Phagocytosis of dying cells is critical in development and immunity1–3. While proteins for recognition and engulfment of cellular debris following cell death are known4,5, proteins that directly mediate phagosome sealing are uncharacterized. Furthermore, whether all phagocytic targets are cleared using the same machinery is unclear. Degeneration of morphologically-complex cells, such as neurons, glia, and melanocytes, produces phagocytic targets of various shapes and sizes located in different microenvironments6,7. Such cells, therefore, offer unique settings to explore engulfment program mechanisms and specificity. Here we report that dismantling and clearance of a morphologically-complex C. elegans epithelial cell requires separate cell-soma, proximal-, and distal-process programs. Similar compartment-specific events govern elimination of a C. elegans neuron. While canonical engulfment proteins drive cell-soma clearance, these are not required for process removal. We find that EFF-1, a protein previously implicated in cell-cell fusion8, specifically promotes distal-process phagocytosis. EFF-1 localizes to phagocyte pseudopod tips, and acts exoplasmically to drive phagosome sealing. eff-1 mutations result in phagocytosis arrest with unsealed phagosomes. Our studies suggest universal mechanisms for dismantling morphologically-complex cells, and uncover a phagosome sealing component promoting cell-process clearance.

The Caenorhabditis elegans cell lineage is nearly invariant. Whether this stereotyped cell-division pattern promotes reproducibility in cell shapes/positions is not generally known, as manual spatiotemporal cell-shape/position alignments are labor-intensive, and fully-automated methods are not described. Here, we report automated algorithms for spatiotemporal alignments of C. elegans embryos from pre-morphogenesis to motor-activity initiation. We use sparsely-labeled green-fluorescent nuclei and a pan-nuclear red-fluorescent reporter to register consecutive imaging time points and compare embryos. Using our method, we monitor early assembly of the nerve-ring (NR) brain neuropil. While NR pioneer neurons exhibit reproducible growth kinetics and axon positions, cell-body placements are variable. Thus, pioneer-neuron axon locations, but not cell-body positions, are under selection. We also show that anterior NR displacement in cam-1/ROR Wnt-receptor mutants is not an early NR assembly defect. Our results demonstrate the utility of automated spatiotemporal alignments of C. elegans embryos, and uncover key principles guiding nervous-system development in this animal.