Neurons were traced in a serial section transmission electron microscopy (ssTEM) volume of a full adult female D. melanogaster brain (FAFB) (Zheng et al., 2018 (link)) using CATMAID, a web-based software for collaborative neural circuit reconstruction from large image datasets (https://catmaid.readthedocs.io/en/stable/ ) (Saalfeld et al., 2009 (link), Schneider-Mizell et al., 2016 ). Consistent with previous studies (Eichler et al., 2017 (link), Schlegel et al., 2016 , Schneider-Mizell et al., 2016 ), tracing followed the centerline of a neuron’s profiles through the dataset to reconstruct neurite morphology and annotate synaptic sites. We used an iterative approach established and tested by Schneider-Mizell et al. (2016) , where initial reconstruction is followed by a systematic proofreading by at least two experienced reviewers (> 500h of tracing experience).
MBON identification: MBONs were located by sampling downstream of previously identified KC synapses in the respective mushroom body lobe compartments. Their identity was confirmed by comparison with light level data (Aso et al., 2014b (link)).
Synapse annotation: Synaptic sites were identified based on three, previously described criteria (Prokop and Meinertzhagen, 2006 (link)) and reviewed as above: an active zone with (1) T-bar(s) and (2) surrounding vesicle cloud, and (3) a synaptic cleft to which all postsynaptic neurons must have access.
In Drosophila, presynapses have been found on fine axonal processes (Schneider-Mizell et al., 2016 ), boutons (Butcher et al., 2012 (link)), and other neurites that are neither in the dendritic nor the axonal field. Post-synapses have been found on large or fine dendritic processes and fine spine like twigs that are shorter than 3μm (Schneider-Mizell et al., 2016 ). M4β′ M6R and M6L were reconstructed the same way to maintain consistency in the placement of synapses. Schneider-Mizell et al. (2016) estimated that the tracing approach employed typically finds 99.8% of all pre- and 91.7% of all post-synapses. The probability of identifying false-positive post-synapses is 2.2% and negligible for presynapses. Since all synaptic sites on the MVP2 axon and M4β′ and M6 dendrites were annotated, we can estimate the upper and lower bounds of the number of synapses between MVP2 and M4β′ or M6 neurons (Note only integer numbers of synapses are expected):
Reconstructed neurons were visualized using Blender 3D, an open-source 3D software (https://www.blender.org/ ). Neuron data from CATMAID were imported and shaded by Strahler order using an existing CATMAID plugin for Blender (https://github.com/schlegelp/CATMAID-to-Blender ; Schlegel et al., 2016 ).
Volumetric reconstruction of synapse architecture was achieved by importing and annotating FAFB image data into ImageJ using the TrakEM2 plugin (Cardona et al., 2012 (link)). Reconstructions were exported for rendering to Blender 3D.
Analysis: All analyses were performed in R and Python using open-source software. PyMaid (https://github.com/schlegelp/PyMaid ) and RCatmaid (http://jefferis.github.io/rcatmaid /; http://jefferis.github.io/elmr/ ) were used to interface with CATMAID servers and perform morphological analyses. Dendrogram representations of neural arbors were generated using new code (https://github.com/markuspleijzier/AdultEM/tree/master/Dendrogram_code ) the graphviz library (https://graphviz.gitlab.io/ ; Gansner and North, 2000 via Python bindings provided by NetworkX, https://networkx.github.io/ ; Hagberg et al., 2008 ). M4β′ axonlets were defined as distal parts of neurites originating from the dendritic field, which made exclusively presynaptic connections. Axonlets were isolated and imported into Blender 3D using PyMaid. The root of the dendritic field was defined as the point at which the neuron’s main neurite branched into proximal dendrites and distal axon. Geodesic (along the arbor) distances between synapses and dendritic root were calculated using RCatmaid.
MBON identification: MBONs were located by sampling downstream of previously identified KC synapses in the respective mushroom body lobe compartments. Their identity was confirmed by comparison with light level data (Aso et al., 2014b (link)).
Synapse annotation: Synaptic sites were identified based on three, previously described criteria (Prokop and Meinertzhagen, 2006 (link)) and reviewed as above: an active zone with (1) T-bar(s) and (2) surrounding vesicle cloud, and (3) a synaptic cleft to which all postsynaptic neurons must have access.
In Drosophila, presynapses have been found on fine axonal processes (Schneider-Mizell et al., 2016 ), boutons (Butcher et al., 2012 (link)), and other neurites that are neither in the dendritic nor the axonal field. Post-synapses have been found on large or fine dendritic processes and fine spine like twigs that are shorter than 3μm (Schneider-Mizell et al., 2016 ). M4β′ M6R and M6L were reconstructed the same way to maintain consistency in the placement of synapses. Schneider-Mizell et al. (2016) estimated that the tracing approach employed typically finds 99.8% of all pre- and 91.7% of all post-synapses. The probability of identifying false-positive post-synapses is 2.2% and negligible for presynapses. Since all synaptic sites on the MVP2 axon and M4β′ and M6 dendrites were annotated, we can estimate the upper and lower bounds of the number of synapses between MVP2 and M4β′ or M6 neurons (Note only integer numbers of synapses are expected):
| Found | Lower Bound | Upper Bound | |
|---|---|---|---|
| MVP2- > M6R | 17 | 16.6 | 18.4 |
| MVP2- > M6L | 16 | 15.6 | 17.3 |
| MVP2- > M4β' | 47 | 46 | 50.8 |
Reconstructed neurons were visualized using Blender 3D, an open-source 3D software (
Volumetric reconstruction of synapse architecture was achieved by importing and annotating FAFB image data into ImageJ using the TrakEM2 plugin (Cardona et al., 2012 (link)). Reconstructions were exported for rendering to Blender 3D.
Analysis: All analyses were performed in R and Python using open-source software. PyMaid (
Full text: Click here