mice, plaques and tangles can arise independently of one another. In mice, neither FAD-APP nor FTDP-17T mutant tau causes both of these pathological hallmarks. Some human patients show amyloid without tau pathology, and FTDP-17T tau mutations cause tangles without plaques in frontotemporal dementia. These observations are consistent with convergent disease pathways, either of which can cause pathology when sufficiently activated. MAPT P301L knockin mutation does not enhance the generation of Ab 1e42 in the brains of 3-month-old TgCRND8 mice (Fig 2B), consistent with previous results using overexpression of tau in FAD-APP expressing cell lines or primary neuronal cultures (Goldsbury et al., 2006). In addition, when tau deletion reduced cognitive deficits and excitotoxic damage in FAD-APP mice, there was no corresponding reduction in Ab 1e42 (Roberson et al., 2007). This finding is important because if the synergistic effect of MAPT P301L knockin tau and FAD-APP were mediated by increasing Ab 1e42 processing, then that would favor a mechanism involving feedback on APP over a common downstream target for Ab 1e42 and MAPT P301L tau.
Tau pathology is not a feature of APP or PS1 mutant mice but has been seen when combined with overexpressing tau mutants. We were interested to see whether such pathology could be induced when mutant tau is expressed at physiological levels. The overexpression of Ab did not induce pathologic aggregation of hyperphosphorylated tau in MAPT P301L tau knockin mice by the age of 6 months; however, we cannot rule out the possibility that tau pathology could develop at an older age. The puncta-like staining of phosphorylated tau in dystrophic neurites around the plaques we observed in our study (Fig. 5A,B) was reported previously in mutant APP and APP/PS1 transgenic mice, therefore is unlikely to reflect an action of MAPT mutation (Boutajangout et al., 2004;Kurt et al., 2003;Tomidokoro et al., 2001).
It is worth noting that alternative models that overexpress both mutant APP and MAPT P301L show late development of tau tangles; a double transgenic requiring 9 months and a different 3ÂTg model not showing pathology until 12 months of age (Lewis et al., 2001;Oddo et al., 2003). It may be due to the knockin nature of the MAPT P301L in our model that neurofibrillary pathology cannot develop during the normal life span of our mice.
Some evidence suggests that increased tau expression and/or altered tau isoform ratio are more important for the development of neurofibrillary pathology in a mouse than is the presence of an FTDP-17T mutation (Adams et al., 2009;Gotz et al., 2001). The presence of tau at physiological levels and with the expected isoform ratios in our MAPT P301L tau knockin mice could be responsible for the absence of tau pathology. In addition, it could be that murine tau has less propensity for aggregation in vivo compared to human tau (Allen et al., 2002;Ando et al., 2011;Sahara et al., 2002). The decline in axonal transport seen when combining the APP and MAPT P301L mutations may anticipate the age-related effect of MAPT P301L tau mutation alone. As Ab levels increase with age even without APP mutation (Lesne et al., 2013), our double mutants may have modeled this process in younger animals. Thus, we suggest that this Ab rise could contribute to age-related decline in frontotemporal dementia. It would be important to repeat our observations with other mutant tau knockin mice (e.g., R406W) as well, to know whether axonal transport impairment in the presence of Ab is a consistent effect of FTDP tau mutations. In addition, it would be interesting to test whether amyloid-independent pathogenic mechanism can also lead to axonal transport impairment observed in our model, by crossing the MAPT P301L tau knockin mice with wild-type human APP overexpressing mice (Mucke et al., 2000). In these mice, the human wild-type APP processing is basically nonamyloidogenic (Simon et al., 2009). Using our mouse model, we found that Ab 1e42 is present in peripheral nerves at the age when the axonal transport measurements were done ( Fig. 2A), similar to previous studies in the peripheral nerves of other APP mutant mice (Jolivalt et al., 2012). In our model, the Ab 1e42 levels in the peripheral nerves were less than 10% of those in the brains of the same animals, which suggest that the effect of Ab in combination with tau in the CNS could be much stronger.
In the presence of the APP mutation, MAPT P301L/P301L knockin failed to enhance axonal transport of mitochondria; therefore, new strategies have to be used to increase transport (Hinckelmann et al., 2013). Drugs increasing MT acetylation may be useful to boost axonal transport in AD. Tubulin acetylation is reduced in the brains of patients with AD (Hinckelmann et al., 2013) and in the brains of various mouse models of AD (Govindarajan et al., 2013;Kim et al., 2012). Studies so far showed that HDAC6 inhibitors reverse the amyloid betaeinduced changes in a-tubulin acetylation and mitochondrial trafficking in cultured hippocampal neurons (Kim et al., 2012) and HDAC6 knockout in a mouse model of AD restores atubulin acetylation in the brain and cognitive function in AD mice (Govindarajan et al., 2013). Another approach to improve axonal transport could be to inhibit cJun N-terminal kinase (JNK) activity. JNK hyperactivity has been reported in models of AD, and the inhibition of JNK activity in vivo was shown to be neuroprotective in animal models (Mehan et al., 2011). JNK3 inhibition rescued axonal transport defects induced by pathogenic fragments of huntingtin in a squid model of axonal transport (Morfini et al., 2009).
In summary, we confirmed and extended our previous observation of an increase in mitochondrial transport in young MAPT P301L knockin homozygous mice to animals aged 3 months and observed a nonsignificant trend also in heterozygotes. This could represent early consequences of the tau dysfunction that may be relevant to FTDP-17T pathogenesis in humans as they age. Our results demonstrate that TgCRND8 and MAPT P301L/P301L mutations in combination reduce the axonal transport of mitochondria in peripheral nerves, in contrast to the enhancement of transport in young MAPT P301L/P301L mice. This exciting finding suggests that Ab and tau converge in axons to influence axonal transport or that tau itself is a downstream target of amyloid beta on the same pathway. MAPT P301L mutation did not significantly reduce the number of plaque associated axonal swellings or the number of amyloid plaques in TgCRND8 brains and did not alter Ab levels in brain. Our FAD-APP and MAPT P301L tau knockin mouse model may be useful to study how amyloid beta peptide and tau combine in pathogenesis.

Disclosure statement
The authors have no actual or potential conflicts of interest.