Data for Revised Manuscript

The contribution of microglia to neuronal circuit refinement and pruning during development has been the subject of intensive investigation in the past few years. Numerous lines of evidence imply that microglia may actively phagocytose bits of developing neuronal processes, including synapses, through a behavior that has been described as “nibbling” or trogocytosis. Influential electron micrographic studies have found probable synaptic material localized within microglial phagocytic compartments. However, as microglia are well known to ravenously remove debris left behind by apoptotic cells in the developing CNS, such evidence from snapshots of fixed microglia is far from conclusive. Other studies have shown circuit development defects in transgenic animals in which microglial phagocytic activity has been suppressed. But given the many signaling molecules known to be released by microglia, it is difficult to know with certainty whether the defect reflects the participation of myeloid cells in synaptic trogocytosis or in more conventional local interactions involving the release of chemical signals. Finally, ex vivo slice culture preparations have offered more direct evidence for synaptic nibbling, but the highly invasive process of preparing such brain slices may not accurately reflect the physiological activity of microglia in a healthy developing brain.

The retinotectal system of the Xenopus laevis tadpole is a classic model for studying the activity-dependent refinement of the developing visual system due to the ease with which in vivo imaging and visually-evoked behavioral studies can be performed in intact translucent tadpoles without the need for any inflammation-inducing surgical procedures. In the current study, we have exploited this simple in vivo model to directly follow microglia and neurons over many hours as they interact within the developing retinotectal circuit. Our results include direct observation and quantification of the uptake of synaptic material from healthy, remodeling axons by microglia over time. We further went on to test the consequences of microglial depletion in circuit development and found evidence for both structural exuberance as well as the disruption of visually-mediated behaviors. Finally, we examined the hypothesis that complement C3 may in part underlie the microglia-mediated structural remodeling of retinotectal axons by showing that the endogenously expressed complement inhibitor protein CD46 appears to promote arbor elaboration whereas decorating axons with C3 causes a failure to ramify arbors within the optic tectum.
This in vivo study is the first, to our knowledge, to directly observe and quantify the trogocytosis of presynaptic material by microglia in the intact developing central nervous system. We introduce a novel and highly efficient assay for testing visuomotor behaviors in Xenopus tadpoles. And our study identifies complement C3 and its inhibition by the complement regulator protein CD46 as important regulators of axonal pruning by microglia in circuit development. We believe that this work constitutes a major advance both by conclusively demonstrating and by mechanistically investigating how microglia participate in circuit refinement during development.