Presynaptic dystrophic neurites surrounding amyloid plaques are sites of microtubule disruption, BACE1 elevation, and increased Aβ generation in Alzheimer’s disease

Alzheimer’s disease (AD) is characterized by amyloid plaques composed of the β-amyloid (Aβ) peptide surrounded by swollen presynaptic dystrophic neurites consisting of dysfunctional axons and terminals that accumulate the β-site amyloid precursor protein (APP) cleaving enzyme (BACE1) required for Aβ generation. The cellular and molecular mechanisms that govern presynaptic dystrophic neurite formation are unclear, and elucidating these processes may lead to novel AD therapeutic strategies. Previous studies suggest Aβ may disrupt microtubules, which we hypothesize have a critical role in the development of presynaptic dystrophies. To investigate this further, here we have assessed the effects of Aβ, particularly neurotoxic Aβ42, on microtubules during the formation of presynaptic dystrophic neurites in vitro and in vivo. Live-cell imaging of primary neurons revealed that exposure to Aβ42 oligomers caused varicose and beaded neurites with extensive microtubule disruption, and inhibited anterograde and retrograde trafficking. In brain sections from AD patients and the 5XFAD transgenic mouse model of amyloid pathology, dystrophic neurite halos with BACE1 elevation around amyloid plaques exhibited aberrant tubulin accumulations or voids. At the ultrastructural level, peri-plaque dystrophies were strikingly devoid of microtubules and replete with multi-lamellar vesicles resembling autophagic intermediates. Proteins of the microtubule motors, kinesin and dynein, and other neuronal proteins were aberrantly localized in peri-plaque dystrophies. Inactive pro-cathepsin D also accumulated in peri-plaque dystrophies, indicating reduced lysosomal function. Most importantly, BACE1 accumulation in peri-plaque dystrophies caused increased BACE1 cleavage of APP and Aβ generation. Our study supports the hypothesis that Aβ induces microtubule disruption in presynaptic dystrophic neurites that surround plaques, thus impairing axonal transport and leading to accumulation of BACE1 and exacerbation of amyloid pathology in AD. Electronic supplementary material The online version of this article (doi:10.1007/s00401-016-1558-9) contains supplementary material, which is available to authorized users.

particle motility. Live-cell images were acquired on a motorized Nikon TE 2000 microscope maintained at 37°C in a custom-designed environment chamber, at the rate of one frame per second, using 60X (NA 1.49) objective and Cascade II:512 CCD camera (Photometrics). Stacks of images were opened in ImageJ, and individual neurites traced with the segmented line tool. For each neurite traced, as indicated by the arrows along the representative neurites in the left panels of Figure S1, a kymograph was generated (shown in right panels) using the "MultipleKymograph" plugin. For each kymograph, the stationary and motile particles were counted, using the "Cell Counter" plugin in ImageJ. Motile particles were defined as puncta that exhibited lateral displacement in the kymograph. For each neurite, the percent of motile particles was calculated from these data. Each point in the scatter plot indicates a neurite and the yaxis value is the percentage of motile particles in that given neurite. Kymographs were generated from 11 neurites of vehicle-treated neurons, which contained a total of 245 NPY-mCherry and 325 BACE1-YFP particles, and from 20 Aβ42-treated neurites, which contained a total of 524 NPY-mCherry and 691 BACE1-YFP particles. We observed a significant decrease in the average percentage of motile particles per neurite for both NPY-mCherry and BACE1-YFP puncta.

Quantification of tubulin immunofluorescence in peri-plaque dystrophies:
After capturing multichannel confocal images for BACE1 (red) and tubulin isoform (green) immunofluorescence signals with NIS Elements software, the single channel tiff files were opened in ImageJ for quantification. Using the region of interest (ROI) manager, free-hand ROIs were drawn around dystrophies in the halo of high BACE1 signal around plaque cores, and the same number of similar sized ROIs was drawn in normal appearing nearby neuropil in the same image. Using the "MultiMeasure" tool, the total intensity for each ROI was measured in the BACE1 channel, and in the tubulin channel, and the ratio between the two channels determined in Excel. For human tissue 16 dystrophies and corresponding non-dystrophic regions were measured per case. For murine tissue, BACE1/tubulin isoform ratios were determined for 11-20 BACE1-positive dystrophic regions and a corresponding number of nearby normal neuropil areas. Mouse dystrophies were much more numerous and single dystrophies harder to distinguish than human dystrophies, so the whole halo of high BACE1 signal was sometimes used a single ROI for mice. BACE1/tubulin ratios for individual dystrophies or normal neuropil regions are represented by dots in the scatter plots. Mean ratios were calculated for each AD case or mouse and compared with a twotailed t-test, with Welch correction for unequal variance in Prism software.
To generate the intensity plots shown in Fig. 3c, a two-channel BACE1 and tubulin immunofluorescence image tiff file was opened in ImageJ, and the "RGB profiles tool" macro was used to draw a line and measure fluorescence intensity in each channel for each pixel along that line (Fig. 3c, top panel). In Excel, the intensities in arbitrary units (AU) were graphed against distance along the line in microns (Fig. 3c, bottom panel).
Multiphoton confocal microscopy live imaging of BACE1-YFP transgenic mouse brain slices BACE1-YFP transgenic mice were generated as previously described [3]. Briefly, eYFP was fused to the coding region of human BACE1, and cloned into the PMM400 tetO expression vector (gift of M. Mayford, The Scripps Institute, La Jolla, CA). This construct was used to make several tetO promoter BACE1-YFP transgenic mouse lines, which were then crossed to transgenic mice expressing the tet transactivator (tTA) from the forebrain pyramidal neuron-specific CaMKII promoter (line B, gift of M. Mayford) generating CaMKII:BACE1-YFP bigenic mice that express BACE1-YFP in forebrain neurons (referred to as BACE1-YFP mice). Bigenic females from a high expressing BACE1-YFP mouse line (#429) were crossed to 5XFAD males to generate 5XFAD;BACE1-YFP and non-5XFAD BACE1-YFP littermates.
For live multiphoton confocal microscopy imaging, 400µm coronal slices were prepared from adult 5XFAD;BACE1-YFP mice anesthetized with ketamine/xylazine, then perfused with ice cold sucrose-ACSF solution containing: 85 mM NaCl, 2.5 mM KCl, The slices were then maintained at room temperature in a bath perfused with aerated ACSF, and for 1 hour prior to imaging were incubated in aerated ACSF containing 1:20,000 dilution of 1mg/ml Thiazine red (Sigma). For live-cell imaging, each slice was positioned on the microscope stage, perfused with ACSF and maintained at 30°C. Slices were excited at 920 nm with Chameleon Vision titanium sapphire laser (laser range 690-1040 nm) using a Nikon A1R-MP+ multiphoton confocal microscope. Images were acquired using a 25X (NA 1.1) water immersion objective with NIS elements software. Zstacks of images were assembled and 3D reconstructions of neuritic plaques were generated and movies made using NIS elements software.

Dystrophic axons appear in physical contact with amyloid deposits in 3D
reconstructions of live brain slices of BACE1-YFP transgenic mice Our immunofluorescence confocal and electron microscopic analyses suggested that BACE1-positive dystrophic neurites are in close physical proximity to amyloid deposits, while normal neuropil exists a short distance way, thus implying that the dystrophy-promoting neurotoxic effects of Aβ are very short-range. To confirm this conclusion and to establish a model for the study of dystrophic neurite formation in living tissue, we generated a multi-transgenic mouse in which the 5XFAD transgenes are expressed together with transgenic BACE1-YFP fusion protein under control of the forebrain-specific CamKII promoter (5XFAD; BACE1-YFP mice). As we have described previously [12] 5XFAD; BACE1-YFP mice display BACE1-YFP accumulation in dystrophic neurites in a halo pattern around plaques similar to that of endogenous BACE1 in 5XFAD mice (Fig. S3a). We then performed multiphoton confocal microscopy imaging of live 5XFAD; BACE1-YFP brain slices co-stained with Thiazine Red to visualize amyloid deposits. The combination of the thick brain slices together with the large depth of focus inherent in mulitphoton microscopy allowed us to image normal appearing processes as they merged into peri-plaque dystrophies (Fig. S3b).
Additionally, we were able to reconstruct 3-dimensional images of individual dystrophic neurites (Fig. S3c, Video S1) and whole neuritic plaques (Fig. S3d, Video S2). While many BACE1-YFP positive dystrophic neurites appeared as individual bulb-like swellings, possibly due to the very strong tendency of BACE1 to localize to presynaptic terminals [7], some appeared connected together exhibiting a "beads on a string" morphology ( Fig. S3c, Video S1), reminiscent of the neuritic beading observed in Aβ42treated primary neurons (Figs. 1 and 2). Interestingly, all BACE1-YFP positive dystrophic neutites appeared to be in direct physical contact with amyloid deposits at some point along their length, usually at the largest part of the dystrophy, suggesting contact is necessary for dystrophy formation. These observations are consistent with our hypothesis that microtubules and BACE1 trafficking are disrupted in the very near vicinity of plaques.

The potential role of η-secretase cleavage of APP in dystrophic neurite formation
The recently discovered η-secretase processing of APP [14] produces an Nterminal soluble APP ectodomain (sAPPη) that is analogous to that generated by BACE1 (sAPPβ), however η-secretase cuts APP ~92 amino acids N-terminal to the BACE1 cleavage site in APP (Fig. S5). The membrane-bound C-terminal fragment generated by η-secretase is subsequently cut by BACE1 or α-secretase to produce Aη-β or Aη-α, respectively. For assessing BACE1 cleavage of APP in peri-plaque dystrophic neurites (Figs. 8,9), we used the sAPPβ neoepitope antibody (ANJJ) that detects only the BACE1-cleaved free C-terminus of sAPPβ with the Swedish FAD mutation. Therefore, ANJJ does not recognize sAPPη or Aη-α, but ANJJ does not distinguish between sAPPβ and Aη-β in immunostained 5XFAD brain sections (green rectangles, Fig. S5). Note that ANJJ immunoreactivity is abolished in 5XFAD; BACE1-/-negative control mice (Fig. 8a). Despite ANJJ cross-recognition of both sAPPβ and Aη-β, the main conclusion of our study is not changed, namely that BACE1 cleavage is increased in dystrophies surrounding amyloid deposits. ANJJ immunoreactivity serves as a surrogate marker of BACE1 activity. Since we observe increased ANJJ immunostaining in peri-plaque dystrophies, we conclude that BACE1 activity is elevated in these dystrophies. In support of this conclusion, we also observe increased immunoreactivity for the free N-and C-terminal neoepitopes of Aβ, indicating that Aβ generation is also elevated in peri-plaque dystrophies. Even though we cannot distinguish between sAPPβ and Aη-β, increased ANJJ signal indicates elevated BACE1 activity in peri-plaque dystrophies. We also note that Aη-β may be increased in BACE1-positive peri-plaque dystrophies. However, in contrast to Aη-α, Aη-β does not impair LTP or suppress neuronal activity [14]. Since Aη-α would not be expected to increase (and may actually decrease) with elevated BACE1 in dystrophic neurites, the η-secretase pathway may not have a large role in toxicity associated with neuritic dystrophy. However, a potential role for Aη in dystrophic neurite formation will be an interesting avenue to pursue in future studies. Tau Tau aggregates are present around senile plaques, mainly in the form of neuropil threads and neurofibrillary tangles consisting of paired helical filaments of Tau (reviewed in [2]). Neuropil threads and neurofibrillary tangles are likely dendrite and cell body in origin, respectively, and appear to be distinct from swollen dystrophies that appear to be axonal in nature [4,5,15]. Thus, we suggest that swollen BACE1-positive peri-plaque dystrophies are presynaptic in origin and mostly lack Tau aggregates, while dendritic dystrophies mainly exhibit Tau aggregates. This conclusion is supported by the observation in this article and others [7,18,19] that BACE1-positive dystrophies colocalize significantly with the synaptic vesicle protein synaptophysin but completely lack the somatodendritic marker MAP2 (Fig. 6, Fig. S3), and that Aβ causes Tau to mis-sort into the dendrite [16,17].  Table S1: List of antibodies, dilutions, and sources used in this study.  which proteins are plaque-associated and co-localize with BACE1 in dystrophies in 5XFAD mice. All antibodies were used in a co-stain with anti-BACE1 antibody.    floating sections from two 6-month old male, and two 6-month old female mice were immunostained as described in Methods with the antibodies listed in Table S3 along with either anti-BACE1 mouse monoclonal 3D5 or anti-BACE1 rabbit monoclonal (EPR3956) at the concentrations listed in Table S1. Immunostaining for BACE1 is shown in red, other proteins in green, and DAPI stain for nuclei is shown in blue. Plaque core autofluorescence in blue. All images were collected on Nikon A1 confocal microscope  Together, these results suggest that direct physical contact with the amyloid plaque may cause dystrophy formation. Scale bars = 5µm in all frames.  In this diagram, adapted from [14], we show that the ANJJ antibody [9,10], which recognizes the free C-terminal neoepitope of sAPPβ (green vertical line) created by BACE1 cleavage of APP carrying the Swedish mutation, will detect Aη-β and sAPPβ (green rectangles) but not sAPPη or Aη-α. Since ANJJ detects only BACE1 cleaved fragments, it is an effective surrogate marker of BACE1/β-secretase activity.

Supplementary Videos:
Video S1 and S2: Movies of 3D reconstructions of BACE1-YFP positive dystrophic neurites. Live 5XFAD; BACE1-YFP brain slices were imaged by multiphoton confocal microscopy as described in