812A Poster - 13. Neural development and physiology
Thursday April 07, 2:00 PM - 4:00 PM

Authors:
Egemen Agi ¹; Charlotte B. Wit ¹; Eric Reifenstein ^2,3; Max von Kleist ^2,3; P. Robin Hiesinger ¹

Affiliations:
1) Division of Neurobiology, Institute for Biology, Freie Universitaet Berlin, Germany; 2) Systems Medicine of Infectious Disease (P5), Robert Koch Institute, Berlin, Germany; 3) Bioinformatics (MF1), Robert Koch Institute, Berlin, Germany

Keywords:
a. axon guidance; q. vision

The fly visual map represents ~800 neighboring points in space as synaptic ensembles (cartridges) in the lamina of the fly brain. These 800 visual axes are represented by six photoreceptor axons (R1-R6) in each cartridge, where R1-R6 provide pooled presynaptic input onto their main postsynaptic partner lamina cells L1-L3. Due to the arrangement of separate light-sensing elements that receive input from different visual axes in each single ommatidium of the fly eye, R1-R6 axons from six different ommatidia have to grow towards a shared cartridge in the lamina, a principle called 'neural superposition'. In effect, six times 800 R1-R6 axons synchronously grow away from their original arrival points in the lamina by elongating their growth cones orthogonally to their axons in a sorting plane. At the time of growth cone sorting, the postsynaptic L cells are present in the lamina in an apparent target grid. Surprisingly, genetic ablation of L cells does not affect early pattern formation, growth cone elongation nor the initial formation of the neural superposition pattern. Only after superposition patterning through R1-6 growth cones alone, the fronts of all elongated growth cones are stabilized by the 'target grid' in an N-Cadherin-dependent manner. To understand how initial neural superposition pattern formation is established by six times 800 growth cones in the absence of any other cell type, we performed non-invasive two-color two-photon live imaging and developed a computational model of the process based on photoreceptor filopodial dynamics. Both data and model reveal a dynamic pattern formed by the 'heels' (back ends) of the elongating growth cones with shifting membrane densities throughout the sorting process. Front filopodia predominantly sample the space between the highest heel densities, i.e. the valleys of the heel density landscape. In turn, the weighted directions and lifetimes of front filopodia predict the subsequent growth cone elongation vectors. We conclude that six times 800 photoreceptor growth cones self-organize into the neural superposition pattern by separating growth cone front and heel patterns, with the former appearing to chase the latter. This self-organization process can be explained and modeled based on a minimal number of guidance cues by largely relying on intrinsic and dynamic cell and tissue properties.