Proapoptotic RHG proteins and mitochondria are required for sensory dendrite pruning in Drosophila

Caspases are best known for their role in apoptosis, but they are also deployed sub-lethally during development, where they contribute to cell fate specification, migration and terminal differentiation. How neurons activate executioner caspases without dying remains unresolved. At the onset of Drosophila metamorphosis, neighbouring sensory neurons in the dorsal body wall adopt opposite fates: class III dendritic arborization (da) neurons die by apoptosis, while the class IV da neuron ddaC prunes its larval dendrites before regrowing an adult-specific arborization.

Their proximity allows lethal and non-lethal caspase signalling to be compared side by side in an intact animal. Using SR4VH, a sensitive dual-colour executioner caspase probe, we show that caspase substrates are cleaved in ddaC within 20 min of puparium formation, well before the proximal dendrites are severed, and at consistently lower levels than in its dying neighbours. Single-cell H99 mutant clones reveal that the proapoptotic RHG proteins are cell-autonomously required for pruning.

Although reaper, hid and grim are all transcribed in ddaC as remodelling begins, mutant combinations suggest that reaper and grim, rather than hid, cooperate to drive branch severing and clearance. In addition, disrupting mitochondrial function or transport depletes dendritic mitochondria, prevents caspase activation, and blocks both pruning in ddaC and apoptosis in neighbouring class III neurons. Taken together, our data show that Reaper and Grim are required cell-autonomously for pruning, and that mitochondria are required for both caspase activation and branch severing.

We would like to thank Nicolas Loncle for his initial observations on caspase activation in sensory neurons with SR4VH. We would like to thank Bloomington Stock Centre, Tadashi Uemura, Kristin White, Joe Bateman and Alex Whitworth for providing several fly lines. Thanks to FengWei Yu for providing us with the antibody against Sox 14.

Thanks to Paul Conduit for his generosity in providing reagents, imaging facility and time to generate some key data for the manuscript. Lajos Kalmar for his help with the statistics. Many thanks to Joe Bateman, Alberto Baena-Lopez and Jon Clarke for reading and giving feedback on the manuscript.

Finally, we thank all the Williams lab members for their help and support. This work was supported by the BBSRC funding from grants BB/P025552/1, BB/L022672/1 and BB/T013869/1. Centre for Developmental Neurobiology, King’s College London, London, SE1 1UL, UK Amrita Mukherjee, Ryan Pei-Yen Cheng, Connor J.