Eaters of the dead

The phagocytic removal of apoptotic cells is essential for proper development and tissue homeostasis. A failure to properly remove cell corpses can lead to an inflammatory response and eventually autoimmunity. Ironically, almost nothing was known about how the massive number of neuronal corpses are scavenged during the development of the peripheral nervous system (PNS), despite the fact that apoptosis was first described in this system some 60 years ago.¹ Programmed cell death has been well characterized as a normal pruning process during the development of the vertebrate nervous system, with nearly 50% of the neurons generated eventually undergoing apoptosis. Because neuronal death contributes to many neuropathologies, the mechanisms regulating this process, both in development and disease, have been extensively studied. However, much less is known about how the dead neurons are removed. In the central nervous system microglia, the resident macrophages, are the primary phagocytes clearing the dead neurons,² but the mechanisms for corpse removal in the periphery remained a mystery.

In a recent study, we investigated the cellular and molecular mechanisms responsible for clearing dead neurons in the developing dorsal root ganglia (DRG).³ Since macrophages are professional phagocytes and share a common origin with microglia, we expected these to be the cells vacuuming up the neuronal debris; however, there were only a few sporadic macrophages in the ganglia, even at the peak of developmental apoptosis. Instead, glial precursors were the primary cell type engulfing the apoptotic neurons. Nearly 90% of the degenerating cells were surrounded by a glial precursor at embryonic day 11 (E11). Even in mice lacking neurotrophin 3, which lose 70% of their DRG neurons, glial cells were the predominant phagocytes.

The glial cells in the ganglia at E11 are primarily satellite glial cell (SGC) precursors, which are likely a common precursor for satellite cells and Schwann cells, although exactly when these glial lineages diverge is not clear. Interestingly, mature Schwann cells contribute to the clearance of axonal and myelin debris following a nerve injury⁴ and have been implicated in axonal pruning during the formation of the neuromuscular junction.⁵ Hence, the precursors and the mature Schwann cells appear to share a phagocytic ability. In contrast, the role of satellite glia in the adult DRG is not well understood. They are largely thought to buffer the microenvironment around the neuron, but whether they scavenge apoptotic neurons after injury or disease remains to be determined.

In Drosophila, glial cells also phagocytose dead neurons and axonal fragments both in development and after injury.⁶ Hence, there is an emerging theme that glia serve an important role as amateur phagocytes during development. We can only speculate as to why these amateur phagocytes and not macrophages are responsible for clearing the neuronal debris. Perhaps glial cells clear the apoptotic neurons instead of attracting macrophages to the ganglia as a way to prevent a damaging immune response and maintain its relative immunological privilege.

Although many receptors have been identified on macrophages that contribute to the engulfment process,⁷ the mechanisms by which amateur phagocytes engulf apoptotic cells is largely unknown. In C. elegans, which do not have professional phagocytes and rely entirely on neighboring cells to scavenge corpses, two pathways have been identified. Ced-1 is the receptor for one pathway, while the receptor for the other remains to be determined. In Drosophila, Draper was identified as the Ced-1 homolog and shown to mediate neuronal corpse engulfment by glial cells.⁸ Therefore, we searched for homologs of Ced-1 and Draper that were expressed by the SGC precursors and identified MEGF10 and Jedi-1/PEAR1 as putative engulfment receptors. MEGF10 had previously been proposed as a mediator of phagocytosis,⁹ but Jedi-1 represented a novel receptor. We found that overexpressing Jedi-1 or MEGF10 in cultured SGC precursors significantly enhanced their ability to engulf apoptotic neurons, while knock down of either protein blocked the phagocytosis.

Interestingly, there was no additive effect when both Jedi-1 and MEGF10 were overexpressed nor when both were knocked down, suggesting that they lie in a common or convergent pathway. Nevertheless, they exhibit some differences in their modes of action. Jedi-1 overexpressing cells have elongated processes, suggesting regulation of the cytoskeleton. In contrast, MEGF10 overexpressing cells exhibit an increase in the number of engulfed corpses per cell, hinting at a role in endocytosis or vesicular trafficking (Fig. 1). Such a dual receptor system appears to be more the rule than the exception for phagocytosis. As mentioned above, there are two engulfment pathways in C. elegans, and in Drosophila, Draper is thought to function in the degradation of the engulfed corpses while its partner-receptor, SIMU, serves more in the recognition process.¹⁰ The need for multiple receptors highlights the complex nature of engulfment. Our recent findings are a start at delineating some of the players utilized by amateur phagocytes for engulfment in the mammalian nervous system. Of course, there are many questions remaining, such as the identity of the signal produced by the dying neuron and whether defects in this clearance process lead to autoimmunity, which will serve as interesting topics for future research.

Figure 1.

Schematic of a satellite glial precursor engulfing an apoptotic neuron. The engulfment receptors Jedi-1 and MEGF10 are involved in the clearance of dead neurons by glial precursors, likely through the regulation of the cytoskeleton and vesicular trafficking, respectively.