Neurites

TrakEM2 has been written using the Java programming language and uses numerous image processing libraries including ImageJ (Wayne Rasband), mpicbg (Stephan Saalfeld), LOCI bio-formats [37] (link), ImgLib (Stephan Preibisch, Stephan Saalfeld, Tobias Pietzsch and others), ImageJ 3D Viewer [38] (link), Stitching [39] (link), bUnwarpJ [40] , JaMa (Mathworks and NIST), postgresql-jdbc, JFreeChart (jfree.org), edu_mines_jtk (Dave Hale), Level Sets (Erwin Frise) and Simple Neurite Tracer [41] , among others. The source code is released under the General Public License and is under version control with git at http://repo.or.cz/w/TrakEM2.git. Binaries are distributed with Fiji (Schindelin et al, submitted to Nature Methods) via the automatic plugin updater.

Tip tracking was performed using Manual Tracking in NIH ImageJ (FIJI build [Schindelin et al., 2012 (link)]). DF were only tracked if they met the following conditions: no contact with axons, neighboring DF, or debris during the time course; emanated from dendrites at least 50 µm away from the center of the soma; clearly visible by brightfield during time course; if they initiated or retracted during imaging, non-existent timepoints were removed from further analysis; buckling and wagging DF were included in tracking. Using the manually tracked positions of the DF base and tip, the image files were then further analyzed with a custom MATLAB script to determine the centerline path along each DF (Mendeley data hyperlink). This script used the fluorescent intensity in either the LifeAct or GFP space-filler channel in the vicinity of the tip and base coordinates to define the average tangent direction of the long axis of the DF by computing the tangent angle q at pixel i using ${\theta }_i=\frac{1}{2}{\tan }^{-1}\left(\frac{\langle y-y_i\rangle \langle x-x_i\rangle }{\langle x^2-x_i^2\rangle \langle y^2-y_i^2\rangle }\right)\text{,}$ where brackets denote the intensity-weighted average over a 15 × 15 pixel domain centered on the ith pixel. The centerline curve (x(s),y(s)) was then determined by solving $\frac{{\partial }^{2}x}{\partial s^2}=-\sin \theta \frac{\partial \theta }{\partial s};\frac{{\partial }^{2}y}{\partial s^2}=\cos \theta \frac{\partial \theta }{\partial s}$
subject to the constraint that the starting and ending positions were the tracked positions of the base and tip of the DF. Using the centerline curves for each DF at each time point, we then calculated the absolute tip displacement, DF length, and mean tip fluorescence intensity and were able to extract the following metrics: average filopodial tip speed calculated as the average of the instantaneous speeds (absolute tip displacement per 5 s interval) between successive timepoints; percent motile, percent of total DF population with average tip speeds greater than 0.0128 µm/s (motile; one pixel displacement or greater per 5 s interval) or less than 0.0128 µm/s (non-motile); percent time motile, the percent of time per DF in which instantaneous speed was greater than 0.0128 µm/s; average length, the distance from base to tip along the centerline curve, median protrusion or retraction rate, the positive or negative change in length between successive timepoints, when instantaneous change in length was greater than ±0.0128 µm/s (motile); mean fluorescence intensity for a circular area of 384 nm radius surrounding the distal DF tip with non-cell background omitted; fluorescence intensity variance, a measure of the spread of intensity values compared to the mean. Fluorescence intensity values were normalized for expression by the minimum local intensity during the duration of imaging. For defining motile versus non-motile filopodia, or substantiative protrusion/retraction rates, a threshold of 0.0128 µm/s was chosen as it represents one pixel (effective size at 100X = 0.064 µm) displacement per 5-s interval and undistinguishable from tracking error. Neurite morphology was measured using the ImageJ plug-in Simple Neurite Tracer (Longair et al., 2011 (link)). Tracings were used to determine the number and length of primary and higher order neurites, and length of the axon (the longest Tau-positive process). Protrusion and spine density was determined by counting proturbences or dendritic spines along a length of dendrite. PSD95 foci analysis was performed by generating a binary mask of foci, and using the automated 2D tracking module in NIS-Elements (Nikon) to follow their trajectories.

DITNC1 astrocytes were seeded at a density of 5 × 10⁵ cells/cm² and incubated to 90% confluency in 6-well glass bottom plates. Then, the cells were treated with OMVs (2.5 μg/ml) or TNF (10 ng/ml) for 12 h or 48 h, respectively. Next, the medium was eliminated, and cells were washed twice with PBS and fixed with 4% paraformaldehyde for 15 min. The reaction was then stopped with 0.1 mM glycine. Following extensive washing with PBS, CAD cells labeled with Cell Tracker Green CMFDA (10 µM) were seeded over fixed astrocytes at a density of 1 × 10⁴ cell/cm² in DMEM F12 media, supplemented with 8% FBS and antibiotics [55 (link)]. Before 24 h of culture, the media was replaced with fresh SFM containing 50 ng/ml sodium selenite (S5261, Sigma-Aldrich) to induce morphologically visible neuronal differentiation. After 24 h, neuronal processes were captured with an Epifluorescent Spinning disk microscope (Olympus). The neurite extension length was determined using the NeuronJ plug-in of the v1.8 NIH ImageJ Software, as previously described [56 (link), 57 (link)].

SH-SY5Y cells were seeded 24 h prior to differentiation in 24-well plate to a density of 1 × 10³ cells per well. Cells were differentiated in B-27™ Plus neuronal culture system (Life Technologies) supplemented with 20 µM retinoic acid (Merck Life Science), 1% L-Glutamine (Gibco), and 1% penicillin/streptomycin (Gibco). Medium was refreshed every other day. Phase-contrast imaging was done on a Zeiss Axiovert 200 M microscope equipped with a Zeiss AxioCam MR3 camera and 20× phase contrast objective. Three images per well were captured at day 8 of differentiation. To extract the total area covered with neurites and soma in each image, we used a custom-developed ImageJ script to automatically segment both neurites and soma, using combinations of simple image operations that can i) remove noise (noise reduction filters), ii) separate fine from coarse structures (rolling ball algorithm or Fast Fourier Transform), iii) separate bright from dark regions (automatic intensity thresholding) and iv) exclude segmented regions based on size or shape (morphological operations). The resulting segmentation masks were used to calculate the ratio of skeletonized neurites per cell body area. In addition, we manually traced individual neurite structures. To this end, images were converted to 8-bit and analyzed with NeuronJ plugin in ImageJ, a commonly used tool for semiautomatic tracings and measurements of neurites^{67 (link)}. Any projection from SH-SY5Y cell body was considered a “primary neurite”, whereas projections branching from primary neurites were considered a “secondary neurite”. Three wells per genotype and three images per well were analyzed, and data from two experiments (repetition) was pooled, resulting in over 600 tracings in total. The overall distribution of primary and secondary neurites lengths was plotted. For each image, the fraction of secondary neurites (over the total) was also calculated. One-Way Anova was used for statistical analysis.