Relación malestar docente – profesionalización

Stacks of serial confocal sections (0.32 -1µm) were acquired using a Zeiss LSM 710 with 20x(air), 40x(water) and 63x(oil) objectives. An automated table and MultiTime2010 (Zeiss) allowed us to scan multiple locations automatically, in which case 40x (with halogen-free immersion medium, n=1.33) and 63x (n=1.52, oil) objectives were used.

Processing of confocal stacks included maximum density projections, substack projections, orthogonal views, rotations, changes in channel hue and adjustment of brightness and contrast. Confocal images were processed with ImageJ (Abramoff, 2004; Rasband, 1997-2012); 3D reconstructions were created in Imaris (Bitplane). All figures and illustrations were prepared in Adobe Illustrator CS5 version15.0.0.

Counting of nuclei within confocal stacks was carried out as described in Figure 30 using a home-written Definiens XD 2.0 script. Subsequent statistical analysis was done in Excel and finished in GraphPadPrism. Considerations regarding coverage of single generic drivers and additional use of region-specific drivers to count the glial cells of on subtype in the entire brain are explained earlier (also see 6.1.1). The structure density of astrocyte-like glia and synapses in different brain regions was determined and correlated as described in Figure 31 using a home- written Definiens XD 2.0 script. To determine the average density of astrocyte-like glia density processes per brain neuropile, we manually rated the densities for every neuropile as is explained in Figure 32.

 MATERIAL AND METHODS  

Figure 30: Automated counting of immune-histochemically labeled glial cell bodies.

Confocal stacks of brains, in which glial cell bodies were labeled using a nuclear GFP reporter (example here: 28A04Gal4; UASnls-GFP) and anti-REPO AB and neuronal cell bodies were co-labeled with anti-ELAV AB, were processed as follows: The contours of the brain were identified based on the ELAV signal in channel one, the entire set of glial cell bodies was identified based on the REPO signal in channel two, subtype specific glial cell bodies were identified based on the nuclear GFP signal in channel three. Counting was carried out in 3D, however, in order to avoid over-counting, cell bodies were only counted in the z-layer with the highest density. To detect cell bodies, we applied the following strategy. In the raw images (A), which have a size of 1024x1024 pixels with a resolution of 692 nm per pixel in xy and a z slice of 1µm, the signal intensity of GFP- and REPO-labeled nuclei bodies is high, but the considerable variation in background intensity makes a precise identification of cells difficult. Therefore, using a global threshold does not produce good segmentation. Instead, we developed a strategy based on background reduction. As a first step, we applied a 3D-Gaussian filter with a kernel size of 5x5x3 pixels (B, Gaussian filter (1)), then applied a second 3D-Gaussian filter, again with a kernel size of 5x5x3 pixels, (C, Gaussian Filter (2)), and then subtracted C

MATERIAL AND METHODS  

threshold and carried out segmentation using an algorithm implemented in the Definiens XD 2.0 software platform, as shown in E. Briefly, the so-called Multi-Threshold Segmentation algorithm used splits the image domain and classifies the resulting image objects based on a defined pixel value threshold. F shows the segmentation patterns (magenta lines) superimposed on the raw data. Despite strong background heterogeneities, as seen on the right, cells are readily identified (G). The 3D segmentation patterns are displayed for an exemplary confocal plane of a full Drosophila brain. Finally, the data were exported and statistically evaluated in Microsoft Excel and GraphPathPrism.

 MATERIAL AND METHODS  

Figure 31: Automated analysis of the astrocyte-like glia-processes density.

The structure density of astrocyte-like glia processes (ALG channel) and synapses (nc82 channel) were determined and correlated using a home-written Definiens XD 2.0 script. The analysis strategy is illustrated for the ALG channel. First, the astrocyte-like glia channel was segmented similarly to the procedure described in Figure 30. As a first step, we applied a 3D-Gaussian filter with a kernel size of 5x5x3 pixels (Figure 30B), then applied a second 3D-Gaussian filter, again with a kernel size of 5x5x3 pixels (Figure 30C), and then subtracted C from B, which resulted in a background subtracted image (Figure 30D). As a last step, we applied a global threshold and carried out segmentation using an algorithm implemented in the Definiens XD 2.0 software platform. Briefly, the so-called Multi-Threshold Segmentation algorithm used splits the image domain and classifies the resulting image objects based on a defined pixel value threshold. A shows an exemplary confocal slice. The segmented patterns of the brain (in green) as well as of the astrocyte-like glia cells (in red) are superimposed in B. The astrocyte-like glia cell density was subsequently determined as illustrated in the insert of C. For each individual cell the total area of surrounding cells was used as a measurement of the local cell density. For each single image object of interest (in blue) we computed the total area in pixels (in yellow) of the neighboring objects present in a radius of 30 pixels around the center of mass of the image object of interest (region delimited by the blue circle). The image objects were then color-coded

MATERIAL AND METHODS  

according to their density (C), revealing regions with different ALG coverage. Finally, the procedure was reiterated with the nc82 channel.

 MATERIAL AND METHODS  

Figure 32: Annotation of astrocyte-like glia processes in different brain regions.

A.1 The density of astrocyte-like glia processes differs in different brain regions (single frontal section of the anterior protocerebrum). A.2 The density of NC82-labeled synapses differs in different brain regions. Within each brain regions, the variation in density level is low (single frontal section of the anterior protocerebrum).B.1 The density of astrocyte-like glia processes (A.1) was analyzed in Fiji as follows: The astrocyte-like glia signal was filtered using the 3D Gaussian blur filter function with a Kernel size of 4x4x4. Subsequently, the displayed greyscale image was transformed into a 16-color heat map (using the lookup table function) and analyzed. B.2 Previously defined brain regions (Insect Brain Name Working Group, unpublished data) were identified using the annotated mask kindly provided by Arnim Jenett, JFRC. B.1- 2 Matching between the brain region (identified in the annotation channel, B.2) and astrocyte-like glia processes density (A.1, B.1) was performed manually region by region and revealed the astrocyte-like glia densities per brain region shown in Table 2 in the main text. Scale bar = 50um in all images.