(A,B) In vivo GCaMP6f signals recorded in layers M1, M5 and M9/10 of Mi1 (A) and Tm3 (B) neurons, before (blue, green) and after (gray, red) application of 0 (sham), 1, 5 or 100 µM PTX. (C,D) Bar plot showing the quantification of the ON step in (A,B). Sample sizes: sham, n = 5 (89); 1 µM, n = 5 (68); 5 µM, n = 5 (64); 100 µM, n = 5 (89). All traces show mean ± SEM. All sample sizes are given as number of flies (number of cells). *: p<0.05, **: p<0.01, ***: p<0.001, tested with a one-way ANOVA and a post-hoc unpaired t-test with Bonferroni-Holm correction for multiple comparisons

Sensory systems sequentially extract increasingly complex features. ON and OFF pathways, for example, encode increases or decreases of a stimulus from a common input. This ON/OFF pathway split is thought to occur at individual synaptic connections through a sign-inverting synapse in one of the pathways. Here, we show that ON selectivity is a multisynaptic process in the Drosophila visual system. A pharmacogenetics approach demonstrates that both glutamatergic inhibition through GluClα and GABAergic inhibition through Rdl mediate ON responses. Although neurons postsynaptic to the glutamatergic ON pathway input L1 lose all responses in GluClα mutants, they are resistant to a cell-type-specific loss of GluClα. This shows that ON selectivity is distributed across multiple synapses, and raises the possibility that cell-type-specific manipulations might reveal similar strategies in other sensory systems. Thus, sensory coding is more distributed than predicted by simple circuit motifs, allowing for robust neural processing.
eLife digest We rely on our senses to capture information about the world around us. Sense organs convert sensory information – such as light or sound waves – into patterns of neuronal activity. In the mammalian retina, for example, specialized neurons called photoreceptors detect individual photons of light as they hit the back of the eye. The photoreceptors then pass on this information to neurons called bipolar cells for further processing. During darkness, all photoreceptors release the same chemical signal onto bipolar cells, namely a molecule called glutamate. But bipolar cells respond to glutamate in different ways depending on which proteins are present in their outer membrane. So-called ON cells respond to glutamate by decreasing their activity, and thus effectively become more active when light levels increase. By contrast, OFF cells respond to glutamate by increasing their activity. This ON/OFF binary code enables later stages of the visual system to detect more complex visual features, such as shape and movement. A new study in fruit flies, however, suggests that the ON/OFF code may be more complex than previously thought. While fruit fly eyes look very different to our own, the two have much in common. By studying fruit flies, researchers can also take advantage of a variety of genetic and pharmacological tools to manipulate cells and neuronal circuits. Using such tools, Molina-Obando et al. show that the ON/OFF signal separation in fruit flies uses two different molecular mechanisms. The first involves a gene called GluCl-alpha, which encodes a receptor for glutamate. The second involves a gene called Rdl, which encodes a receptor for another brain chemical, GABA. Deleting the gene for GluCl-alpha from the entire fly brain prevented ON cells from responding to an increase in light levels. However, deleting this gene from specific ON cells alone did not. This suggests that flies can use more than one type of neuronal connection to detect an increase in light. Moreover, if one pathway fails, the other can take over. This makes the system more robust. The results of Molina-Obando et al. are consistent with findings from anatomical studies that have mapped connections between neurons. Future studies should explore whether the same mechanisms exist in other sensory systems, and other animals. These experiments could take advantage of the molecular tools developed as part of the current work, which allow precise manipulation of neural networks.