Identifying neural circuits for visual pattern detection in flying Drosophila via activation and inactivation screening

Abstract

Many animals, including humans and fruit flies, use their vision to detect external objects and to respond appropriately. A neural circuit for visual pattern detection, however, is not precisely defined, mainly because of the sheer complexity of visual systems. The visual system of Drosophila melanogaster provides a useful model system for the visual pattern detection for its conciseness: Fruit flies show a relatively large repertoire of visual behaviors with only several tens of thousands of neurons. In this study, we investigated how fruit flies detect visual patterns, focusing on a set of higher-order visual structures, termed optic glomeruli, which consists of ~20 discrete neuropils. We performed large scale inactivation and activation screening experiments to identify the role of individual optic glomeruli in the visually induced flight maneuvers. For the inactivation screening, we implemented a thermogenetic strategy by using highly specific split-GAL4 lines that drives expression of a temperature-dependent synaptic transmission blocking reagent, called Shibirets, in a class of visual projection neurons innervating a single glomerulus[1]. We measured the change in visually triggered wing responses, to 16 different visual patterns from ~1,000 flies, at permissive (26.5 oC) and restrictive (31.5 oC) temperatures. We identified distinct but overlapping sets of optic glomeruli –– LC11, LC12, LC15, LC17 and LC22 for a horizontally moving vertical bar; LC10, LC13, LC16, LC17 and LPLC2 for a horizontally moving small spot, for example –– that reduce wing responses to different visual patterns. For the activation screening, we implemented an asymmetric optogenetic activation strategy, in which we activated a set of visual projection neurons on the right hemishphere, by using a red-shifted version of channelrhodopsin and limiting the 600nm light pulses to one side of the brain. We found that two optic glomeruli –– LPLC2 and LC22 –– when activated, cause strongly aversion from the activated side, whereas one optic glomerulus, LC10, causes a strong attraction toward the activated side. Combining the inactivation and activation screening results, we found that LPLC2 neurons appear to mediate fly’s strong aversion to a horizontally moving small object. The inactivation and activation results for other glomeruli, including LC10, LC11, LC15, LC16 and LC22, however, were not readily interpretable, either because the phenotype was observed only in one type of experiments or because the two results were seemingly contradictory. Together, our results suggest that some glomeruli such as LPLC2 serve a conspicuous function as a labeled line, whereas visual signaling in the majority of glomeruli are further processed by the downstream circuit nonlinearly leading to a specific flight maneuver.