Undertaker, a Drosophila Junctophilin, links Draper-mediated phagocytosis and calcium homeostasis

Phagocytosis is important during development and in the immune response for the removal of apoptotic cells and pathogens, yet its molecular mechanisms are poorly understood. In Caenorhabditis elegans, the CED2/5/10/12 pathway regulates actin during phagocytosis of apoptotic cells, whereas the role of the CED1/6/7 pathway in phagocytosis is unclear. This study reports that Undertaker (UTA), a Drosophila Junctophilin protein, is required for Draper (CED-1 homolog)-mediated phagocytosis. Junctophilins couple Ca²⁺ channels at the plasma membrane to those of the endoplasmic reticulum (ER), the Ryanodine receptors. This study places Draper, its adaptor drCed-6, UTA, the Ryanodine receptor Rya-r44F, the ER Ca²⁺ sensor dSTIM, and the Ca²⁺-release-activated Ca²⁺ channel dOrai in the same pathway that promotes calcium homeostasis and phagocytosis. Thus, these results implicate a Junctophilin in phagocytosis and link Draper-mediated phagocytosis to Ca²⁺ homeostasis, highlighting a previously uncharacterized role for the CED1/6/7 pathway (Cuttell, 2008).

In a deficiency screen, a mutant was characterized in which embryonic macrophages poorly engulfed apoptotic cells. In an RNAi screen using S2 cells, undertaker/retinophilin (uta) was identified as being responsible for this phenotype. uta encodes a membrane occupational and recognition nexus (MORN) repeat-containing protein with homology to mammalian Junctophilins (JPs). JPs form junctional complexes between the plasma membrane (PM) and the endoplasmic/sarcoplasmic reticulum (ER/SR) Ca²⁺ storage compartment (Takeshima 2000). These complexes allow for crosstalk between Ca²⁺ channels at the PM and the ER/SR Ca²⁺ channels, or Ryanodine receptors (RyRs), thus regulating Ca²⁺ homeostasis and functions of excitable cells. Although a role for Ca²⁺ in phagocytosis of various particles by mammalian phagocytes has been previously described, the molecular mechanisms underlying Ca²⁺ fluxes associated with these events are not known (Cuttell, 2008).

This study reports that, as for UTA, the Drosophila Ryanodine receptor, Rya-r44F, plays a role in phagocytosis of apoptotic cells in vivo. A requirement in phagocytosis was found for store-operated Ca²⁺ entry (SOCE) via dSTIM, a Ca²⁺ sensor of the ER/SR lumen, and CRACM1/dOrai, a Ca²⁺-release-activated Ca²⁺ channel (CRAC). uta and rya-r44F genetically interact with drced-6 and drpr, and uta, drced6, and drpr are required for SOCE in S2 cells. Thus, these genes act in the same pathway that plays a general role in phagocytosis; uta, dstim, dorai, drced-6, and drpr are also required for efficient phagocytosis of bacteria. These results provide a link between SOCE and phagocytosis, imply that UTA plays a similar role in macrophages to that of JPs in excitable cells, and shed light on a role for the CED1/6/7 pathway in Ca²⁺ homeostasis during phagocytosis (Cuttell, 2008).

Binding of various particles induces a rise in [Ca²⁺]i in mammalian phagocytes. In dendritic cells, [Ca²⁺]i increases upon apoptotic cell binding via integrin, and inhibition studies have suggested that both Ca²⁺ release from the ER/SR storage pool and extracellular Ca²⁺ entry into the cytosol are required for this process. Neutrophils also rely on such changes to promote particle engulfment. Yet, the molecular mechanisms underlying this rise in [Ca²⁺]i and what role it plays in phagocytes are poorly understood (Cuttell, 2008).

This study found that uta, a Drosophila gene encoding a JP-related protein, is required for phagocytosis of apoptotic cells. Genetic evidence of a role for a Ryanodine receptor, Rya-r44F is provided, and uta and rya-44F were genetically linked. It was also found that SOCE via dstim and dorai promotes efficient apoptotic cell clearance. uta and rya-44F were genetically linked to drpr and drced-6, and a role was found for uta, drpr, and drced-6 in SOCE, thus demonstrating a functional link between the DRPR/drCed-6 pathway and SOCE during phagocytosis (Cuttell, 2008).

A model is proposed whereby apoptotic cell binding via DRPR (and possibly other receptors, such as CRQ) leads to ER Ca²⁺ release via Rya-r44F. DRPR, which bears an immunoreceptor tyrosine-based activation motif (ITAM) that is phosphorylated via Src and Syk family kinase-mediated signaling, appears to behave like an immunoreceptor. In B and T lymphocytes, engagement of Fc immunoreceptors (the signaling of which also relies on phosphorylation on ITAMs) leads to a rise in [Ca²⁺]i. Thus, DRPR might play a similar role to that of Fc receptors in the signaling, leading to a rise in [Ca²⁺]i in macrophages (Cuttell, 2008).

UTA is localized in the ER and at the PM. Thus, it is proposed that, like JPs, UTA forms junctional complexes that link the PM events to the ER and trigger Ca²⁺ release from ER stores. These studies, however, did not address whether the formation of UTA junctional complexes is required to trigger ER Ca²⁺ release via Rya-r44F, nor what triggers ER Ca²⁺ release. The resting membrane potential of mammalian phagocytes is depolarized upon contact with apoptotic cells. As in mammalian muscle cells, such changes in fly phagocytes might initiate ER Ca²⁺ release (Cuttell, 2008).

In S2 cells, Ca²⁺ imaging results with drpr and drced-6 RNAi and the results of drced-6 RNAi suggest that drpr and drced-6 are required for dOrai-mediated Ca²⁺ entry upon TG treatment (which bypasses the need for particle binding to the receptor). It is proposed that ER Ca²⁺ release feeds back onto DRPR and drCed-6 to activate their downstream signaling cascade. Although further studies will be required to test the validity of this proposal, several reports already support it: DRPR-mediated phagocytosis depends on Src and Syk family kinase signaling, and the activity of such kinases can be Ca²⁺ dependent in mammalian cells (Cuttell, 2008).

It is then proposed that signaling downstream of DRPR and drCed-6 promotes and/or maintains the formation of UTA junctional complexes, thereby linking ER Ca²⁺ release to SOCE. dSTIM is indeed likely to act as an ER Ca²⁺ sensor that oligomerizes and redistributes to ER-PM junctions upon ER Ca²⁺ depletion, as for its mammalian counterparts. It is proposed that UTA junctional complexes are needed to maintain a close proximity between the ER Ca²⁺ stores and the PM and to juxtapose dSTIM oligomers and dOrai, thereby promoting conformational changes and opening of dOrai. DRPR- and drCed-6-dependent signaling and/or UTA may also be required for dSTIM oligomerization. The resulting increase in [Ca²⁺]i then promotes engulfment of the particle (Cuttell, 2008).

Ca²⁺ may promote phagocytosis via several ways. It can enhance scavenger receptor (SR) activity: adhesion of mouse macrophages to a fibronectin-coated surface via integrin binding results in an increase in the number of SRs at their PM, which enhances their binding activity. This enrichment in SRs is dependent on extracellular Ca²⁺ influx, arguing in favor of a role for Ca²⁺ in SR trafficking and/or recycling. Several SRs or related receptors play a role in phagocytosis of apoptotic corpses, including the mammalian CD36 and its Drosophila homolog CRQ. Although no change in CRQ expression is seen in uta mutant macrophages, CRQ and UTA colocalize and genetically interact. One possible model is that CRQ is recruited to the phagocytic cup upon apoptotic cell binding after SOCE that depends on UTA, DRPR, and drCed-6, and that this might strengthen the binding and uptake of the corpse (Cuttell, 2008).

In C. elegans, CED-1 (DRPR homolog) is related to the endothelial scavenger receptor SREC. Its recruitment to the phagocytic cup depends on functional CED-7, and may occur by exocytosis (Yu, 2006). Components of the exocyst were implicated in phagocytosis (Stuart, 2007). Moreover, Orai1 is required for degranulation of mast cells, which occurs by exocytosis (Vig, 2008). Thus, like Orai1, dOrai may be required for exocytosis and, whereas DRPR appears to always be present at the PM, CRQ may be recruited from its intracellular vesicular pool to the phagocytic cup by exocytosis, as previously proposed for CED-1, to promote apoptotic cell uptake (Cuttell, 2008).

A rise in Ca²⁺ was observed in mammalian neutrophils upon particle binding, and Ca²⁺ participates in phagocytosis by promoting F-actin breakdown and phagosome maturation. Mycobacterium tuberculosis is able to invade human macrophages without triggering an increase in [Ca²⁺]i: in the absence of Ca²⁺ signaling, phagosomes containing M. tuberculosis fail to mature, perhaps explaining the survival of this bacterium in the cell. A role for Ca²⁺ in particle binding and phagosome maturation in macrophages, however, was once discounted. uta, dstim, dorai, drCed-6, and drpr are required to trigger SOCE. Yet, although they are poorly phagocytic, macrophages in drced-6 hypomorphs and drpr null mutants engulf bacteria into fully matured phagosomes, arguing against Ca²⁺ being involved in phagosome maturation. This maturation, however, might still occur with lower efficiency when SOCE fails, as RNAi-treated S2 cells for all genes in this pathway poorly engulfed bacteria (Cuttell, 2008).

The findings that UTA links DRPR-mediated phagocytosis and Ca²⁺ homeostasis provide the opportunity to pursue the dissection of the DRPR pathway in Drosophila. DRPR is homologous to CED-1, which belongs to the CED1/6/7 pathway where CED-7 is an ABC transporter. Interestingly, an ABC transporter can modulate Ca²⁺ channel activity in plants, further supporting a link between the CED1/6/7-like pathways and Ca²⁺ homeostasis, which appears to have been conserved throughout evolution. Furthermore, a mutation in human Orai1 was found in some patients with severe combined immune deficiency. Thus, pursuing such studies might be relevant to mammalian systems and to human health (Cuttell, 2008).

Transcriptional Regulation

How are the expression of drpr and ced-6 regulated in the specific subset of glial cells? Because the elevated expression of these genes occurred shortly after the onset of the pupal metamorphosis, a most likely candidate mechanism would be ecdysone signal-mediated regulation. Inhibition of the ecdysone receptor in γ neurons by ectopic expression of the dominant negative form of the ecdysone receptor (EcR-DN), however, does not affect the expression of drpr in glial cells (Awasaki, 2004). In situ hybridization confirmed the elevated expression of drpr and ced-6 at 6h APF, as in the control animal. These observations exclude the possibility that the expression of drpr and ced-6 in glia is induced extrinsically by γ neurons (Awasaki, 2006).

Therefore the cell-autonomous role of ecdysone signals in glia were examined using GAL4-repo-driven ectopic expression of EcR-DN in glial cells. Although development of the repo>EcR-DN pupae was arrested in the late prepupal stage, the pupae developed normally until at least 6h APF. In these pupae, neither intense expression of drpr transcripts nor the Drpr protein was detected in the glial cells at 6h APF. In addition, no glial processes infiltrated the MB lobe. Quantification of the expression level indicated that although the amount of drpr transcripts increased by approximately 30% in 6h APF pupae of wild-type and 201Y>EcR-DN compared with that of L3 wild-type larvae, there was no increase of drpr transcripts in the pupae of repo>EcR-DN (Awasaki, 2006).

Interestingly, Ced-6 expression was not affected by EcR-DN expression: Ced-6 was expressed in the same number of glial cells as those that coexpressed Drpr and Ced-6 in the control pupae at 6h APF (between one and three cells, an average of 2.1 cells per lobe). These results suggest that ecdysone-induced stimulation in glia is essential for the elevated expression of drpr, but not ced-6 (Awasaki, 2006).

Protein Interactions

The cellular machinery promoting phagocytosis of corpses of apoptotic cells is well conserved from worms to mammals. An important component is the Caenorhabditis elegans engulfment receptor CED-1 and its Drosophila orthologue, Draper. The CED-1/Draper signalling pathway is also essential for the phagocytosis of other types of 'modified self' including necrotic cells, developmentally pruned axons and dendrites, and axons undergoing Wallerian degeneration. This study shows that Drosophila Shark, a non-receptor tyrosine kinase similar to mammalian Syk and Zap-70 (Ferrante, 1995), binds Draper through an immunoreceptor tyrosine-based activation motif (ITAM) in the Draper intracellular domain. Shark activity is essential for Draper-mediated signalling events in vivo, including the recruitment of glial membranes to severed axons and the phagocytosis of axonal debris and neuronal cell corpses by glia. The Src family kinase (SFK) Src42A can markedly increase Draper phosphorylation and is essential for glial phagocytic activity. It is proposed that ligand-dependent Draper receptor activation initiates the Src42A-dependent tyrosine phosphorylation of Draper, the association of Shark and the activation of the Draper pathway. These Draper-Src42A-Shark interactions are strikingly similar to mammalian immunoreceptor-SFK-Syk signalling events in mammalian myeloid and lymphoid cells. Thus, Draper seems to be an ancient immunoreceptor with an extracellular domain tuned to modified self, and an intracellular domain promoting phagocytosis through an ITAM-domain-SFK-Syk-mediated signalling cascade (Ziegenfuss, 2008).

Developing tissues produce excessive numbers of cells and selectively destroy a subpopulation through programmed cell death to regulate growth. Rapid clearance of cell corpses is essential for maintaining tissue homeostasis and preventing the release of potentially cytotoxic or antigenic molecules from dying cells, and defects in cell corpse clearance are closely associated with autoimmune and inflammatory diseases. In C. elegans the CED-1 receptor is expressed in engulfing cells, where it acts to recognize cell corpses and drive their phagocytosis. CED-1 promotes engulfment through an intracellular NPXY motif, a binding site for proteins containing a phosphotyrosine-binding (PTB) domain, and a YXXL motif, a potential interaction site for proteins containing SH2 domains. The PTB domain adaptor protein CED-6 can bind the NPXY motif of CED-1, is required for cell corpse engulfment and acts in the same genetic pathway as CED-1. CED-1 ultimately mediates actin-dependent cytoskeletal reorganization through the Rac1 GTPase, and Dynamin modulates vesicle dynamics downstream of CED-1 during engulfment, but the molecular signalling cascade that allows CED-1 to execute phagocytic events remains poorly defined (Ziegenfuss, 2008).

Glia are the primary phagocytic cell type in the developing and mature brain. Glia rapidly engulf neuronal cell corpses produced during development, as well as neuronal debris generated during axon pruning or during Wallerian degeneration in the adult brain. In Drosophila, glial phagocytosis of these engulfment targets requires Draper, the fly orthologue of CED-1. Draper, like CED-1, contains 15 extracellular atypical epidermal growth factor (EGF) repeats, a single transmembrane domain, and NPXY and YXXL motifs in its intracellular domain. Drosophila Ced-6 is also required for the clearance of pruned axons, indicating possible conservation of the interaction between CED-1 and CED-6 in flies, but additional signalling molecules acting downstream of Draper have not been identified (Ziegenfuss, 2008).

Draper was identified in a yeast two-hybrid screen for molecules interacting with the regulatory region of Shark. When LexA-Shark, constitutively active Src kinase and AD-Draper are present, Shark and Draper interact physically. In the absence of Src kinase, Shark and Draper fail to interact, indicating that phosphorylation of Draper by Src may be essential for Shark-Draper interactions. The Draper intracellular domain contains an ITAM (YXXI/L-X_6-12-YXXL), a key domain found in many mammalian immunoreceptors including Fc, T-cell and B-cell receptors. SFKs phosphorylate the tyrosines in ITAM domains, thereby allowing ITAM association with SH2-domain-containing signal transduction proteins including Syk and Zap-70. Y-->F substitutions of the tyrosine residues were generated within or near the Draper ITAM, and it was found that Tyr 949 and Tyr 934 are critical for robust Draper-Shark binding. These correspond to the consensus tyrosine residues in the predicted Draper ITAM. Plasmids were transfected with carboxy-terminally haemagglutinin-tagged Draper (Draper-HA) or with Draper-HA and Shark with an amino-terminal Myc tag (Myc-Shark) into Drosophila S2 cells, immunoprecipitated with anti-HA antibodies, and western blots were performed with anti-phosphotyrosine, anti-Myc and anti-HA antibodies. Myc-Shark was found to co-immunoprecipitate with Draper-HA, and that anti-phosphotyrosine antibodies label a band corresponding to the position of Draper-HA that is absent in empty vector controls. Further, it was found that a Y949F substitution markedly reduces Draper-Shark association. Taken together, these data indicate Draper and Shark can associate physically through the Draper ITAM domain (Ziegenfuss, 2008).

Attempts were made to determine whether Shark is required for glial phagocytic activity in vivo. Severing adult Drosophila olfactory receptor neurons (ORNs) initiates Wallerian degeneration of ORN axons. Antennal lobe glia surrounding these severed axons respond to this injury by extending membranes towards severed axons and engulfing degenerating axonal debris. These glia express high levels of Draper, and in draper^δ5 null mutants, glia fail to respond morphologically to axon injury, and severed axons are not cleared from the central nervous system (CNS). Thus, both the extension of glial membranes to severed axons and the phagocytosis of degenerating axonal debris require Draper signalling (Ziegenfuss, 2008).

Whether Shark function in glia is essential for glial responses to axon injury was explored by driving a UAS-regulated double-stranded RNAi construct designed to target shark (shark^RNAi) with the glial-specific repo-Gal4 driver, severing ORN axons, and assaying the recruitment of Draper and green fluorescent protein (GFP)-labelled glial membranes to severed axons. Maxillary palp-derived ORN axons project to 6 of the roughly 50 glomeruli in the antennal lobe. Within hours after maxillary palps have been ablated in control animals, Draper immunoreactivity decorates severed axons projecting to and within maxillary palp ORN-innervated glomeruli, and GFP-labelled glial membranes are recruited to these severed axons. Strikingly, knocking down Shark in glia completely suppresses these events. Next, antennal ORN axons were severed; these axons project to about 44 of the 50 antennal lobe glomeruli. Antennal ablation therefore injures nearly all glomeruli in the antennal lobe and results in the majority of antennal lobe glia in control animals upregulating Draper and undergoing hypertrophy. It was found that knocking down Shark in glia also blocks this glial response to axon injury. Thus, Shark is essential for all axon-injury-induced changes in glial morphology and Draper expression (Ziegenfuss, 2008).

To determine whether Shark is required for glial phagocytosis of severed axons, a subset of maxillary palp ORN axons were labelled with mCD8::GFP, Shark function was knocked down in glia, and the clearance of severed axons was assayed. In control animals severed GFP-labelled ORN axonal debris was cleared from the CNS within 5 days. In contrast, glial-specific shark^RNAi potently suppresses the clearance of degenerating axons, with severed axons lingering in the CNS for at least 5 days. Then whether mutations in the shark gene affected the glial clearance of degenerating axons was examined. The null allele of shark, shark¹, is pupal lethal (Fernandez, 2000). Therefore glial responses to axon injury were assayed in shark¹ heterozygous mutants, and dominant genetic interactions between draper^δ5 and shark¹ were tested. It was found that both draper^δ5/+ and shark¹/+ animals showed defects in glial phagocytic function: 5 days after injury, significant amounts of axonal debris remained within OR85e-innervated glomeruli. Moreover, shark¹/+; draper^δ5/+ animals showed a striking suppression of glial clearance of severed axons almost equivalent to that of draper^δ5 mutants. Thus, shark mutations dominantly suppress the glial clearance of degenerating ORN axons, and this phenotype is strongly enhanced by removing one copy of draper. These data, taken together with shark^RNAi data, show that Shark is essential for the clearance of degenerating axons by glia (Ziegenfuss, 2008).

Is Shark required for the glial clearance of neuronal cell corpses? In embryonic stage 14-15 control animals, 24.4 cell corpses were found per hemisegment. In contrast, it was found that shark¹ null mutants showed a marked increase in CNS cell corpses, with null mutants containing almost twice as many corpses per hemisegment. shark¹/Df(2R)6063 mutants accumulate cell corpses at levels similar to those in shark¹, indicating that this phenotype maps to shark. These cell corpse engulfment phenotypes are indistinguishable from that of draper^δ5 mutants. It is concluded that Shark, like Draper, is also essential for the efficient clearance of embryonic neuronal cell corpses by glia (Ziegenfuss, 2008).

Because it was found that Shark binds Draper only in the presence of an active Src kinase in two-hybrid assays, Drosophila Src kinases were screened for roles in glial phagocytic activity. Interestingly, it was found that glia-specific knockdown of Src42A (src42A^RNAi) potently suppress glial phagocytic activity: in src42A^RNAi animals, Draper is not recruited to severed maxillary palp axons; glial hypertrophy and upregulation of Draper after antennal ablation was blocked; and GFP-labelled severed axons lingered in the CNS for 5 days. Knockdown of two other Drosophila Src kinases, Btk29A and Src64B, had no effect on the glial phagocytosis of severed axons. Thus, Src42A seems to be essential for all morphological responses of glia to axon injury and for the efficient clearance of degenerating axonal debris from the CNS (Ziegenfuss, 2008).

It was predicted that Draper phosphorylation status should be sensitive to the SFK inhibitor PP2. Indeed, addition of PP2 to S2 cultures led to a decrease in the phosphorylation of Draper and Draper-Shark association. Strikingly, co-transfection of Draper and Src42A led to a marked increase in Draper phosphorylation, which was PP2-sensitive and Draper-specific. Draper-Shark interactions are not dependent on Shark kinase activity because kinase-dead Shark (Shark K698R) associates with Draper. These data indicate that Src42A may phosphorylate the Draper intracellular domain, thereby increasing the association of Shark with Draper and the activation of downstream glial phagocytic signalling (Ziegenfuss, 2008).

This study has identified Shark and Src42A as novel components of the Draper pathway. One potential model for Draper-Shark-Src42A interactions is that Shark and Src42A drive the recruitment of Draper to engulfment targets. However, CED-1 has been shown to cluster around cell corpses even in the absence of its intracellular domain. Moreover, Zap-70 and Syk bind phosphorylated ITAM domains in mammalian immunoreceptors when ITAM domains are phosphorylated by Src after ligand-dependent receptor. Therefore a model is favoured in which the engagement of Draper with its ligand (presumably presented by engulfment targets) promotes receptor clustering, tyrosine phosphorylation of Draper by Src42A, association of Shark, and activation of downstream phagocytic signalling events. This work suggests that Draper is an ancient immunoreceptor in which the extracellular domain is tuned to recognize modified self and the intracellular domain signals through ITAM-Src-Syk-mediated mechanisms. This is the first identification of ITAM-Src-Syk signalling in invertebrates, and it suggests that a pathway similar to Draper-Ced-1 may ultimately have given rise to ITAM-based signalling cascades in mammalian myeloid and lymphoid cells, including those regulated by Fc, B-cell and T-cell receptors (Ziegenfuss, 2008).

Pretaporter, a Drosophila protein serving as a ligand for Draper in the phagocytosis of apoptotic cells

Phagocytic removal of cells undergoing apoptosis is necessary for animal development and tissue homeostasis. Draper, a homologue of the C. elegans phagocytosis receptor CED-1, is responsible for the phagocytosis of apoptotic cells in Drosophila, but its ligand presumably present on apoptotic cells remains unknown. An endoplasmic reticulum protein that binds to the extracellular region of Draper was isolated. Loss of this protein, which has been named Pretaporter, led to a reduced level of apoptotic cell clearance in embryos, and the overexpression of Pretaporter in the mutant flies rescued this defect. Results from genetic analyses suggested that Pretaporter functionally interacts with Draper and the corresponding signal mediators. Pretaporter was exposed at the cell surface after the induction of apoptosis, and cells artificially expressing Pretaporter at their surface became susceptible to Draper-mediated phagocytosis. Finally, the incubation with Pretaporter augmented the tyrosine-phosphorylation of Draper in phagocytic cells. These results collectively suggest that Pretaporter relocates from the endoplasmic reticulum to the cell surface during apoptosis to serve as a ligand for Draper in the phagocytosis of apoptotic cells (Kuraishi, 2009).

Throughout the life of multi-cellular organisms, cells of particular types die and are eventually eliminated in certain places in the body and at certain developmental stages under physiological and sub-physiological conditions. Such 'unwanted cells' are mostly induced to undergo apoptosis and subjected to phagocytic elimination. Prompt and selective phagocytosis of apoptotic cells is prerequisite for morphogenesis, establishment of tissue functions, tissue renewal, avoidance of diseases, and effective progress of tissue functions (Kuraishi, 2009 and references therein).

There are two genetically identified signalling pathways that lead to the induction of phagocytosis of dying cells in C. elegans: one is made up of the proteins CED-2, CED-5, and CED-12, and the other consists of CED-1, CED-6, and CED-7. The pathways converge at CED-10, a small G-protein responsible for the rearrangement of cytoskeleton in phagocytes (Kinchen, 2005). The fact that all these C. elegans proteins possess counterparts in Drosophila and mammals (Lettre, 2006; Kinchen, 2007) suggests that these two partially redundant signalling pathways are evolutionally conserved. However, the mode of action of those proteins still remains to be clarified, and there are missing components in the pathways. The selectivity in the recognition of apoptotic cells by phagocytes is due to the specific interaction between receptors of phagocytes and their ligands present at the surface of target cells. Presumably, there are two phagocytosis receptors in C. elegans, and CED-1 is likely the one located upstream of CED-6 and CED-7 (Zhou, 2001; Yu, 2006; Venegas, 2007) whereas the other has not been found in genetic studies. Draper (Drpr), a Drosophila homologue of CED-1, has been shown to act as a receptor in the phagocytosis of apoptotic cells and more recently molecules that act with Drpr to accomplish the engulfment and subsequent processing of apoptotic cells in Drosophila phagocytes were reported (Cuttell, 2008; Kurant, 2008; Ziegenfuss, 2008). However, a molecule(s) present at the surface of apoptotic cells and recognized by Drpr is yet to be identified. The study by Venegas (2007) has implied that the membrane phospholipid phosphatidylserine, the best-characterized phagocytosis marker in mammals, could be a ligand for CED-1. In contrast, this study showed that phosphatidylserine is not required in the Drpr-mediated phagocytosis of apoptotic cells. In this study, a phagocytosis marker(s) recognized by Drpr was sought, and an endoplasmic reticulum (ER) protein was identified as a strong candidate (Kuraishi, 2009).

This study reports the identification of a Drosophila protein, Prtp, serving as a ligand for Drpr in the phagocytosis of apoptotic cells. The level of phagocytosis in embryos of the null mutants for prtp and drpr was almost the same and about half of that in controls. This suggests that Prtp is solely responsible for the Drpr-mediated phagocytosis by embryonic haemocytes, and that there are additional ligands and receptors for phagocytosis. The following pathway is proposed for the induction of the Prtp-dependent, Drpr-mediated phagocytosis of apoptotic cells: Prtp relocates from the ER to the cell surface during apoptosis; Prtp binds to Drpr of haemocytes and induces tyrosine phosphorylation; Ced-6 binds to phosphorylated Drpr and further transmits the signal leading to the activation of Rac1 and/or Rac2; and cytoskeletons are reorganized for engulfment. Additional molecules could participate in this pathway: a membrane protein named Six-microns-under reportedly acts upstream of Drpr in the phagocytosis of apoptotic neurons by glia (Kurant, 2008}; another signal mediator, named Shark, located proximal downstream of Drpr has been reported (Ziegenfuss, 2008); and Undertaker, an intracellular membrane protein involved in the regulation of calcium homeostasis, was shown to function downstream of Drpr (Cuttell, 2008}. In particular, Six-microns-under has been suggested to bridge glia and apoptotic neurons (Kurant, 2008}. The results suggested that Prtp still serves as a ligand for Drpr in the phagocytosis by glia that express Six-microns-under. Further investigation is necessary before a complete picture is obtained of the Prtp/Drpr-initiated signalling pathway for the phagocytosis of apoptotic cells. More molecules are involved in the phagocytosis of apoptotic cells in Drosophila: the receptor Croquemort, the phagocytosis marker Calreticulin, and Pallbearer, a component of ubiquitin ligase, which might be integrated into the other engulfment pathway (Kuraishi, 2009).

Prtp normally resides in the ER, most likely in its lumen, and a portion of it seems to relocate to the cell surface during apoptosis. The ER and plasma membranes exchange their components, and this might be enhanced upon the induction of apoptosis. In fact, a variety of proteins and lipids of the ER are exposed at the surface of apoptotic human cells (Franz, 2007). More recently, it was reported that the exposure of ER proteins at the surface of apoptotic mammalian cells occurs by SNARE-dependent exocytosis (Panaretakis, 2009). It is thus probable that together with other ER components Prtp moves to the cell surface during apoptosis and serves as a ligand for the phagocytosis receptor Drpr. Prtp seemed dispensable for the removal of degenerated neural axons during metamorphosis, which requires the action of Drpr. It is speculated that Prtp is not exposed on the surface of degenerated axons because the removal of these axons occurs independently of caspases, and that another ligand for Drpr exists at the surface of degenerated axons. In contrast, Drpr appears to serve as a receptor in the phagocytosis of bacteria (Cuttell, 2008; Hashimoto, 2009). Taken together, it is likely that Drpr is a multi-ligand receptor for phagocytosis responsible for the maintenance of tissue homeostasis through the removal of degenerated own cells and invading microbial pathogens (Kuraishi, 2009).

A counterpart for Prtp in C. elegans could be a ligand for CED-1, but there seems no complete homologue of prtp in its genome. The current analysis has suggested that the presence of two thioredoxin-like domains is sufficient for the binding of Prtp to Drpr. Therefore, a C. elegans protein containing such a structure, PDI for example, is a candidate for the CED-1 ligand. In contrast, the mammalian homologue of Prtp seems to exist; a mouse protein called ERp46 and human proteins belonging to the TXNDC5 family possess a domain composition similar to that of Prtp. It is important to examine if C. elegans PDI and mammalian ERp46 and TXNDC5 act as ligands for CED-1 and its mammalian homologue MEGF10, respectively, in the phagocytosis of apoptotic cells (Kuraishi, 2009).

Identification of lipoteichoic acid as a ligand for draper in the phagocytosis of Staphylococcus aureus by Drosophila hemocytes

Phagocytosis is central to cellular immunity against bacterial infections. As in mammals, both opsonin-dependent and -independent mechanisms of phagocytosis seemingly exist in Drosophila. Although candidate Drosophila receptors for phagocytosis have been reported, how they recognize bacteria, either directly or indirectly, remains to be elucidated. Staphylococcus aureus genes required for phagocytosis by Drosophila hemocytes were sought in a screening of mutant strains with defects in the structure of the cell wall. The genes identified included ltaS, which encodes an enzyme responsible for the synthesis of lipoteichoic acid. ltaS-dependent phagocytosis of S. aureus required the receptor Draper but not Eater or Nimrod C1, and Draper-lacking flies showed reduced resistance to a septic infection of S. aureus without a change in a humoral immune response. Finally, lipoteichoic acid bound to the extracellular region of Draper. It is proposed that lipoteichoic acid serves as a ligand for Draper in the phagocytosis of S. aureus by Drosophila hemocytes and that the phagocytic elimination of invading bacteria is required for flies to survive the infection (Hashimoto, 2009).

Larval

The lobes of the larval MB are formed by a bundle of axon branches of the MB γ neurons. The distal tip of the MB dorsal lobe is surrounded by a few glial cells, which extend glial processes that cover the tip and shaft region of the lobe. Shortly after the onset of pupal metamorphosis, these glial processes infiltrate the lobes to engulf the axon branches (Awasaki, 2006).

Infiltrating glial processes are labeled with anti-Drpr antibody (Awasaki, 2004). To identify the glial cells that extend these Drpr-positive processes, the expression of drpr was examined by in situ hybridization, focusing on the area around the MB dorsal lobe. The drpr transcript was observed in a small number of cells adjacent to the dorsal lobe. Whereas the expression was weak in wandering larvae (L3) and pupae at 12h APF, it was elevated in pupae at 6h APF. The cells that express drpr were also labeled with anti-Repo antibody and expressed the nuclear-localizing signal LacZ fusion reporter (nLacZ), which was driven by GAL4-repo (repo>nLacZ), indicating that they were glia (Awasaki, 2006).

In C. elegans, the nematode ortholog of drpr, ced-1, interacts genetically with ced-6 in the engulfment of apoptotic cells, and CED-1 and CED-6 are colocalized in the cells around early apoptotic cells (Ellis, 1991; Liu, 1998; Kinchen, 2005). These proteins also interact biochemically (Su, 2002). Therefore, the expression of Drosophila ced-6 was examined in larval and pupal brains. In situ hybridization revealed weak expression of ced-6 in the cells adjacent to the MB dorsal lobe at L3 and 12h APF and elevated expression at 6h APF. The ced-6-expressing cells were glial cells, because they were labeled with anti-Repo antibody and repo>nLacZ (Awasaki, 2006).

Although both drpr and ced-6 transcripts are expressed in the glial cells adjacent to the MB dorsal lobes with a similar pattern, only a specific subset of the glial cells in this area expressed these genes. To determine whether they are expressed in the same glial cells, the distribution of the Drpr and Ced-6 proteins were examined using anti-Drpr and anti-Ced-6 antibodies. Drpr and Ced-6 were detected in the same glial cells around the MB lobe. Between one and three glial cells in this area coexpressed Drpr and Ced-6 at 6h APF (an average of 2.2 glial cells per lobe). Other glial cells expressed neither Drpr nor Ced-6. Drpr and Ced-6 were, however, detected in all of the infiltrating glial processes. This finding strongly suggests that only a specific subset of the glial cells in which these two proteins are expressed extend their processes into the MB lobes. In addition to the MB dorsal lobe, glial processes infiltrating the medial lobe were labeled with both anti-Drpr and anti-Ced-6 antibodies. These processes derived from the glial cells that were arranged in the posterior region of the medial lobe (Awasaki, 2006).

These findings indicate that elevated expression of both drpr and ced-6 was transiently induced in the specific subset of glial cells that send their processes into the MB lobes early in metamorphosis. Although more than eight glial cells are arranged around the MB dorsal lobe at 6h APF, only a subset of them contribute to infiltration by glial processes (Awasaki, 2006).

Adult

Ensheathing glia function as phagocytes in the adult Drosophila brain

The mammalian brain contains many subtypes of glia that vary in their morphologies, gene expression profiles, and functional roles; however, the functional diversity of glia in the adult Drosophila brain remains poorly defined. This study defines the diversity of glial subtypes that exist in the adult Drosophila brain, show they bear striking similarity to mammalian brain glia, and identify the major phagocytic cell type responsible for engulfing degenerating axons after acute axotomy. Beuropil regions were found to contain two different populations of glia: ensheathing glia and astrocytes. Ensheathing glia enwrap major structures in the adult brain, but are not closely associated with synapses. Interestingly, it was found that these glia uniquely express key components of the glial phagocytic machinery (e.g., the engulfment receptor Draper, and dCed-6), respond morphologically to axon injury, and autonomously require components of the Draper signaling pathway for successful clearance of degenerating axons from the injured brain. Astrocytic glia, in contrast, do not express Draper or dCed-6, fail to respond morphologically to axon injury, and appear to play no role in clearance of degenerating axons from the brain. However, astrocytic glia are closely associated with synaptic regions in neuropil, and express excitatory amino acid transporters, which are presumably required for the clearance of excess neurotransmitters at the synaptic cleft. Together these results argue that ensheathing glia and astrocytes are preprogrammed cell types in the adult Drosophila brain, with ensheathing glia acting as phagocytes after axotomy, and astrocytes potentially modulating synapse formation and signaling (Doherty, 2009).

In an effort to identify distinct morphological subtypes of glial cells in the mature Drosophila brain, the MARCM system was used to generate small clones of glial cells labeled with membrane-tethered GFP. Larvae containing a hs-flipase allele and a wild-type chromosome arm for recombination were subjected to a short heat shock (37°C) and glial cells within clones were visualized by use of the pan-glial driver, repo-Gal4. The analysis of glial subtype morphology was focused mainly on the adult antennal lobe region owing to its well defined histology and accessibility to genetic manipulation. In the adult brain clones were identified resembling each of the three main types of glial cells found in embryos and larvae: (1) cortex glia, which resided outside the neuropil in regions housing neuronal cell bodies, ramified dramatically to surround individual cell bodies; (2) surface glia, which appeared as large flat cells enveloping the surface of the brain, did not extend any processes into the brain; and (3) neuropil glia, which were closely associated with the neuropil, extended membranes into synaptic regions, and surrounded large bundles of axons. As in the embryo and larva, glial cell bodies were not found within the neuropil, rather they resided at the edge of the neuropil (neuropil glia), in the cortex (cortex glia), or at the surface of the brain (surface glia) (Doherty, 2009).

Interestingly, the single-cell resolution provided by MARCM analysis allowed further subdivision of neuropil glia into two distinct morphological classes, 'ensheathing glia' and 'astrocytic glia.' Ensheathing glia appeared as flattened cells that lined the borders of the neuropil and subdivided regions of the brain by isolating neuropil from the surrounding cortex. Within the antennal lobe, ensheathing glial membranes surrounded individual glomeruli (the functional units of the antennal lobe) but did not extend into the synaptic regions of the glomeruli. In addition, an astrocyte-like cell type was identified that extended membrane processes deeply into the neuropil and ramified profusely in synaptic-rich regions. This latter cell type is referred to as the fly 'astrocyte', based on its striking morphological similarity to mammalian astrocytes, as well as the conserved expression of a number of molecular markers used to identify astrocytes in the mammalian brain. Mammalian astrocytes remove excess amounts of extracellular glutamate through the high-affinity excitatory amino acid transporters (EAATs), GLAST and GLT-1, which transports the glutamate into glial cells where it is then converted into glutamine by glutamine synthetase. It was found that Drosophila astrocytes also express the transporter EAAT1. While the adult antennal lobe was used as the primary model tissue in this study, the morphological glial subtypes described above were observed in all brain regions examined, suggesting that the results are generally applicable to glial populations throughout the adult Drosophila brain (Doherty, 2009).

Attempts were made to identify Gal4 driver lines that would allow unique labelling and and manipulation of these glial populations, with a major focus being to genetically subdivide neuropil glia (i.e., ensheathing glia versus astrocytes). To accomplish this, UAS-mCD8::GFP was crossed to a previously described collection of embryonic and larval glial drivers (Ito, 1995), as well as a number of drivers generated for this study. The adult antennal lobe was studied to examine the morphology and spatial distribution of cell types marked by these drivers in a background with glial nuclei ({alpha}-Repo) and the neuropil ({alpha}-nc82) also labeled. The repo-Gal4 driver labeled all Repo+ glial subtypes in the adult brain, as evidence by {alpha}-Repo immunostaining in the nuclei of GFP+ cells. Membrane processes from Repo+ cells are found throughout the adult brain, and together they constitute the diverse collection of glial subtypes identified in single-cell MARCM analysis. Upon examination of a single glomerulus within the antennal lobe, membranes from Repo+ cells were found to both surround and invade glomeruli. All GFP expression in the adult brain driven by the repo-Gal4 driver can be suppressed by coexpression of Gal80 (a Gal4 inhibitor) under control of the repo promoter (repo-Gal80), arguing that repo-Gal80 can efficiently block Gal4-mediated activation of UAS-reporters in all adult brain glia (Doherty, 2009).

Two drivers, mz0709-Gal4 and alrm-Gal4, appeared to show very specific expression in ensheathing glia and astrocytes, respectively. Glial processes labeled by mz0709-Gal4 were found at the edge of the antennal lobe and extended deeply into the neuropil region. These flattened glial processes surrounded, but did not invade, individual glomeruli, and did not extend into the cortex region. With the exception of variable expression in a small number of neurons, all mz0709-Gal4-induced expression was suppressed by repo-Gal80, indicating that mz0709-Gal4 is largely specific to ensheathing glia. The generation of MARCM clones labeled with the mz0709-Gal4 driver resulted in the consistent labeling of ensheathing glia, but not astrocytes, within the antennal lobes. Reciprocally, alrm-Gal4 was found to be expressed exclusively in astrocytes. All cellular processes from cells labeled with alrm-Gal4 extended into the neuropil, showed a highly branched or tufted morphology, invaded individual glomeruli, and all alrm-Gal4-driven expression was suppressed by repo-Gal80. Additionally, single cell MARCM clones were found labeled with the alrm-Gal4 driver resulted in the consistent labeling of astrocytes, but not ensheathing glia. Together, these drivers are excellent tools to manipulate and functionally distinguish different subtypes of glia in the adult Drosophila brain (Doherty, 2009).

What are the functional roles for each glial subtype in the adult brain? Is each subtype responsible for a unique collection of tasks, or are all glial subtypes functionally equivalent? As a first step to determining the in vivo functional differences between adult brain glial subtypes the cell autonomy of glial phagocytic function was explored. Severing olfactory receptor neuron (ORN) axons by surgical ablation of maxillary palps leads to axon degeneration (termed Wallerian degeneration), recruitment of glial membranes to fragmenting axons, and glial engulfment of axonal debris. These glial responses are mediated by Draper, the Drosophila ortholog of the C. elegans cell corpse engulfment receptor cell death defective-1 (CED-1). In draper null mutants, glia fail to extend membranes to degenerating ORN axons and axonal debris is not removed from the CNS. Thus, Draper function should be autonomously required in phagocytic glial subtypes and Draper expression is predicted to act as a molecular marker for glial cells capable of performing engulfment functions (Doherty, 2009).

To define the precise cell types that express Draper, all glial membranes were labeled with mCD8::GFP driven by repo-Gal4, stained with α-Draper antibodies, and assayed for colocalization of Draper and GFP. Extensive overlap of Draper and GFP was found in this background. Draper and GFP signals overlapped at the edge of the neuropil, in membranes surrounding antennal lobe glomeruli, and in all cortex glia. This labeling was specific to Draper since expression of a UAS-draper^RNAi construct with repo-Gal4 led to the elimination of all Draper immunoreactivity in the adult brain. Thus, the entire population of cortex glia appear to express Draper and are likely to be phagocytic. However, cortex glia do not extend membranes into the antennal lobe neuropil, even after ORN axon injury. Therefore, cortex glia are not likely responsible for clearing severed ORN axonal debris from the antennal lobe neuropil (Doherty, 2009).

Interestingly, when mCD8::GFP was driven by mz0709-Gal4 extensive overlap of Draper and GFP was observed in neuropil-associated ensheathing glia. A high-magnification view of the antennal lobe revealed Draper and mz0709-Gal4 labeled membranes colocalizing and surrounding, but not innervating individual glomeruli. Moreover, expression of UAS-draper^RNAi in ensheathing glia with mz0709-Gal4 led to a dramatic reduction in Draper immunoreactivity in the neuropil, but the weaker Draper immunoreactivity in the cortex remained unchanged. Conversely, no overlap was observed between Draper and GFP when astrocytic membranes were labeled using the alrm-Gal4 driver. Furthermore, driving the expression of UAS-draper^RNAi in astrocytes had no obvious effect on Draper expression in the brain. These results indicate that Draper is expressed in cortex glia and ensheathing glia but not in astrocytes (Doherty, 2009).

The specific expression of Draper in antennal lobe ensheathing glia suggests that this glial subset is the phagocytic cell type responsible for engulfing degenerating axonal debris after ORN axotomy. To explore this possibility, it was asked whether ensheathing glia or astrocytes extend membranes to severed axons after injury, and in which cell type Draper was required for clearance of axonal debris from the CNS. To assay extension of glial membranes to severed axons, glial membranes were labeled with mCD8::GFP, maxillary palp axons were severed, and colocalization of Draper and GFP was assayed in glomeruli housing severed ORN axons. Within 1 d after injury, Repo⁺ glial membranes were found to localize to glomeruli housing severed maxillary palp axons and these membranes were decorated with Draper immunoreactivity. Similarly, mz0709⁺ glial membranes also localized to severed axons and colocalized with intense Draper immunoreactivity 1 d after injury. Knockdown of Draper with UAS-draper^RNAi using repo-Gal4 or mz0709-Gal4 completely suppressed the recruitment of both Draper and glial membranes to severed axons. In contrast, when astrocyte membranes were labeled with GFP no colocalization of GFP and Draper immunoreactivity was observed 1 d after axotomy. In addition, knockdown of Draper in astrocytes with UAS-draper^RNAi did not suppress the recruitment of Draper to severed axons. In an effort to identify any indirect role for astrocytes during the injury response, the morphology of astrocytes was examined both before and after injury to determine whether they exhibited any overt changes in morphology or retracted their membranes from the site of injury to accommodate the recruitment of ensheathing glial membranes. However, no obvious changes were detected in morphology or in the positions of the astrocyte glial cells in response to axon injury. Together, these data indicate that Draper is required in ensheathing glia for recruitment of glial membranes and accumulation of Draper on severed ORN axons, and suggest that Drosophila astrocytic glia do not undergo any dramatic changes in morphology in response to ORN axotomy (Doherty, 2009).

From the above data it is predicted that ensheathing glia would act as phagocytes to engulf degenerating ORN axonal debris from the CNS. To test this a subset of maxillary palp ORN axons was labeled with mCD8::GFP using the OR85e-mCD8::GFP transgene, Draper function was knocked down in glial subsets using subset-specific driver lines, maxillary palp ORN axons were severed, and clearance of axons was assayed 5 d after injury. GFP-labeled axons were first severed in control animals with each driver and GFP⁺ axonal debris was found to be efficiently cleared from the CNS within 5 d after injury, confirming that glial phagocytic function is not affected in the driver lines. Strikingly, RNA interference (RNAi) knockdown of Draper using UAS-draper^RNAi in a background with repo-Gal4 or mz0709-Gal4 completely blocked clearance of GFP-labeled axonal debris from the CNS, while RNAi knockdown of Draper in astrocytes with alrm-Gal4 had no effect on axon clearance. Thus, Draper is required autonomously in ensheathing glia for the clearance of degenerating ORN axonal debris from the CNS. In addition, knockdown of Draper in ensheathing glia with mz0709-Gal4 had no measurable effect on Draper expression in cortex glia, arguing that cortex glia are not capable of compensating for the loss of phagocytic activity in ensheathing glia during the clearance of axonal debris from the antennal lobe neuropil after axotomy. From these data on morphogenic responses to injury and phagocytic function, it is concluded that astrocytic, cortex, and ensheathing glia represent functionally distinct subsets of glial cells in the adult Drosophila brain (Doherty, 2009).

The Draper signaling pathway is a central mediator driving glial engulfment of neuronal cell corpses, and axons undergoing Wallerian degeneration in Drosophila. These studies show that within the adult brain neuropil, ensheathing glial cells are the only cell type that expresses Draper. Draper expression overlaps precisely with ensheathing glial membranes (when labeled with GFP driven by the ensheathing glia-specific driver mz0709-Gal4), and RNAi for draper in ensheathing glia leads to the elimination of Draper immunoreactivity in the neuropil. Similarly, it was found that dCed-6, the fly ortholog of C. elegans CED-6, a PTB-domain binding protein that functions genetically downstream of worm CED-1, is expressed in the adult brain in a pattern indistinguishable from Draper. dCed-6 immunoreactivity is strongly reduced in the neuropil through mz0709-Gal4-mediated knockdown (with UAS-dCed-6^RNAi), and eliminated from the entire brain when dCed-6^RNAi treatment is performed with the pan-glial driver repo-Gal4. Thus, in the adult brain dCed-6 appears to be expressed exclusively in glia, including cortex glia and ensheathing glia of the neuropil (Doherty, 2009).

Components of the Draper signaling pathway are required specifically in ensheathing glia for glial membrane recruitment to severed ORN axons, and clearance of degenerating axonal debris from the brain. Knocking down either draper or shark in ensheathing glia is sufficient to block recruitment of glial membranes and Draper to severed axons and clearance of degenerating axonal debris from the CNS. It is suspected that dCed-6 is also required in ensheathing glia for a number of reasons. First, dCed-6 and Draper show perfect overlap in expression in the neuropil, and Draper is only expressed in ensheathing glia. Second, ensheathing glia are recruited to degenerating axon injury, and dCed-6, like Draper, is specifically recruited to degenerating maxillary palp ORN axons after maxillary palp ablation. Third, dCed-6 expression is dramatically increased in ensheathing glia surrounding the antennal lobe after antennal ablation, similar to what has been found for Draper. Surprisingly, RNAi knock down of dced-6 in ensheathing glia (with mz0709-Gal4) failed to suppress the recruitment of Draper to severed ORN axons or the clearance of axonal debris from the brain. It is suspected that this is because mz0709-Gal4-mediated knockdown of dced-6 is incomplete in ensheathing glia, based on reduced but not eliminated staining in this background. Nevertheless, the observations that dCed-6 is localized with Draper in immunostains, is required in glia by RNAi knockdown with repo-Gal4, and that a null allele of dced-6 genetically interacts with draper null mutations in axon engulfment, argues strongly for a role for dCed-6 in ensheathing glia (Doherty, 2009).

Moreover, knockdown of draper in ensheathing glia, or suppressing glial engulfing activity by blocking endocytic function with Shibire^ts fully suppresses the recruitment of Draper to severed axons and clearance of degenerating axonal debris from the brain. In contrast, astrocytes do not express Draper (or dCed-6), fail to respond morphologically to axon injury, and knockdown of draper in astrocytes has no effect on the recruitment of Draper to severed axons or clearance of axonal debris from the CNS. Such a separation of phagocytic function in neuropil glial cells suggests similarities to the assigned functional roles in mammalian glia, with ensheathing glia as resident phagocytes, engulfing the majority or all axonal debris and astrocytes perhaps playing a less important role in phagocytosis of degenerating axons (Doherty, 2009).

It is concluded that ensheathing glia are the phagocytes of the central brain, responsible for engulfing degenerating ORN axons after axotomy. Based on their expression of Draper and dCed-6 it is proposed that cortex glia play a similar role in the cortex, perhaps engulfing degenerating axons in this tissue, or cell corpses generated during neuronal development or after brain injury. Drosophila astrocytes, in contrast, are in close association with synapse rich regions of the brain and it is speculated they likely play an important role in neural circuit and synapse physiology. This work represents the first functional dissection of glial subtypes in the central brain of adult Drosophila, and lays the foundation for future functional studies of these diverse classes of glia (Doherty, 2009).

Effects of Mutation or Deletion

Glia are the most abundant cell type in the mammalian brain. They regulate neuronal development and function, CNS immune surveillance, and stem cell biology, surprisingly little is known about glia in any organism. This study identified over 40 new Drosophila glial genes. glial cells missing (gcm) mutants and misexpression were used to verify they are Gcm regulated in vivo. Many genes show unique spatiotemporal responsiveness to Gcm in the CNS, and thus glial subtype diversification requires spatially or temporally restricted Gcm cofactors. These genes provide insights into glial biology: unc-5 (a repulsive netrin receptor) is shown to orient glial migrations and the draper gene mediates glial engulfment of apoptotic neurons and larval locomotion. Many identified Drosophila glial genes have homologs expressed in mammalian glia, revealing conserved molecular features of glial cells. 80% of these Drosophila glial genes have mammalian homologs; these are now excellent candidates for regulating human glial development, function, or disease (Freeman, 2003).

An antibody was generated to the Draper intracellular domain, which is predicted to recognize all known protein isoforms. Draper protein was detected on the plasma membrane of all glia, including glial membranes ensheathing axon tracts, and outside the CNS on the membranes of macrophages. All Drpr immunoreactivity is absent in drpr^Δ5 mutant embryos. Several draper mutations were generated: a predicted null allele (drpr^Δ5) that is embryonic lethal and two hypomorphic alleles (drpr^EP(3)522 and drpr^Δ19) that are larval lethal. Embryos homozygous for the drpr^Δ5 null allele showed no defects in early CNS development (e.g., glial and neuronal specification, migration, proliferation), but did show clear defects in CNS cell corpse engulfment. Wild-type embryos had 25.9 ± 0.8 cell corpses per hemisegment, while drpr^Δ19 had 31.0 ± 0.9, and drpr^Δ5 had 43.6 ± 1.7. Consistent with this phenotype, strong Draper immunoreactivity was found on vesicles within glia that contain neuronal cell corpses. Together, these data indicate that draper is a downstream target of Gcm in glia and macrophages, it encodes a Ced-1-like transmembrane domain receptor expressed on glial and macrophage membranes, and it is required for cell corpse removal in the CNS (Freeman, 2003).

Draper protein is detected at high levels on glial membranes that ensheath motor nerves, so its role in motor neuron function was tested. Flies homozygous for hypomorphic draper alleles died as first instar larvae, which allowed assay of the role of draper in larval locomotion. It was found that drpr^EP(3)522 and drpr^Δ19 larvae were profoundly uncoordinated as judged by quantitative assays or by time lapse recordings. This phenotype is characteristic of a failure in glial ensheathment of neurons and suggests that Draper may be required for maintaining proper motor axonal physiology (Freeman, 2003).

Essential role of the apoptotic cell engulfment genes draper and ced-6 in programmed axon pruning during Drosophila metamorphosis

Inhibition of glial engulfment suppresses pruning of the larval axon branches of γ neurons (Awasaki, 2004). Thus, if drpr has an essential role in glial engulfment, the loss of drpr function would affect pruning. To test this, the axon pruning of γ neurons using a drpr^D5 (Freeman, 2003) null mutant (Awasaki, 2006).

The axon branches of γ neurons were labelled specifically with cytoplasmic green fluorescent protein (cGFP) driven by GAL4-201Y (201Y>cGFP). In the control pupae, most of the larval axons in the MB dorsal lobe were pruned at 18h APF. Only a very small amount of larval γ axons were labeled with GFP at this stage. In comparison, however, most larval axons appeared to remain in the lobes of the drpr^D5 mutant. To quantitatively examine the amount of remaining larval axons, the number of pixels labeled with the GFP signal were compared within the volume of the dorsal lobe using stacks of confocal serial sections. In the control animals, the number of labeled pixels at 18 hr decreased to less than 5% of that of L3, while approximately 60% of the pixels remain labeled in the drpr^D5 mutant pupae at 18h APF. Thus, drpr mutation strongly suppressed axon pruning of the γ neurons. The development of the unpruned γ neurons was further examined in the drpr^D5 mutant in later stages (Awasaki, 2006).

The 201Y>cGFP labels γ neurons intensely through larval to adult stages. In the drpr^D5 mutants, the larval axon branches of the γ neurons remained in both the dorsal and medial lobes at 24h and 48h APF, though their total volume appeared to decrease. From 48h APF onward, 201Y>cGFP labeled a subset of newly emerging α/β neurons. To distinguish these neurons from the γ neurons, two more molecular markers were combined. In the adult brain, anti-FasII antibody labeled γand α/β neurons, but not α'/β' neurons. Anti-Trio antibody, in contrast, labeled γ and α'/β' neurons, but not α/β neurons. In the drpr^D5 young adult within 2 days after eclosionand also in 1 week old mature adults, there were abnormal axon branches outside of the α/β lobes. Because these axon branches were labeled by 201Y>cGFP, anti- FasII, and anti-Trio antibodies, they are likely to be the remnants of larval γ neurons. This finding indicates that without drpr function, at least some of the larval axon branches persist in the adult brain (Awasaki, 2006).

Of note, whereas the axons of the left and right β lobes never cross the midline in the control animals, these lobes were frequently fused in the midline in the drpr^D5 mutants. Fusion of the β lobes was observed after 48h APF, and was likely due to an overshoot of the medial axons of the α/β neurons. Thus, the drpr^D5 mutation affects not only the pruning of the larval axons, but also the precise formation of the adult neural circuits (Awasaki, 2006).

Next, the effect of the drpr^D5 mutation on the engulfing action of glia was examined. Once glial processes infiltrate the MB lobes, they form lumps to engulf and degrade clusters of γ neuron axon branches (Awasaki, 2004). To evaluate this process quantitatively, the amount of the infiltrating glial processes in the distal tip region of the MB dorsal lobe was measured. The glial processes and γ neurons were labeled with cGFP driven by GAL4-repo (repo>cGFP) and anti-FasII antibody, respectively. In the control pupae at 6h APF, the number of pixels occupied by the infiltrating glial processes corresponded to approximately 25% of the whole volume of the distal tip region. In the homozygous drpr^D5 mutant, less than 4% of the pixels were labeled by infiltrating glia. Though infiltrating glial processes were observed in a slightly larger number of pixels at 18h APF, it was still less than 7% of the whole volume of the distal tip. At this stage only a small amount of glial infiltration was observed in the control pupae. In the drpr^D5 mutant, the dorsal lobes were occupied by unpruned axon branches of the larval γ neurons. In contrast, the lobe structure was essentially collapsed in the control pupae, because most of the larval axon branches had disappeared (Awasaki, 2006).

In the early phase of axon pruning, glial processes form lumps that engulf clusters of varicosities on the axon branches (Awasaki, 2004). Because the diameter of the largest varicosities was less than 2 mm, the number of glial lumps that were larger than 2 mm were counted. In the control pupae at 6h APF, an average of 10.2 lumps was observed in the distal tip region. Several glial lumps were very large in size; an average of 2.0 large glial lumps were over 5 mm in diameter (Awasaki, 2006).

In the homozygous drpr^D5 mutant pupae, the formation of the glial lumps was greatly suppressed. There were very few glial lumps over 2 mm and no lumps over 5 mm at both 6h and 18h APF. There were a few small lumps that were less than 2 mm in diameter in the MB lobe at these stages. Although such small lumps were frequently observed in the control pupae, it was difficult to count their number precisely, because some of them were fused with larger lumps (Awasaki, 2006).

The engulfing action of glia, however, was not completely suppressed; there was an average of 0.2 and 0.5 glial lumps over 2 mm per MB lobe in the drpr^D5 pupae at 6h and 18h APF, respectively. In the drpr^D5 mutant, however, glial lumps were observed only near the periphery of the MB lobes in the vicinity of the glial cell bodies. In contrast, in the control pupae, engulfing glial lumps formed deep inside the lobes. In the control MB lobes, several clusters of regions were observed where FasII-positive axons were absent, apparently because of glial cell engulfment. Such FasII-negative clusters were not observed in the lobes of the drpr^D5. These findings suggest that both the infiltration of glial processes and the engulfment of larval axon branches are severely impaired in the drpr^D5 mutants (Awasaki, 2006).

It is possible that glial infiltration does occur, but that these processes become so thin that the GFP fluorescence is below the limits of detection. In fact, cGFP tends to visualize thicker processes strongly but thinner ones relatively weakly. To visualize thin glial processes more specifically, expression of the cytoskeleton-targeted Actin::GFP and membrane-tagged mCD8::GFP reporters were induced in glial cells using GAL4-repo in the drpr^D5 mutant. Although these reporters should label thin cellular processes very well, no signal of the infiltrating glial processes was detected in the lobes. Thus, the infiltration and engulfing action of glia is suppressed in drpr^D5 mutants (Awasaki, 2006).

Compared to Drpr, functional analysis of Ced-6 was difficult, because there are no ced-6 loss-of-function mutants available. Moreover, careful examination of confocal serial sections revealed that the anti-Drpr antibody labeled not only glial cells, but also weakly labeled the cell bodies of most neurons. It is therefore possible that some of the drpr^D5 mutant phenotypes observed might actually be due to the loss of neuronal drpr expression. Weak neuronal expression was also observed with the anti-Ced-6 antibody. To disturb drpr and ced-6 function specifically in the glial cells, RNA interference (RNAi) was induced in the glial cells by expressing double-stranded RNA of each gene with GAL4-repo. The effect of RNAi was examined by quantitative RT-PCR of the RNA extracted from the whole brain of 6h APF pupae. Whereas little drpr transcript was detected in the homozygous drpr^D5 mutant, the expression level of the drpr and ced-6 mRNA was reduced by approximately 25% and 65% with respective glial RNAi. Antibody staining no longer detected Drpr and Ced-6 proteins in the glial cells surrounding the MB dorsal lobe, whereas the weak staining in neurons was not affected. Therefore, it was concluded that glial RNAi effectively suppressed the specific expression of each gene in the glial cells. In glial RNAi of drpr, infiltration of the glial processes was severely suppressed, and there were only a few glial lumps over 2 mm and no lumps over 5 mm at 6h and 18h APF. There were only a few small glial lumps smaller than 2 mm. The phenotypes caused by glial RNAi were essentially the same as that of the homozygous drpr^D5 mutant. These results strongly suggest that the expression of drpr in the specific subset of the glial cells, but not in neurons, is essential for the engulfing action of glia (Awasaki, 2006).

The infiltration of glial processes and the formation of glial lumps were also severely suppressed in glial RNAi of ced-6 at 6h APF. At 18h APF, the amount of glial infiltration and the number of observed glial lumps somewhat increased. The number of observed glial lumps, however, was less than half of that observed in the control pupae at 6h APF (Awasaki, 2006).

Both Drosophila drpr and C. elegans ced-1 are essential for the engulfment of apoptotic cells (Ellis, 1991; Zhou, 2001; Freeman, 2003; Manaka, 2004). Thus, the glial cells might infiltrate the MB lobes to engulf the axons of the apoptotic neurons rather than to prune the axons of living neurons. Because γ neurons re-extend their axons after pruning of their larval axon branches, it is likely that most of them do not undergo apoptosis. This, however, has not been confirmed (Awasaki, 2006).

To test this, the existence of apoptotic cells was examined with anti-active caspase-3 antibody, which detects activated drICE and DCP1 caspases, in the cell body cluster of γ neurons at 0h, 6h, 12h, 18h, and 24h APF. Although there was a small number of cells labeled around the cluster of the MB neuron cell bodies, no cells were colabeled with 201Y>cGFP at the stages examined. This suggests that most of the γ neurons did not undergo apoptosis during early pupal stages. Furthermore, no axons of γ neurons in the MB lobes were labeled with the antibody, suggesting that few cells contained activated caspases. To further confirm the independence between glial engulfment and apoptosis, two caspase inhibitors, virus-derived p35 and Drosophila DIAP1, were expressed in the γ neurons. This did not affect the pruning of the axon branches and the engulfing action of glia. These results strongly suggest that glial engulfment of MB axon branches is independent of apoptosis and activation of caspases in the MB γ neurons. Therefore, glial cells do not engulf apoptotic neurons, but rather axons of the living neurons (Awasaki, 2006).

Drpr and Ced-6 were colocalized in the infiltrating glial processes, and the loss of either protein caused similar phenotypes. It was therefore asked whether there is a genetic interaction between drpr and ced-6 by examining the effect of the heterozygous drpr^D5 mutation on glial RNAi of ced-6. In the heterozygous drpr^D5 mutant, glial infiltration and lump formation were not affected. In ced-6 RNAi pupae at 18h APF, the suppression of the glial actions was weaker than that at 6h APF. When ced-6 RNAi and heterozygous drpr^D5 mutation coincided in the pupae, however, both glial infiltration and lump formation were significantly reduced. These findings suggest that drpr genetically interacted with ced-6 in the glial engulfment of the larval axon branches. Ced-6, an adaptor molecule, contains the phosphotyrosine binding (PTB) domain, which potentially interacts with the intracellular domain of Drpr, a scavenger receptor-like molecule. The purified intracellular region of Drpr was retained more effectively by Sepharose containing the GST-fused N-terminal half of Ced-6 than Sepharose containing either GST alone or the GST-fused C-terminal half of Ced-6. These suggest that the N-terminal half of Ced-6 might bind the intracellular region of drpr (Awasaki, 2006).

Even when the glial engulfing action was disrupted in the drpr^D5 mutant, the amount of the residual larval axon branches decreased only gradually during metamorphosis. This suggests that some of the larval axon branches can be pruned, albeit very slowly, even when the engulfing action of the glial cells is impaired (Awasaki, 2006).

Disruption of the microtubule cytoskeleton occurs in the early phase of larval axon pruning. Therefore whether this disruption is induced by glial engulfment or occurs independently from the glial action was examined. To visualize the microtubule cytoskeleton, Myc::α-tubulin fusion protein was expressed together with cGFP in γ neurons. In the control animals, Myc::α-tubulin was detected abundantly in the γ neuron axons in L3 but diminished at 6h APF. When Myc::α-tubulin was expressed in the drpr^D5 mutant, it was detected abundantly at L3 and diminished at 6h APF, just as in controls. There were no significant differences in the ratio of Myc::α-tubulin and cGFP signals in the MB dorsal lobe between control and drpr^D5 mutant pupae at L3 or 6h APF, indicating that disruption of the cytoskeleton occurs normally in the drpr^D5 mutant. This finding suggests that microtubule disruption occurs independent of glial engulfment (Awasaki, 2006).

If disruption of the microtubule cytoskeleton is not regulated by glial cells, it might be cell-autonomously regulated. Thus, Myc::α-tubulin and EcR-DN were coexpressed in γ neurons. The Myc::α-tubulin level was not decreased in the MB lobe at 6h APF. The ratio of Myc::α-tubulin and GFP signals in the MB dorsal lobe was not significantly different between L3 and 6h APF. This suggests that disruption of the microtubule cytoskeleton is cell-autonomously regulated by ecdysone through its receptors in γ neurons. These findings demonstrate that ecdysone orchestrates the disruption of the microtubule cytoskeleton in larval axons and the expression of drpr in a specific subset of glial cells in the same period to enable the pruning of unwanted larval axon branches with precise developmental timing. Glial cells therefore engulf degenerating axons in which the microtubule cytoskeleton has already been disrupted (Awasaki, 2006).

The Drosophila cell corpse engulfment receptor Draper mediates glial clearance of severed axons

Neuron-glia communication is central to all nervous system responses to trauma, yet neural injury signaling pathways remain poorly understood. This study explores cellular and molecular aspects of neural injury signaling in Drosophila. Transected Drosophila axons undergo injury-induced degeneration that is morphologically similar to Wallerian degeneration in mammals and can be suppressed by the neuroprotective mouse Wallerian degeneration slow (Wld^s) protein. Axonal injury elicits potent morphological and molecular responses from Drosophila glia: glia upregulate expression of the engulfment receptor Draper, undergo dramatic changes in morphology, and rapidly recruit cellular processes toward severed axons. In draper mutants, glia fail to respond morphologically to axon injury, and severed axons are not cleared from the CNS. Thus Draper appears to act as a glial receptor for severed axon-derived molecular cues that drive recruitment of glial processes to injured axons for engulfment (MacDonald, 2006).

Whether severed Drosophila axons undergo Wallerian degeneration and whether Drosophila glia respond to axon injury were tested. This study used the adult olfactory system to study neuron-glia interactions following nerve transection. This tissue is well-defined histologically and a number of useful reagents are available to label and manipulate olfactory receptor neurons (ORNs) and glia (MacDonald, 2006).

ORN cell bodies are housed in the third antennal segments or maxillary palps of adult Drosophila with axons projecting to the antennal lobe of the brain via the antennal or maxillary nerve, respectively. Axon projections from ORNs expressing the same odorant receptor (OR) gene converge on common, spatially discrete glomerular targets in the antennal lobe. Conveniently, these subsets of ORNs and their axons can be labeled and genetically manipulated using a number of available OR gene promoter-Gal4 driver lines, which drive UAS-regulated expression in highly reproducible subsets of ORNs. Glial cells in the antennal lobe can be identified based on their expression of the reversed polarity (repo) gene and labeled or manipulated using the repo-Gal4 driver. Glial cell nuclei were found at the edge of the antennal lobe neuropil; glial membranes delineate the boundaries of the antennal lobe and extend into the neuropil where they ensheath individual glomeruli (MacDonald, 2006).

Axon injury was induced by nonlethal surgical ablation of third antennal segments or maxillary palps. This ablation completely removed ORN cell bodies and fully transected the antennal or maxillary nerve, respectively. Degeneration of severed ORN axons and glial responses to axonal injury were monitored in vivo for several weeks after injury. For the majority of the experiments the OR22a-Gal4 driver was used to label subsets of antennal ORNs, or OR85e-Gal4 was used to label subsets of maxillary palp ORNs; however, similar results were obtained using additional OR-Gal4 driver lines (MacDonald, 2006).

Attempts were made to determine the time course and morphology of ORN axon degeneration after nerve transection. Three markers were used to visualize axons: UAS-mCD8::GFP (axonal membranes); UAS-α-tubulin::GFP (axonal microtubule cytoskeleton); and UAS-n-Synaptobrevin::YFP::YFP::YFP (UAS-n-Syb::3XYFP) (axon terminals). In control uninjured animals, OR22a⁺ axons (labeled with mCD8::GFP) had a smooth morphology as they projected across the antennal lobe, and GFP intensity in glomeruli was very strong. However, 1 day after ablation of third antennal segments, axon fibers outside glomeruli appeared highly fragmented and were undetectable at 3 or 5 days after injury. GFP signals in the OR22a-innervated glomeruli (termed DM2) remained near control levels 1 day after injury and were reduced to ~30% of control levels 3 days after injury, and nearly all GFP⁺ material was cleared from the CNS within 5 days. A similar profile of axon removal was observed with α-tubulin::GFP-labeled axons. It was found that axonal fibers were largely fragmented 1 day after injury and were absent by 3 days after injury. Interestingly, 1 day after injury, GFP signals within DM2 glomeruli were already reduced to ~50% of control levels. This observation suggests that the microtubule cytoskeleton may degenerate more rapidly than axonal membranes. GFP signals within DM2 glomeruli were near 15% of control levels; by 5 days after injury nearly all α-tub::GFP was cleared from the CNS. Finally, when OR22a⁺ axon terminals were marked with n-Syb::3XYFP, it was found that clearance of YFP⁺ axonal material occurred within 9 days after injury. Thus, axonal membranes and the microtubule cytoskeleton rapidly degenerate when Drosophila ORN axons are severed, clearance of axonal material within the DM2 glomerulus (i.e., membrane and cytoskeletal elements) was slightly delayed relative to axons outside glomeruli, and nearly all degenerating GFP⁺/YFP⁺-labeled axonal material was cleared from the antennal lobe within 5–9 days after axons were severed. A similar time course and morphology of axon fragmentation was observed when ORN axons were marked with the OR85e-Gal4 driver and ablated maxillary palps (MacDonald, 2006).

Axon degeneration appears to occur simultaneously in all portions of the severed axon rather than in a linear fashion along the length of the axon. For example, maxillary palp ORN axons in the maxillary nerve are closest to the site of injury (i.e., the ablated maxillary palp), while midline-crossing ORN axons within the antennal lobe commissure are the most distal to the site of injury. OR85e⁺ maxillary palp ORNs were labeled with mCD8::GFP, ablated maxillary palps, and it was found that bright GFP⁺ puncta (indicative of fragmentation) first appeared ~4 hr after injury. This occurred coincidently in the maxillary nerve and in midline crossing axons, suggesting that axon fragmentation can be initiated at any point along the axon. It was also found that axon destruction was specific to severed ORN axons. For example, if OR85e⁺ maxillary palp ORNs were labeled with mCD8::GFP and then ablated third antennal segments, OR85e⁺ ORN axons showed normal morphology even 9 days after injury. At this time point all antennal ORN axons were removed from the antennal lobe. Thus the cellular mechanisms mediating clearance of severed axons can discriminate between healthy and injured neurons, which suggests that severed axons are autonomously tagged for clearance from the CNS (MacDonald, 2006).

Degeneration of Severed ORN Axons Is Suppressed by Wld^s or Nmnat

The above data show severed Drosophila axons undergo injury-induced fragmentation that is morphologically similar to mammalian Wallerian degeneration. How similar are these events at the molecular level? Wallerian degeneration had long been thought to represent a simple wasting away of the severed axons due to a lack of nutrients supplied by the cell body. However, when nerves are transected in the spontaneous mutant mouse line C57BL/Wld^s (for Wallerian degeneration slow) severed axons survive in a functionally competent state for weeks after injury (Glass, 1993; Lunn, 1989). This observation suggests that severed axons, rather than simply wasting away, may use an active autodestruction program to drive degeneration. Axon-sparing activity in the C57Bl/Wld^s line has recently been found to map to a novel chimeric protein termed Wld^s, which is generated from the fusion of two genes: Ube4b, an E4 ubiquitin ligase, and Nmnat1, a nicotinamide mononucleotide adenylyltransferase, a NAD⁺ biosynthetic and salvaging enzyme (Conforti, 2000; Mack, 2001). To determine whether Wld^s-modulated mechanisms regulate axon degeneration in Drosophila, the ability of mouse Wld^s to protect severed Drosophila axons from injury-induced degeneration was tested (MacDonald, 2006).

UAS-Wld^s and UAS-mCD8::GFP were coexpressed in antennal ORNs using OR22a-Gal4, ablated third antennal segments, and subsequently scored axon morphology. Control animals (OR22a-Gal4 driving only UAS-mCD8::GFP) exhibited a normal rate of axon degeneration with all mCD8::GFP being absent by 5 days after injury. In striking contrast, Wld^s+ axons did not undergo Wallerian degeneration: animals bearing a single copy of the UAS-Wld^s transgene exhibited normal mCD8::GFP fluorescence in severed ORN axons 5 days after injury, and Wld^s+ axons maintained normal morphology. Neuroprotective effects of Wld^s was further explored by labeling axon terminals (with n-Syb::3XYFP) and the axonal microtubule cytoskeleton (with α-tubulin::GFP) and inducing injury. Impressively, in severed Wld^s+ axons, n-Syb::3XYFP signals appeared morphologically normal and YFP intensity remained at control levels, even 9 days after injury. Similarly, in Wld^s+ axons α-tubulin::GFP labeling appeared morphologically normal 5 days after injury, and GFP intensities in glomeruli remained at levels comparable to uninjured animals. Similar levels of protection of axons and their terminals were found with an independent UAS-Wld^s insertion line (MacDonald, 2006).

How long can Wld^s protect severed Drosophila axons? Severed axon morphology was examined in Wld^s+ ORNs at 20, 30, and 50 days after injury. Remarkably, even 20 or 30 days after antennal ablation, Wld^s+ ORN axons maintained largely normal morphology, and GFP intensity within glomeruli remained close to control levels. By 50 days after injury, the axons of Wld^s+ neurons had degenerated significantly; only ~20% of antennal lobes contained detectable GFP⁺ axonal fibers, but mCD8::GFP signals within glomeruli remained at ~35% of control levels. These observations indicate that mouse Wld^s can protect severed Drosophila axons from Wallerian degeneration for weeks, but severed Drosophila axons ultimately degenerate between 30–50 days after injury. It was also noted that these data show that mCD8::GFP is stable in axons for up to 50 days in the absence of transcription (MacDonald, 2006).

In vitro studies with mammalian DRG neurons suggest that (Nicotinamide mononucleotide adenylyltransferase 1) Nmnat1 provides the activity essential for Wld^s-mediated neuroprotection (Araki, 2004; Wang, 2005). To determine whether Drosophila Nmnat (dNmnat) could protect severed axons from Wallerian degeneration in vivo, a UAS-dNmnat transgenic line was generated, dNmnat was coexpressed in OR22a⁺ ORNs with mCD8::GFP, ablated third antennal segments, and axon morphology was assayed 5 days after injury. dNmnat was found to protect severed axons from Wallerian degeneration 5 days after injury. The efficacy of protection may be reduced compared to Wld^s, since 5 days after injury axon degeneration occurs at some level in dNmnat⁺, but not Wld^s+, axons. Nevertheless, these data argue that Wld^s-mediated axon protection in vivo occurs through dNmnat-dependent mechanisms. Together these data show that severed Drosophila axons undergo Wallerian degeneration, and the observation that Wld^s potently suppresses this process in Drosophila argues strongly that Wld^s-modulated mechanisms of severed axon autodestruction are conserved in Drosophila and mammals (MacDonald, 2006).

Drosophila Glia Rapidly Respond to ORN Axon Injury: Requirements for draper in glial morphological responses to antennal ORN injury

Glial cells are responsible for mediating postinjury events in the mammalian nervous system, but glial responses to injury have not been explored in Drosophila. This study explored morphological and molecular responses of Drosophila antennal lobe glia to axon injury. In control, uninjured animals, glial membranes were found to delineate the borders of the antennal lobe and were intimately associated with antennal lobe glomeruli. However, 1 day after antennal ablation, glia exhibited dramatic changes in morphology and appeared to significantly increase their membrane surface area. This expansion of glial membranes appeared to be a local response since antennal lobe glia, but not glia in surrounding brain regions, responded to ablation of third antennal segments in this way. During this injury response glial nuclei consistently remained at the periphery of the antennal lobe, indicating that glia responded to ORN injury by extending membranes toward severed axons rather than by migrating as an entire cell into the antennal lobe (MacDonald, 2006).

Mammalian microglia and Schwann cells, but not astrocytes, normally proliferate in response to neuronal injury; therefore Drosophila glia were assayed for injury-induced proliferative events. Third antennal segments were ablated, adult brains were costained with anti-Repo antibodies and the mitotic marker anti-phosphohistone H3 (PHH3), and Repo⁺/PHH3⁺ cells were assayed at 1, 3, 5, 7, and 9 days after ORN injury. No examples of mitotic glia were found at any of these time points after ORN injury, no gross increase in glial numbers surrounding the antennal lobe, nor evidence for proliferation in any other cell types in the brain after injury at these time points. Peripheral glial responses along the antennal nerve were assayed. In response to antennal ablation these peripheral glia do not proliferate, and no gross changes were observe in glial numbers along the antennal nerve (MacDonald, 2006).

To gain insight into how Drosophila glia might be responding to severed axons at the molecular level, a collection of embryonically expressed glial genes were assayed for those specifically enriched in antennal lobe glia 1 day after antennal ablation. Among these candidates draper, the Drosophila ortholog of the C. elegans cell corpse engulfment gene ced-1 was detected; draper has been shown to be required for glial engulfment of apoptotic neuronal cell corpses during embryonic development (Freeman, 2003). Such an engulfment receptor is an excellent candidate for driving glial removal of severed axons, though it would be somewhat surprising in light of the fact that CED-1/Draper is thought to play a role in the recognition of cell corpses, and severed axons degenerate via mechanisms that are molecularly distinct from apoptosis (MacDonald, 2006).

Using an antibody specific to Draper it was found that Draper is expressed in all Repo⁺ adult brain glia, including antennal lobe glia. Interestingly, 1 day after ablation of third antennal segments a dramatic increase was observed in Draper immunoreactivity in antennal lobe glia. Similar to the expansion of antennal lobe glial membranes after antennal ablation, increased glial Draper was found to be a local response to injury as only antennal lobe glia exhibited increased Draper after antennal ORN injury. All observed staining represents glial expression of Draper: anti-Draper immunoreactivity was absent in drpr^Δ5 null mutants, and glial specific knockdown of draper mRNA by dsRNAi removes all detectable Draper immunoreactivity in adult brains even after antennal ablation. Together these data indicate that Drosophila glia rapidly respond to ORN axon injury with changes in morphology and Draper protein levels. How glial membranes specifically interact with severed axons has been further explored, and the Draper receptor is shown to be essential for all glia responses to ORN axon injury (MacDonald, 2006).

The third antennal segments each house ~600 ORNs, and their ablation severs axons that project to ~44/50 antennal lobe glomeruli. Such ablations resulted in a robust upregulation of Draper and a dramatic expansion of antennal lobe glial membranes, but precisely how antennal lobe glia were interacting with severed axons under this conditions was unclear. For example, were glial membranes invading the antennal lobe and extending specifically toward severed axons? To explore glial membrane dynamics after ORN axon injury with higher resolution a second ORN injury assay was used: maxillary palp ablation. The maxillary palp houses only ~60 ORNs which collectively innervate ~6 glomeruli, all positioned in the ventro-medial region of the antennal lobe. Ablation of maxillary palps should therefore result in the injury of only this small subset of ORNs. If glial processes are recruited to severed ORNs, it is predicted that glial Draper and membranes would be specifically targeted to this subset of maxillary palp-innervated antennal lobe glomeruli after maxillary palp ablation (MacDonald, 2006).

Prior to injury Draper immunoreactivity was only weakly detectable within antennal lobe glomeruli and along the length of the maxillary nerve. However, 1 day after maxillary palp ablation, Draper immunoreactivity was found at high levels in ~6 ventro-medially positioned antennal lobe glomeruli—one of these was positively identified as maxillary palp-innervated with the OR85e-Gal4 driver—and along the entire length of the maxillary nerve that was visible in preparations. Draper localization to severed axons was observed beginning as early as 4 hr after maxillary palp ablation, colocalizing coincidently with the appearance of GFP⁺ puncta from degenerating axons, and Draper levels on severed axons appeared to be strongest between 12 hr and 1 day after injury. Draper protein was maintained at elevated levels on all injured axonal elements while they were still detectable (by GFP signals) and returned to control levels after axons had been cleared from the CNS. Interestingly, maxillary palp ablation did not lead to a dramatic upregulation of Draper protein in all antennal lobe glia as was seen after ablation of the third antennal segment. Thus, injuring a smaller number of ORN axons through maxillary palp ablation leads to a qualitatively different response by glia in the antennal lobe (MacDonald, 2006).

Glial membranes showed a similar pattern of rapid and specific localization to severed maxillary palp ORN axons after injury. Prior to injury GFP-labeled glial membranes were detectable at low levels around (but not within) glomeruli and at low levels along the maxillary nerve. However, 1 day after ablation of maxillary palps, glial membranes were found to be enriched within ~6 ventro-medially positioned antennal lobe glomeruli and along the entire maxillary nerve. Glial membrane-decorated glomeruli perfectly overlapped with those showing high-level Draper immunoreactivity and presumably represent glomeruli housing severed maxillary palp ORN axons (MacDonald, 2006).

In summary, glial-expressed Draper and glial membranes are rapidly and specifically recruited to severed ORN axons; their localization is coincident with the initiation of severed axon fragmentation; and Draper protein remains associated with degenerating axons until they are cleared from the CNS. These data further demonstrate that severed Drosophila axons generate molecular cues that elicit potent responses from glia and that these neuron→glia injury signals are sufficient to drive the selective recruitment of glial processes to severed axons (MacDonald, 2006).

In C. elegans, CED-1 has been shown to be essential for the engulfment of cell corpses and is believed to act as a recognition receptor for molecular cues presented by corpses (Zhou, 2001). Additional cellular targets for CED-1 have not been identified. This study shows that Draper is required to drive injury-induced changes in glial morphology and for glial clearance of severed axons from the CNS (MacDonald, 2006).

The requirements for draper in glial morphological responses to antennal ORN injury were determined. In control animals 1 day after antennal ablation a widespread expansion of antennal lobe glial membranes was observed. In contrast, when third antennal segments were ablated in drpr^Δ5 mutants, antennal lobe glia showed no noticeable changes in morphology or GFP intensity. It is noted that antennal lobe morphology was grossly normal in drpr^Δ5 mutants: ORN axons projected to the correct glomeruli, glomeruli appeared well-defined morphologically, and glial membrane processes were found in their normal positions ensheathing antennal lobe glomeruli. Together these data indicate that Draper is not required for antennal lobe development, but that glial morphological responses to antennal ORN axon injury require Draper signaling (MacDonald, 2006).

Maxillary palps were ablated and glial recruitment to the severed maxillary palp ORN axons was assayed. In control animals glial membranes were found to be highly enriched in maxillary palp-innervated glomeruli and on the maxillary nerve 1 day after injury. However, these responses were blocked in drpr^Δ5 mutants: glial processes did not accumulate in glomeruli housing severed maxillary palp ORN axons, nor were they recruited at high levels to the maxillary nerve. The autonomy of Draper function was further assayed by glial-specific RNAi knockdown of Draper. It was found that blocking Draper function in glia also fully suppressed glial responses to antennal or maxillary palp ablation, consistent with a requirement for Draper in glia. Thus all morphological changes exhibited by glia after injury (i.e., membrane expansion and process extension toward severed axons) require Draper signaling. These observations suggest that Draper may act as a glial receptor for severed axon-derived cues that drive glial responses to axon injury (MacDonald, 2006).

CED-1/Draper encodes a receptor essential for engulfment of cell corpses in both C. elegans (Zhou et al., 2001) and Drosophila (Freeman, 2003). Is Draper required for glial clearance of degenerating axons from the CNS? To test this the OR22a⁺ subset of antennal ORN axons in the drpr^Δ5 null mutant background was marked with mCD8::GFP and removal of injured axons was assayed. In control animals severed ORN axons outside glomeruli were cleared from the CNS by 3 days after injury (20/20 antennal lobes), and GFP⁺ axonal material was cleared from the CNS within 5 days. In striking contrast, the majority of axonal debris lingered in the CNS of drpr^Δ5 mutants after ablation of the third antennal segments. For example, 3 days after injury there was an abundance of GFP-labeled OR22a⁺ axon fibers remaining in the antennal lobe (20/20 antennal lobes). Similarly, 5 days after injury in drpr^Δ5 mutants, ~65% of antennal lobes retained GFP⁺ axonal fibers and GFP levels in DM2 glomeruli remained at levels comparable to control uninjured animals. Severed axon fragmentation was not blocked in drpr^Δ5 mutants. Severed axons showed signs of fragmentation as early as 5 hr after injury in drpr^Δ5 mutants—this approximates the earliest time points at which axons fragmented in control animals—and axon fibers were highly fragmented in these animals 3 and 5 days after injury. These data indicate that Draper is essential for glial clearance of degenerating axons from the Drosophila CNS, and axons thus represent a new engulfment target for this receptor. These observations also indicate that severed axons actively communicate with glia (via Draper) to promote their clearance from the CNS.

It is proposed that Draper acts as a glial receptor for severed axon-derived molecular cues, and that a Draper ligand encodes a neuron→glia injury signal that drives glial responses to axon injury. The ligand on cell corpses recognized by CED-1 remains to be identified, but is thought to encode the “eat-me” signal that initiates engulfment. Do severed axons, developmentally pruned axons, and cell corpses present similar molecular cues for engulfment? This may indeed be the case since CED-1/Draper is required for the clearance of each of these engulfment targets. However, Wallerian degeneration, developmental axon pruning, and apoptotic cell death are clearly distinct at the molecular level. For example, Wld^s expression in neurons does not block apoptosis in neuronal cell bodies, nor the developmental pruning of Drosophila mushroom body γ neurons. Reciprocally, inhibition of canonical cell death pathways is not sufficient to block Wallerian degeneration. Nevertheless, it is possible that Wallerian degeneration, developmental axon pruning, and apoptotic cell death, though promoted by different molecular mechanisms, lead to the production of the same engulfment cue that acts as a ligand for CED-1/Draper (MacDonald, 2006).

Degenerating axons and cell corpses could also generate distinct engulfment cues that are each recognized by CED-1/Draper. One model would be that CED-1/Draper would not discriminate these targets, but only drive their nonspecific engulfment. Alternatively, distinct Draper receptor isoforms might recognize specific ligands presented by a cell corpse or a severed axon and potentially play an important role in discriminating these engulfment targets. Intriguingly, at least two isoforms of Draper have been identified in Drosophila that vary significantly in their extracellular domains; perhaps these bind distinct ligands on engulfment targets in vivo. Future studies aimed at defining the functional requirements of specific Draper receptor isoforms in the engulfment of cell corpses, developmentally pruned axons, and severed axons, as well as studies aimed at identifying the CED-1/Draper ligand(s) presented by these engulfment targets, should help resolve these issues (MacDonald, 2006).

Glial Responses to Severed Axons Are Suppressed by Wld^s and dNmnat

Draper becomes localized to severed axons ~4 hr after injury, coincident with the initiation of axon fragmentation. Is axon degeneration essential to recruit glial processes in Drosophila? Mouse Wld^s and dNmnat were used as tools to block ORN axon degeneration and glial responses were assayed to these severed, but nondegenerating, axons. UAS-Wld^s or UAS-dNmnat were coexpressed with UAS-n-Syb::3XYFP in OR85e⁺ maxillary palp ORNs, ablated maxillary palps, and subsequently Draper localization to severed axons was assayed. Interestingly, while Draper was detected at high levels 1 day after maxillary palp ablation in 5/6 maxillary palp ORN-innervated glomeruli, Wld^s-expressing axons showed control levels of Draper immunoreactivity. Thus Wld^s expression is sufficient to potently suppress glial recruitment to severed axons. dNmnat was also found to suppress glial recruitment to severed axons, though, as was found in severed axon protection experiments, dNmnat was not as efficient at suppressing glial recruitment to severed axons. Nevertheless, this observation suggests that dNmnat activity is an important requirement for Wld^s to block glial responses to severed axons. Together these data indicate that production of the molecular cues in severed axons that elicit glial responses is genetically downstream of Wld^s and are consistent with axon fragmentation being essential for robust glial responses to injury. In addition, the results indicate that Wld^s acts in a cell-autonomous fashion in Drosophila, as the recruitment of glial processes to severed axons is suppressed in Wld^s-expressing axons but not in adjacent wild-type injured axons (MacDonald, 2006).

The mechanism by which Wld^s exerts its neuroprotective effects remains unclear. One model proposes that Wld^s acts prior to injury in the nucleus through the NAD binding histone deacetylase Sirt1 to effect neuroprotective changes in gene expression. In contrast, a second model proposes that Wld^s acts locally in axons after injury to maintain high NAD levels which ultimately block axonal degeneration. Although transfection of Nmnat1 into DRG explant cultures indeed suppressed Wallerian degeneration in vitro, it was not known whether Nmnat1 could protect severed axons in vivo. This study has shown that dNmnat can suppress severed axon autodestruction in Drosophila; however, dNmnat appears less efficient than Wld^s in protecting axons and suppressing glial responses to axon injury. The results therefore support the notion that Nmnat activity is an important component of Wld^s neuroprotective function in vivo, but indicate that Wld^s somehow provides more effective neuroprotection than dNmnat (MacDonald, 2006).

Wld^s protection distinguishes axon degeneration following injury from naturally occurring developmental pruning

Axon pruning by degeneration remodels exuberant axonal connections and is widely required for the development of proper circuitry in the nervous system from insects to mammals. Developmental axon degeneration morphologically resembles injury-induced Wallerian degeneration, suggesting similar underlying mechanisms. As previously reported for mice, this study shows that Wld^s protein substantially delays Wallerian degeneration in flies. Surprisingly, Wld^s has no effect on naturally occurring developmental axon degeneration in flies or mice, although it protects against injury-induced degeneration of the same axons at the same developmental age. By contrast, the ubiquitin-proteasome system is intrinsically required for both developmental and injury-induced axon degeneration. The glial cell surface receptor Draper is required for efficient clearance of axon fragments during developmental axon degeneration, similar to its function in injury-induced degeneration. Thus, mechanistically, naturally occurring developmental axon pruning by degeneration and injury-induced axon degeneration differ significantly in early steps, but may converge onto a common execution pathway (Hoopfer, 2005).

Selective pruning of exuberant axons and axon branches occurs widely from insects to mammals and plays an essential role in the proper wiring of the nervous system. A major form of developmental axon pruning occurs via degeneration, qualitatively characterized by axon fragmentation and the subsequent removal of fragments by other cells. Axon degeneration was first reported in the large-scale remodeling of the retinotopic map in the optic tectum/superior colliculus (SC). Recently, pruning of Drosophila mushroom body (MB) γ neuron axons was demonstrated to occur via degeneration, as shown by axon fragmentation and glial engulfment of axon fragments. Developmental axon pruning can also be achieved by a distal-to-proximal retraction of the axon. Axon retraction is used in the stereotyped pruning of the infrapyramidal bundle of hippocampal mossy fibers, and has been directly observed for short branches of Cajal-Retzius and thalamocortical axons in the developing cerebral cortex. In the same study, pruning of long branches of thalamocortical axons was found to occur via degeneration, supporting the general trend that pruning of long portions of primary axons or their collateral branches occurs via degeneration. Additionally, synapse elimination at the developing neuromuscular junction in mammals occurs via an “axosome shedding” mechanism: motor axon terminals appear to undergo a distal-to-proximal retraction, but shed membrane-enclosed axosomes at the distal ends, which are engulfed by surrounding Schwann cells. Thus, although multiple mechanisms exist, axon degeneration is commonly used for developmental pruning, particularly for longer axon segments (Hoopfer, 2005 and references therein).

Axon degeneration during developmental pruning in flies and mammals occurs within a short period of time along the entire length of the axon segments to be pruned. This resembles fragmentation of distal axons in response to axon severing, a process known as Wallerian degeneration. Wallerian degeneration is characterized by the rapid breakdown of the cytoskeleton and fragmentation along the entire length of axon segments distal to the injury, usually occurring after a variable waiting period after the initial insult. Indeed, the progression of axonal fragmentation occurs with such rapidity that most axons appear either fully intact or completely fragmented at a given time postinjury. Axon fragmentation is followed by a series of responses from surrounding cells, including recruitment of microglia and macrophages to the severed axons. These morphological similarities between axon degeneration during developmental pruning and after injury raise the possibility that they share similar mechanisms (Hoopfer, 2005 and references therein).

Studies of Wallerian degeneration slow (Wld^s) mice have provided major mechanistic insights into Wallerian degeneration. Compared to the rapid degeneration of severed axons within days in wild-type (wt) mice, Wld^s mice exhibit much slower Wallerian degeneration: distal parts of severed axons persist for one to two weeks without obvious signs of fragmentation. Wld^s is a dominant mutation resulting in the overexpression of a fusion protein of the first 70 amino acids of UFD2/E4, an evolutionarily conserved protein used in protein polyubiquitination, and the full-length nicotinamide mononucleotide adenylyltransferase (Nmnat), an enzyme that facilitates NAD synthesis. The effect of Wld^s is autonomous to injured neurons, and neuronal expression of Wld^s protein protects against the degeneration of severed distal axons in a dose-dependent manner. In vitro studies suggest that the Nmnat part of the fusion protein is responsible for its protective effect (Araki, 2004; Wang, 2005). Remarkably, Wld^s also has protective effects in a variety of disease models, including motor axon “dying back” caused by a tubulin chaperone mutation, axon atrophy caused by demyelination, axonal spheroid pathology in gracile axonal dystrophy mice, and axon degeneration in a Parkinson's disease model (Hoopfer, 2005 and references therein).

Given its widespread protective effect against injury- and disease-induced axon degeneration, can Wld^s interfere with developmental axon pruning that occurs by degeneration? How similar are the mechanisms involved in axon degeneration during developmental pruning and after injury? This study reports that in Wld^s mice, pruning of both the retinocollicular and Layer 5 subcortical projections occur normally, indistinguishable from wt mice in the extent and timing. By contrast, Wld^s protects axons of developing retinal ganglion cells (RGCs) from degeneration following complete transection of the optic nerve in the first postnatal week, the same developmental age at which Wld^s has no effect on the naturally occurring developmental degeneration of these same axon projections. It is further shown that the expression of mouse Wld^s protein has no effect on developmental Drosophila MB γ axon pruning, despite its potent effect in protecting fragmentation of severed axons in a Drosophila injury model (see also MacDonald, 2006. By comparing axon degeneration during developmental pruning and after injury, it was found that they share similarities in the neuron-intrinsic requirement of the ubiquitin-proteasome system (UPS) and Draper-mediated clearance of degenerating axon fragments by glia, but differ substantially in early steps that trigger axon fragmentation. These data provide insights into both developmental axon pruning and Wallerian degeneration (Hoopfer, 2005).

Therefore naturally occurring developmental pruning of axons resembles injury-induced Wallerian degeneration of axons in adults. These two processes share some molecular mechanisms: they both depend on UPS action in neurons and axon fragment clearance by the cell corpse engulfment receptor Draper in glia. However, this analysis also revealed fundamental differences between developmental and injury-induced axon degeneration. Whereas expression of Wld^s protein potently delays axon degeneration after injury in adult flies and in both developing and adult mice, it does not affect axon degeneration in three developmental axon-pruning paradigms in either organism. It is proposed that axon degeneration during developmental pruning and after injury differ in the early steps leading to axon fragmentation but may later converge onto a common pathway (Hoopfer, 2005).

Beyond the morphological features of axon fragmentation and engulfment of fragments by extrinsic cells, axon degeneration during developmental pruning and after injury share additional similarities. Both processes require cell-autonomous programs, neither requires caspase-dependent apoptotic machinery, and microtubule fragmentation/disappearance is an early sign of axon degeneration in both cases. By systematically comparing developmental and injury-induced axon degeneration in Drosophila, additional mechanistic similarities were found that likely extend to vertebrate systems (Hoopfer, 2005).

MB γ axon pruning requires the intrinsic activity of the UPS (Watts, 2003). UPS inhibition significantly delays injury-induced axon degeneration. At least at early stages, UPS inhibition appears to delay axon fragmentation. This is similar to an in vitro Wallerian degeneration model in mammals, in which inhibition of UPS by application of a proteasome inhibitor or expression of UBP2 delays axon fragmentation. These studies suggest that UPS action in facilitating axon fragmentation after injury is a conserved feature from flies to mammals (Hoopfer, 2005).

Although UPS inhibition can block developmental axon pruning completely under appropriate conditions (Watts, 2003), it cannot prevent injured axons from eventual degeneration. One possible explanation for this difference could be because of the short window of time in which developmental pruning occurs: if pruning is delayed beyond this “critical period,” neurons may no longer be in a competent state to carry out the degenerative pruning program. Consistent with this notion is the recent finding of a critical period for large-scale axon degeneration of the retinocollicular projection coincident with the brief time window during which the projection normally undergoes pruning. In the case of Wallerian degeneration after complete axon transection, since the injury physically separates the axons from their cell bodies, the most that can be done is to prolong the inevitable onset of fragmentation of the already severed axon. However, many neurodegenerative disease models exhibit Wallerian-like axon degeneration, which can be ameliorated by Wld^s expression. Thus, understanding the mechanisms that trigger axon fragmentation may lead to therapeutic strategies to stop axon degeneration before it begins (Hoopfer, 2005).

MacDonald (2006) has demonstrate a role for glia in the clearance of axon fragments in a Drosophila model of Wallerian degeneration. In particular, a requirement of the cell corpse engulfment receptor Draper in glia for this function was demonstrated. This study finds that removal of Draper in glia markedly delays clearance of axon fragments during MB axon pruning. In addition, ultrastructural analysis of degenerating axons after injury reveals degenerating profiles similar to what have been found during MB axon pruning (Watts, 2004). These similarities strongly suggest that mechanisms for this late phase of axon degeneration and clearance are shared between developmental pruning and Wallerian degeneration (Hoopfer, 2005).

Glial function has not been systematically explored in developmental axon pruning in mammals. However, the transient appearance in early postnatal cats of clusters of macrophages contiguous to the corpus callosum during the period of callosal axon pruning suggests that these cells might serve a phagocytic function in vertebrate developmental axon pruning analogous to that of Draper-expressing glia in flies (Hoopfer, 2005 and references therein).

An important finding of this study is that developmental axon degeneration and Wallerian degeneration can be clearly distinguished mechanistically by their differential sensitivity to the protective effect of Wld^s. Remarkably, this mouse fusion protein can also potently protect axon degeneration after injury in flies. Despite the phylogenetically conserved function of Wld^s in protecting injured axons, this study shows in three diverse developmental degeneration paradigms in mice and flies that Wld^s expression has no effect on naturally occurring developmental axon pruning. In particular, the injury-induced degeneration of transected RGC axons is markedly protected in Wld^s mice at the same age when naturally occurring degenerative pruning of the same developing RGC axons is not protected. This finding demonstrates unequivocally the fundamental difference between axon degeneration during developmental pruning and after injury in the same type of neuron at the same developmental stage. Similarly, Wld^s protects young motor axons from degeneration after injury while having no effect in the remodeling of neuromuscular junction, although such remodeling uses a distinct mechanism (Hoopfer, 2005).

What might be the reasons behind this difference between developmental and injury-induced axon degeneration? The trigger for naturally occurring developmental axon degeneration and injury-induced Wallerian degeneration differs, as do their respective roles. Wallerian degeneration is triggered extrinsically by axonal insult. Because degeneration is restricted to the portion of the axon distal to the injury site, it must be executed without new transcription or transport from the cell body. These properties suggest that the effectors of the program are continuously present along the entire length of axons. The function of Wallerian degeneration could be to remove the damaged axons as a step in facilitating regeneration and repair (Hoopfer, 2005).

In contrast to Wallerian degeneration, the role of developmental axon degeneration is to remodel initially exuberant neuronal connections into a functionally appropriate adult circuit. It is likely triggered by diverse mechanisms including cell-autonomous transcriptional regulation, patterned neuronal activity, the local environment of the axon, or their combination. For example, MB γ axon pruning is triggered by activation of the steroid hormone receptor EcR to cell-autonomously regulate gene expression. In the case of the retinocollicular projection, spatial control of pruning is instructed by the distribution and levels of EphA receptors along the length of RGC axons, which is controlled in part by transcriptional regulation autonomous to the RGC. Non-RGC autonomous mechanisms, including the level of ephrin-As to which an RGC axon segment is exposed within the target as well as correlated patterns of RGC activity generated by networks of RGCs and cholinergic amacrine cells that produce spontaneous retinal waves, are also essential factors for the pruning of inappropriate axon segments (Hoopfer, 2005).

It remains a future challenge to investigate mechanistically how diverse triggers in developmental pruning, injury, and disease lead to similar late stage execution of axon degeneration. The findings of both a differential effect of Wld^s in these processes and similar roles for the intrinsic activity of UPS and glial clearing of fragmented axons provide an essential step toward this goal (Hoopfer, 2005).

Draper-mediated and phosphatidylserine-independent phagocytosis of apoptotic cells by Drosophila hemocytes/macrophages

The mechanism of phagocytic elimination of dying cells in Drosophila is poorly understood. This study was undertaken to examine the recognition and engulfment of apoptotic cells by Drosophila hemocytes/macrophages in vitro and in vivo. In the in vitro analysis, l(2)mbn cells (a cell line established from larval hemocytes of a tumorous Drosophila mutant) were used as phagocytes. When l(2)mbn cells were treated with the molting hormone 20-hydroxyecdysone, the cells acquired the ability to phagocytose apoptotic S2 cells, another Drosophila cell line. S2 cells undergoing cycloheximide-induced apoptosis exposed phosphatidylserine on their surface, but their engulfment by l(2)mbn cells did not seem to be mediated by phosphatidylserine. The level of Croquemort, a candidate phagocytosis receptor of Drosophila hemocytes/macrophages, increased in l(2)mbn cells after treatment with 20-hydroxyecdysone, whereas that of Draper, another candidate phagocytosis receptor, remained unchanged. However, apoptotic cell phagocytosis was reduced when the expression of Draper, but not of Croquemort, was inhibited by RNA interference in hormone-treated l(2)mbn cells. Whether Draper is responsible for the phagocytosis of apoptotic cells in vivo was examined using an assay for engulfment based on assessing DNA degradation of apoptotic cells in dICAD mutant embryos (which only occurred after ingestion by the phagocytes). RNA interference-mediated decrease in the level of Draper in embryos of mutant flies was accompanied by a decrease in the number of cells containing fragmented DNA. Furthermore, histochemical analyses of dispersed embryonic cells revealed that the level of phagocytosis of apoptotic cells by hemocytes/macrophages was reduced when Draper expression was inhibited. These results indicate that Drosophila hemocytes/macrophages execute Draper-mediated phagocytosis to eliminate apoptotic cells (Manaka, 2004).

It has been suggested that there are two signaling pathways that lead to the induction of engulfment in phagocytes. These pathways are most likely to be triggered by distinct ligands, to be mediated by distinct receptors, and to involve partially overlapping signal mediators and downstream uptake mechanisms. The membrane receptors called CED-1 and phosphatidylserine receptor PSR-1 (Wang, 2003) have been genetically shown to act farthest upstream in the pathways in C. elegans (Manaka, 2004 and references therein).

In Drosophila, Croquemort, a member of the CD36 family of proteins, is the only protein known to be involved in phagocytosis of apoptotic cells by hemocytes/macrophages. However, Croquemort appears to be structurally unrelated to either CED-1 or PSR-1, and the identity of its ligand or how it transduces signal(s) for phagocytosis remains unknown. A Drosophila homologue of C. elegans CED-1 has been suggested to play a role in phagocytosis by glia (i.e. loss of Draper expression caused an increase in the number of apoptotic neurons in the central nervous system of Drosophila embryos). This study confirms the role of Draper in glia in a more direct analysis and, more importantly, found that Draper is also involved in the phagocytosis of apoptotic cells by Drosophila hemocytes/macrophages. This indicates that at least one of the two signaling pathways for phagocytosis of apoptotic cells in C. elegans exists in Drosophila. In contrast, the existence of a Drosophila homologue of the mammalian PS receptor has been noted, but its role in the phagocytosis of apoptotic cells in Drosophila remains to be determined. The finding that Draper was not solely responsible for the phagocytosis by larval hemocyte-derived cell line l(2)mbn, embryonic hemocytes/macrophages, or embryonic glia suggests the presence of other phagocytosis receptors in those phagocytes. It is, however, unlikely, at least for l(2)mbn cells, that a homologue of the PS receptor is a putative additional phagocytosis receptor, because phagocytosis by 20-hydroxyecdysone-treated l(2)mbn cells was shown not to be mediated by PS present on the surface of target apoptotic cells. At this point, it is not clear whether or not PS serves as a phagocytosis marker at all in Drosophila (Manaka, 2004).

It is most probable that Draper induces engulfment of apoptotic cells in phagocytes by recognizing a specific ligand (i.e. a 'phagocytosis marker' molecule present on the surface of target cells). The ligand for Draper/CED-1, however, remains to be identified. PS appears not to be recognized by this receptor; this study that phagocytosis of PS-exposing apoptotic S2 cells by 20-hydroxyecdysone-treated l(2)mbn cells is mediated by Draper with no dependence on PS. A mammalian membrane protein called low density lipoprotein receptor-like protein (LRP, also known as CD91) shares some structural similarity with CED-1 and is considered to be its mammalian homologue (Su, 2002). Ogden (2001) has suggested that CD91 acts as a phagocytosis receptor due to its association with calreticulin, a molecular chaperone present in the endoplasmic reticulum. In addition, the involvement of calreticulin in the phagocytosis of yeast by hemocytes of another insect, Pieris rapae, has been reported. A Dictyostelium mutant lacking both calreticulin and calnexin, another molecular chaperone in the endoplasmic reticulum, shows a reduced level of phagocytosis of yeast, although both of these proteins appear to act not at the recognition step but at some downstream events during phagocytosis by Dictyostelium. Furthermore, like calreticulin, calnexin has been shown to translocate to the surface of mammalian cells under certain circumstances. Calreticulin and calnexin, which both exist in Drosophila, are thus good candidates for the ligand of Draper/CED-1 (Manaka, 2004 and references therein).

Treatment with 20-hydroxyecdysone does not alter the general phagocytic activity of l(2)mbn cells but augments the level of phagocytosis of apoptotic cells or zymosan particles. This suggests that the hormone causes quantitative and/or qualitative changes in l(2)mbn cells that are specific, if not solely responsible, for phagocytosis of apoptotic cells. A simple explanation would be an increase in the amount of Draper present at the surface of the phagocyte, but this was shown not to be the case. In contrast, the amount of Croquemort in l(2)mbn cells increases upon treatment with the hormone, but the results of an RNA interference experiment showed that this receptor was not involved in the phagocytosis of apoptotic cells by l(2)mbn cells. Another possibility is that Draper is functionally activated in hormone-treated l(2)mbn cells. This might be true, because it was found that migration of Draper protein in a polyacrylamide gel becomes a little bit faster after treatment with 20-hydroxyecdysone. Although no evidence of modification of CED-1 or Draper has been reported so far, the results suggest that Draper is structurally modified, through for example dephosphorylation or deglycosylation, and activated by 20-hydroxyecdysone (Manaka, 2004).

The innate immune system, like adaptive immunity, consists of humoral and cellular reactions. In Drosophila, the humoral responses, most of which are conducted through the action of the fat body (a functional equivalent of the mammalian liver), include production of antimicrobial peptides and melanization at wound sites or around microorganisms. In contrast, the cellular responses in Drosophila consist of phagocytosis and encapsulation of microorganisms by hemocytes, both of which events lead to killing of the invaders. Although the mechanism of the humoral responses has been well characterized, how Drosophila hemocytes act against invading microorganisms remains to be clarified. Phagocytosis of apoptotic cells by hemocytes is considered part of the cellular immune response of Drosophila. The data showed that Draper is involved in hemocyte phagocytosis of apoptotic cells but not of zymosans. It is thus anticipated that distinct systems are utilized by Drosophila hemocytes for the recognition of endogenous apoptotic cells and invading microorganisms. To clarify these recognition systems is necessary for a better understanding of cellular innate immunity in Drosophila (Manaka, 2004).

Six-microns-under acts upstream of Draper in the glial phagocytosis of apoptotic neurons

The removal of apoptotic cells by phagocytic neighbors is essential for metazoan development but remains poorly characterized. This study reports the discovery of a Drosophila phagocytosis receptor, Six-microns-under (SIMU), which is expressed in highly phagocytic cell types during development and required for efficient apoptotic cell clearance by glia in the nervous system and by macrophages elsewhere. SIMU is part of a conserved family of proteins that includes CED-1 and Draper (DRPR). Phenotypic analysis reveals that simu acts upstream of drpr in the same pathway and affects the recognition and engulfment of apoptotic cells, while drpr affects their subsequent degradation. SIMU strongly binds to apoptotic cells, presumably through its EMILIN-like domain, but requires no membrane anchoring, suggesting that it can function as a bridging molecule. This study introduces an important factor in tissue-resident apoptotic clearance and underscores the prominent role of glia as 'semiprofessional' phagocytes in the nervous system (Kurant, 2008).

The elimination of superfluous or damaged cells through programmed cell death plays an essential role in metazoan development and tissue homeostasis; its critical final stage is the clearance of the apoptotic cells through phagocytosis. The proper recognition, uptake, and degradation of dying cells is accomplished either by 'professional' phagocytes, such as macrophages and immature dendritic cells, or by 'nonprofessional' tissue-resident neighboring cells. While professional phagocytes have been studied extensively, relatively little is known about the biological significance and the molecular underpinnings of tissue-resident phagocytosis (Kurant, 2008 and references therein).

Apoptotic cell clearance is a complex process, involving recognition, engulfment, phagosome formation and maturation as distinct steps. The apoptotic cell displays distress ('eat me') signals that are recognized by the phagocyte, either directly by phagocytic receptors or indirectly through bridging molecules (opsonins), supplied systemically through the serum or secreted locally by the phagocyte. Two types of phagocytic receptors have been implicated in this recognition process: tethering receptors without a significant intracellular domain, such as CD36 or SRA, and docking receptors with noncatalytic intracellular domains permitting interaction with other proteins, such as CED-1 and its homolog Draper (DRPR), or LRP. The clustering of both types of receptors is thought to be required for the recruitment of the downstream machinery to the docking sites, which leads to cytoskeletal reorganization and engulfment of the apoptotic cell. The phagocytosis process is completed by the formation of a phagosome and its maturation to a phagolysosome, effecting the degradation of the apoptotic particle (Kurant, 2008).

Insight into the molecular mechanisms by which nonprofessional tissue-resident cells effect apoptotic clearance came initially from the worm, which does not have professional phagocytes. Genetic screens identified several phagocytosis genes that fall into two partially redundant pathways: one consists of a phagocytic docking receptor (CED-1), its adapter (CED-6), an ABC transporter (CED-7), and dynamin; the other pathway consists of an actin-regulating protein complex (CED-2/5/10/12) that presumably acts downstream of an unknown phagocytic receptor. Recently, the phosphatidylserine receptor BAI1 was found to act upstream of the homologous complex in mouse. In Drosophila, only three factors have been implicated in apoptotic cell clearance to date: the macrophage-specific CD36 homolog Croquemort (CRQ), the F Box protein Pallbearer, and the broadly expressed CED-1 homolog DRPR, which plays a role in glial phagocytosis of apoptotic neurons. A recent study (Hamon, 2006) suggests that the apoptotic clearance function of CED-1/DRPR is conserved in vertebrates (Kurant, 2008).

Thus, most of the players involved in tissue-resident clearance are still unknown, in particular the phagocytic receptors and their cognate ligands on the apoptotic cell. More generally, a better understanding of the cellular and molecular underpinnings of apoptotic clearance in vivo is highly desirable. In the work presented in this study, it is demonstrated that glia are the main phagocytes in the late developing nervous system of the fly. A new tethering receptor named Six-microns-under (SIMU), is described that functions in apoptotic clearance by both glia and macrophages; it acts upstream of DRPR in the same pathway. This study reveals an evolutionary expansion and concomitant specialization of the repertoire of CED-1-like receptors, with SIMU required for the recognition and engulfment, and DRPR for the degradation of corpses (Kurant, 2008).

The simu gene is predicted to encode a 377 aa transmembrane protein, comprising an N-terminal signal peptide, a large extracellular portion, a single putative transmembrane (TM) domain, and a short cytoplasmic tail at the C terminus. The extracellular portion of the protein consists of an N-terminal EMILIN (EMI)-like domain (Callebaut, 2003) and four EGF domains with highly stereotyped spacing of Cys residues, recently named Nimrod (NIM) repeats (Kurucz, 2007). Intriguingly, the combination of an EMI(-like) domain, with a highly conserved CC×GY motif at its C-terminal end and immediately followed by a NIM repeat, is found across the vertebrate/invertebrate divide in a small number of both secreted and transmembrane factors, some of which have been shown to function in phagocytosis. They include the CED-1 homologs in all species (DRPR, human MEGF10, mouse Jedi), the fly bacterial phagocytosis factor Eater (Kocks, 2005), and the large Nimrod gene cluster located at 34E (Kurucz, 2007), the most proximal member of which is simu (= NimC4), as well as human SREC (Kurant, 2008).

This study has demonstrated that tissue-resident clearance by neighboring glia plays a major role in the removal of apoptotic neurons during nervous system development and has characterized the function of the new tethering receptor SIMU in this important process. Using several different markers, it was shown that, once the nerve cord is ensheathed and macrophages no longer have access, astrocytic glia avidly phagocytose apoptotic cells in their vicinity. This pronounced phagocytic activity is reflected on a molecular level by a strong differential expression of phagocytosis genes in the glia. In vivo time-lapse recordings indicate that the stationary glia are as efficient and fast in corpse uptake as the highly motile macrophages, and differences in clearance are largely a reflection of differential motility. Notably, glia also phagocytose axon branches during neuronal remodeling and injury-induced Wallerian degeneration. Collectively, these findings indicate that glia are potent 'semiprofessional' phagocytes and that, at least during development, tissue-resident apoptotic clearance equals macrophage-mediated clearance in importance. In vertebrates, tissue-resident phagocytosis by neighbors has been demonstrated in vivo in some contexts including glia, but its overall contribution to apoptotic clearance and the underlying molecular mechanisms have not been established (Kurant, 2008).

The investigation of the gene simu was originally motivated by two intriguing features: its expression during all major epochs of developmental apoptosis and in the major phagocytic cell types -- in particular the glia -- and a protein domain structure that is similar to but distinct from that of the CED-1 homolog DRPR. The two proteins are expressed in strongly overlapping patterns throughout embryogenesis and largely colocalize at the plasma membrane of glia and macrophages. Genetic analysis now reveals that they are components of the same phagocytic pathway, with simu acting upstream of drpr. Observations both in fixed material and in time-lapse recordings show simu affecting the early steps of recognition and engulfment of apoptotic particles, while drpr affects their subsequent degradation; the double-mutant phenotype closely resembles that of simu alone, indicating epistasy of simu over drpr. Unlike DRPR, SIMU lacks a large cytoplasmic domain with docking sites for downstream effectors, and the results suggest that its molecular function is to recognize the apoptotic cell and tether it to the phagocyte. Purified SIMU protein strongly binds to apoptotic cells in vitro; the in vivo analysis indicates that the N-terminal EMI-like domain, which is essential for SIMU function, is likely involved in the recognition and binding. To date, EMI domains have only been implicated in protein-protein interaction, suggesting that SIMU may recognize a protein rather than a lipid or carbohydrate component on the apoptotic cell. The lack of a cytoplasmic domain and the dispensability of membrane anchoring for in vivo function imply that SIMU must interact, directly or indirectly, with docking receptors on the phagocyte surface to effect engulfment and subsequent degradation. Given its phenotype and action downstream of simu, drpr is the prime candidate for mediating degradation. However, the difference in the simu and drpr phenotypes (in particular the essentially unimpaired engulfment of apoptotic particles observed in drpr mutants) strongly argues for the existence of an intermediary factor that controls the cytoskeletal reorganization necessary for engulfment downstream of simu. The existence of a SIMU binding partner other than DRPR is supported by the finding that, in drpr null mutants, secreted SIMUDeltaTM still accumulates on macrophage cell surfaces, and by the fact that no physical interaction was observed between SIMU and DRPR in coimmunoprecipitation experiments. Thus, the following model seems most likely: As part of a linear pathway, SIMU interacts with a factor X, leading to the recruitment of cytoskeletal components for engulfment and phagosome formation, followed by X interacting with DRPR, which leads to the recruitment of degradation components for phagosome maturation. The identification of factor X and its extra- and intracellular binding partners will be the subject of future investigation, and understanding the mechanisms by which the different phagocytic receptors cluster and coordinate their function will be of particular interest (Kurant, 2008).

The existence of the SIMU protein was unexpected both genomically and genetically: the C. elegans genome does not contain a SIMU homolog and in fact no other EMI(-like)+NIM domain protein. Moreover, in contrast to drpr, CED-1 is crucial for recognition, engulfment, and the degradation of corpses in the worm, suggesting that biological functions that are performed by a single factor in the worm become distributed among more specialized proteins in the fly. Intriguingly, when the worm CED-1 mutant is rescued by a transgene that lacks the intracellular docking domain, recognition and engulfment, but not degradation, are restored, resulting in a drpr-like phenotype. Several recent papers have suggested that DRPR acts as an engulfment receptor in different developmental contexts in the fly. Most observed an increase in the number of apoptotic cells or axon fragments in drpr mutants, which is consistent with the current findings. Unlike the current analysis, however, none of these studies established whether the excess apoptotic particles accumulate outside or inside the respective phagocytes. It is also noteworthy that apoptotic clearance is much faster (minutes versus hours or days) than axon pruning/Wallerian degeneration and possibly involves distinct signals from the degrading cells (caspase versus non-caspase-mediated mechanism), making it likely that differences exist at the level of the recognition factors (Kurant, 2008).

The comparison of apoptotic clearance in worm, fly, and vertebrate provides interesting evolutionary perspectives. Between worm and fly, the demand for apoptotic clearance increases both in quantitative and qualitative terms: During worm development, only 10% of cells are fated to die, and they are removed shortly after their birth through phagocytosis by immediate neighbors, apparently without need for dedicated phagocytes. During fly development, a much larger proportion of cells die through apoptosis (>30%), and their death is spread out over a wide range of stages in their life cycle: immediately after their birth, during fating and differentiation, and even after long-term function. To meet this challenge, macrophages emerge as professional phagocytes, and in addition, several cell types differentially increase their phagocytic capacity and become semiprofessional phagocytes, including the ectoderm and the glia. In vertebrates, the apoptotic load is presumably even greater, and the cellular complexity of the innate immune system is of course greatly increased. Strikingly, these changes are accompanied by an increase in complexity at the molecular level, both in terms of a greater structural diversity of phagocytic receptors generally and of CED-1-like factors in particular. While C. elegans seems to have only one CED-1 gene encoding a single protein form, Drosophila has two different DRPR isoforms with distinct extracellular domains and the entire Nimrod gene family, including SIMU and Eater (Kurucz, 2007). This family contains both secreted and transmembrane proteins, and receptors for the clearance of apoptotic cells (SIMU) and pathogens (Eater, NIM C1), but the function of most family members is not known. The in vivo function of the vertebrate CED-1-like factors is also unknown, but cell culture experiments suggest that MEGF10 is a phagocytic receptor for apoptotic cells and SREC for (oxidized) LDL. The molecular comparisons indicate that these phagocytic factors all share an N-terminal EMI(-like)+NIM core, presumably for recognition, and that invertebrates and vertebrates amplified different EGF-type repeats present in the ancestral CED-1, either to achieve appropriate spacing or to facilitate homotypic cis interaction in factor clustering (Kurant, 2008).

This molecular expansion is likely accompanied by an increase in functional redundancy between phagocytic pathways: already in the worm, at least two (CED-1/6/7 and CED-2/5/10/12) and likely a third (currently unknown) pathway participate in the phagocytosis process. As shown in this study for the fly, joint removal of simu and drpr results in viable adults whose apoptotic clearance is impaired but not abrogated, indicating the existence of additional as-yet-unidentified pathways. Finally, in vertebrates, a large number of different types of macrophage receptors have been identified, which under single-mutant conditions either only reduce apoptotic clearance or have no effect at all (Kurant, 2008).

Overall, this study has shown that the differences between professional and nonprofessional phagocytes are perhaps more fluid than originally thought. Not only can tissue-resident phagocytic cells differentially upregulate a similar molecular repertoire of phagocytic pathways, but they also can use it as quickly and efficiently as the macrophages when confronted with apoptotic neighbors. This suggests a simple division of labor -- in freely accessible spaces, actively patrolling macrophages clear the corpses, while in sequestered spaces (such as the late nervous system) the task falls to resident neighbors like the glia. In addition to its role in development, the efficient clearance of apoptotic cells is essential for avoiding secondary necrosis and exposure of cytoplasmic components to the immune system, acutely causing inflammation and, chronically, autoimmune disease. While apoptotic clearance by tissue-resident cells has received relatively little attention in vertebrate research, its comprehensive analysis in genetic models such as the fly will continue to provide new and interesting insight into the underlying cellular and molecular mechanisms (Kurant, 2008).

Glia and muscle sculpt neuromuscular arbors by engulfing destabilized synaptic boutons and shed presynaptic debris

Synapse remodeling is an extremely dynamic process, often regulated by neural activity. During activity-dependent synaptic growth at the Drosophila NMJ many immature synaptic boutons fail to form stable postsynaptic contacts, are selectively shed from the parent arbor, and degenerate or disappear from the neuromuscular junction (NMJ). Surprisingly, the widespread appearance of presynaptically derived 'debris' was observed during normal synaptic growth. The shedding of both immature boutons and presynaptic debris is enhanced by high-frequency stimulation of motorneurons, indicating that their formation is modulated by neural activity. Interestingly, it was found that glia dynamically invade the NMJ and, working together with muscle cells, phagocytose shed presynaptic material. Suppressing engulfment activity in glia or muscle by disrupting the Draper/Ced-6 pathway results in a dramatic accumulation of presynaptic debris, and synaptic growth in turn is severely compromised. Thus actively growing NMJ arbors appear to constitutively generate an excessive number of immature boutons, eliminate those that are not stabilized through a shedding process, and normal synaptic expansion requires the continuous clearance of this material by both glia and muscle cells (Fuentes-Medel, 2009. Full text of article).

Previous studies have shown that the PTB-domain protein dCed-6 functions downstream of Draper. Therefore, RNAi knockdown of dCed-6 in muscle or glia was used as a second approach to blocking glial and muscle engulfment activity. As in draper mutants, downregulating dCed-6 in either muscle or peripheral glia resulted in significant decrease in the number of synaptic boutons. In contrast, no effect was observed when dCed-6-RNAi was expressed in motorneurons. Similar to Draper RNAi knockdown, expressing dCed-6-RNAi in muscles or glia had differential consequences for the appearance of presynaptic debris versus ghost boutons. Decreased levels of dCed-6 in muscles led to an increase in the number of ghost boutons, but had no influence in the deposition of presynaptic debris. Downregulating dCed-6 in glia, on the other hand, led to a significant increase in presynaptic debris deposition, but the number of ghost boutons remained unaltered. These results are consistent with the notion that dCed-6 functions downstream of Draper during the development of the NMJ. Further, they support the model that both muscle and glia contribute differentially to the clearance of debris versus ghost boutons at the NMJ (Fuentes-Medel, 2009).

Interestingly, it was found that in draper mutants both disconnected ghost boutons and presynaptic debris accumulated, and this accumulation had a negative effect on NMJ expansion and bouton morphology. Moreover, synaptic growth appeared to be highly sensitive to both types of shed presynaptic material since the accumulation of either ghost boutons or presynaptic debris (when engulfment activity was blocked in muscles or glia, respectively) led to reductions in bouton growth similar to that seen in draper null mutants. As mentioned above, shed material might contain important signaling factors that potently stimulate or inhibit new synapse formation. If, for example, presynaptic debris contains molecules that inhibit synaptogenesis, the accumulation of such material would be expected to negatively regulate synaptic growth. Perhaps a similar type of inappropriate modulation of synaptogenesis by the membrane fragments of pruned terminals also accounts for their rapid clearance from the central nervous system after degeneration (Fuentes-Medel, 2009).

Drosophila glial cells also engulf neuronal cell corpses and pruned or degenerating axons. Each of these targets is generated by a unique degenerative molecular cascade: cell corpses are produced by canonical apoptotic cell death pathways, pruned axons undergo degeneration through a ubiquitin proteasome-dependent mechanism, and severed axons undergo Wallerian degeneration via Wld^s-modulated mechanisms (Wld^sis a dominant mutation resulting in the overexpression of a fusion protein of the first 70 amino acids of UFD2/E4, an evolutionarily conserved protein used in protein polyubiquitination, and the full-length nicotinamide mononucleotide adenylyltransferase, an enzyme that facilitates NAD synthesis). Both corpses and degenerating axons are engulfed by glia through Draper-dependent mechanisms, implying that these engulfment targets autonomously tag themselves with molecularly similar 'eat me' cues. The observations that mutations in draper led to accumulation of presynaptic debris and detached ghost boutons suggests that these new glial/muscle engulfment targets also produce similar cues for phagocytic cells to promote their destruction. If so, these data argue that all the necessary machinery essential for tagging membrane fragments for engulfment are present in a ghost bouton or fragment of presynaptic membrane. Importantly, while a lack of glial-mediated clearance of several targets has been observed in vivo -- cell corpses, pruned axons or dendrites, and axons undergoing Wallerian degeneration -- almost nothing is known about phenotypic consequences of a lack of glial engulfment function in the nervous system. This study demonstrates that failure of glia and muscle to clear presynaptically derived material negatively regulates synaptic growth (Fuentes-Medel, 2009).

In conclusion these studies demonstrate that the process of synaptic growth includes a significant degree of membrane/synaptic instability, and that growing terminals are constantly sloughing off undifferentiated boutons and fragments of membrane. These observations demonstrate that growing NMJs generate an excess number of undifferentiated synaptic boutons and that only a fraction becomes stabilized and drive the assembly of the postsynaptic apparatus. Exuberant synapses that have failed to form successful postsynaptic contacts are shed, and cleared from the NMJ by glia and muscle cells. The presence of such a pool ensures a continuous supply of nascent synapses available for use to rapidly increase input into the muscle if dictated by dynamic changes in signaling at the NMJ (Fuentes-Medel, 2009).

Reference names in red indicate recommended papers.

Araki, T., Sasaki, Y. and Milbrandt, J. (2004). Increased nuclear NAD biosynthesis and SIRT1 activation prevent axonal degeneration. Science 305(5686): 1010-3. 15310905

Awasaki, T., and Ito, K. (2004). Engulfing action of glial cells is required for programmed axon pruning during Drosophila metamorphosis. Curr. Biol. 14: 668-677. 15084281

Awasaki, T., Tatsumi, R., Takahashi, K., Arai, K., Nakanishi, Y., Ueda, R. and Ito, K. (2006). Essential role of the apoptotic cell engulfment genes draper and ced-6 in programmed axon pruning during Drosophila metamorphosis. Neuron 50(6): 855-67. 16772168

Bishop, D. L., Misgeld, T., Walsh, M. K., Gan, W. B., and Lichtman, J. W. (2004). Axon branch removal at developing synapses by axosome shedding. Neuron 44: 651-661. 15541313

Callebaut, I., Mignotte, V., Souchet, M. and Mornon, J.P. (2000). EMI domains are widespread and reveal the probable orthologs of the Caenorhabditis elegans CED-1 protein. Biochem. Biophys. Res. Commun. 300: 619-623. PubMed Citation: 12507493

Carninci, P., et al. (1996). High-efficiency full-length cDNA cloning by biotinylated CAP trapper. Genomics 37: 327-336. 8938445

Conforti, L., et al. (2000). A Ufd2/D4Cole1e chimeric protein and overexpression of Rbp7 in the slow Wallerian degeneration (WldS) mouse, Proc. Natl. Acad. Sci. 97: 11377-11382. 11027338

Cuttell, L., et al. (2008). Undertaker, a Drosophila Junctophilin, links Draper-mediated phagocytosis and calcium homeostasis. Cell 135(3): 524-34. PubMed Citation: 18984163

Doherty, J., Logan, M. A., Tasdemir, O. E. and Freeman, M. R. (2009). Ensheathing glia function as phagocytes in the adult Drosophila brain. J. Neurosci. 29(15): 4768-81. PubMed Citation: 19369546

Ellis, R. E., Jacobson, D. M. and Horvitz, H. R. (1991). Genes required for the engulfment of cell corpses during programmed cell death in Caenorhabditis elegans. Genetics 129(1): 79-94. 1936965

Ferrante, A. W., Reinke, R. and Stanley, E. R. (1995). Shark, a Src homology 2, ankyrin repeat, tyrosine kinase, is expressed on the apical surfaces of ectodermal epithelia. Proc. Natl Acad. Sci. 92: 1911-1915. PubMed citation: 7892198

Fernandez, R. et al. (2000). The Drosophila shark tyrosine kinase is required for embryonic dorsal closure. Genes Dev. 14: 604-614. PubMed citation: 10716948

Franz, S., et al. (2007). After shrinkage apoptotic cells expose internal membrane-derived epitopes on their plasma membranes. Cell Death Differ. 14: 733-742. PubMed Citation: 17170754

Freeman, M. R., Delrow, J., Kim, J., Johnson, E. and Doe, C. Q. (2003). Unwrapping glial biology: Gcm target genes regulating glial development, diversification, and function. Neuron 38(4): 567-80. 12765609

Glass, J. D.. Brushart, T. M., George, E. B. and Griffin, J. W. (1993). Prolonged survival of transected nerve fibres in C57BL/Ola mice is an intrinsic characteristic of the axon, J. Neurocytol. 22: 311-321. 8315413

Fuentes-Medel, Y., Logan, M. A., Ashley, J., Ataman, B., Budnik, V. and Freeman, M. R. (2009). Glia and muscle sculpt neuromuscular arbors by engulfing destabilized synaptic boutons and shed presynaptic debris. PLoS Biol. 7(8): e1000184. PubMed Citation: 19707574

Hamon, Y., et al. (2006). Cooperation between engulfment receptors: the Case of ABCA1 and MEGF10. PLoS ONE 1: e120. PubMed Citation: 17205124

Hashimoto, Y., et al. (2009). Identification of lipoteichoic acid as a ligand for Draper in the phagocytosis of Staphylococcusaureus by Drosophila hemocytes. J. Immunol. 183(11): 7451-60. PubMed Citation: 19890048

Hoopfer, E. D., McLaughlin, T., Watts, R. J., Schuldiner, O., O'Leary, D. D. and Luo, L. (2006). Wld^s protection distinguishes axon degeneration following injury from naturally occurring developmental pruning. Neuron 50(6): 883-95. 16772170

Ito, K., Urban, J. and Technau, G. M. (1995). Distribution, classification and development of Drosophila glial cells in the late embryonic and early larval ventral nerve cord. Roux's Arch. Dev. Biol. 204: 284-307

Kinchen, J. M., Cabello, J., Klingele, D., Wong, K., Feichtinger, R., Schnabel, H., Schnabel, R. and Hengartner, M. O. (2005). Two pathways converge at CED-10 to mediate actin rearrangement and corpse removal in C. elegans. Nature 434: 93-99. 15744306

Kinchen, J. M., et al. (2005). Two pathways converge at CED-10 to mediate actin rearrangement and corpse removal in C. elegans. Nature 434: 93-99. PubMed Citation: 15744306

Kinchen, J. M. and Ravichandran, K. S. (2007). Journey to the grave: signaling events regulating removal of apoptotic cells. J. Cell Sci. 120: 2143-2149. PubMed Citation: 17591687

Kocks, C., et al. (2005). Eater, a transmembrane protein mediating phagocytosis of bacterial pathogens in Drosophila. Cell 123: 335-346. PubMed Citation: 16239149

Kuraishi, T., et al. (2009). Pretaporter, a Drosophila protein serving as a ligand for Draper in the phagocytosis of apoptotic cells. EMBO J. 28(24): 3868-78. PubMed Citation: 19927123

Kurant, E., Axelrod, S., Leaman, D. and Gaul, U. (2008). Six-microns-under acts upstream of Draper in the glial phagocytosis of apoptotic neurons. Cell 133(3): 498-509. PubMed Citation: 18455990

Kurucz, E. et al. (2007). Nimrod, a putative phagocytosis receptor with EGF repeats in Drosophila plasmatocytes. Curr. Biol. 17: 649-654. PubMed Citation: 17363253

Lettre, G. and Hengartner, M. O. (2006). Developmental apoptosis in C. elegans: a complex CEDnario. Nat. Rev. Mol. Cell Biol. 7: 97-108. PubMed Citation: 16493416

Liu, Q. A. and Hengartner, M.O. (1998). Candidate adaptor protein CED-6 promotes the engulfment of apoptotic cells in C. elegans. Cell 93: 961-972. 9635426

Lunn, E. R., et al. (1989). Absence of Wallerian degeneration does not hinder regeneration in peripheral nerve. Eur. J. Neurosci. 1: 27-33. 12106171

Luo, L. and O'Leary, D. D. M. (2005). Axon retraction and degeneration in development and disease, Annu. Rev. Neurosci. 28: 127-156. 16022592

MacDonald, J. M., Beach, M. G., Porpiglia, E., Sheehan, A. E., Watts, R. J. and Freeman, M. R. (2006). The Drosophila cell corpse engulfment receptor Draper mediates glial clearance of severed axons. Neuron 50(6): 869-81. 16772169

Mack, T. G., et al. (2001). Wallerian degeneration of injured axons and synapses is delayed by a Ube4b/Nmnat chimeric gene, Nat. Neurosci. 4: 1199-1206. 11770485

Manaka, J., et al. (2004). Draper-mediated and phosphatidylserine-independent phagocytosis of apoptotic cells by Drosophila hemocytes/macrophages. J. Biol. Chem. 279(46): 48466-76. 15342648

Ogden, C. A., deCathelineau, A., Hoffmann, P. R., Bratton, D., Ghebrehiwet, B., Fadok, V. A., and Henson, P. M. (2001). C1q and mannose binding lectin engagement of cell surface calreticulin and CD91 initiates macropinocytosis and uptake of apoptotic cells. J. Exp. Med. 194: 781-795. 11560994

Panaretakis, T., et al. (2009). Mechanisms of pre-apoptotic calreticulin exposure in immunogenic cell death. EMBO J. 28: 578-590. PubMed Citation: 19165151

Stuart, L. M., et al. (2005). Response to Staphylococcus aureus requires CD36-mediated phagocytosis triggered by the COOH-terminal cytoplasmic domain, J. Cell Biol. 170: 477-485. PubMed Citation: 16061696

Su, H. P., Nakada-Tsukui, K., Tosello-Trampont, A. C., Li, Y., Bu, G., Henson, P. M., and Ravichandran, K. S. (2002). Interaction of CED- 6/GULP, an adapter protein involved in engulfment of apoptotic cells with CED-1 and CD91/low density lipoprotein receptor-related protein (LRP). J. Biol. Chem. 277: 11772-11779. 11729193

Takeshima, H., et al. (2000). Junctophilins: a novel family of junctional membrane complex proteins. Mol. Cell 6: 11-22. PubMed Citation: 10949023

Venegas, V. and Zhou, Z. (2007). Two alternative mechanisms that regulate the presentation of apoptotic cell engulfment signal in Caenorhabditis elegans. Mol. Biol. Cell 18: 3180-3192. PubMed Citation: 17567952

Vig, M., et al. (2006). CRACM1 is a plasma membrane protein essential for store-operated Ca2+ entry. Science 312: 1220-1223. PubMed Citation: 16645049

Wang, X., et al. (2003). Cell corpse engulfment mediated by C. elegans phosphatidylserine receptor through CED-5 and CED-12. Science 302(5650): 1563-6. 14645848

Wang, J., et al. (2005). A local mechanism mediates NAD-dependent protection of axon degeneration. J. Cell Biol. 170(3): 349-55. 16043516

Watts, R. J., Hoopfer, E. D. and Luo, L. (2003). Axon pruning during Drosophila metamorphosis: evidence for local degeneration and requirement of the ubiquitin-proteasome system, Neuron 38: 871-885. 12818174

Watts, R. J., Schuldiner, O., Perrino, J., Larsen, C., and Luo, L. (2004). Glia engulf degenerating axons during developmental axon pruning. Curr. Biol. 14: 678-684. 15084282

Wu, Y. C., and Horvitz, H. R. (1998a). The C. elegans cell corpse engulfment gene ced-7 encodes a protein similar to ABC transporters. Cell 93: 951-960. 9635425

Wu, Y. C., and Horvitz, H. R. (1998b). C. elegans phagocytosis and cell-migration protein CED-5 is similar to human DOCK180. Nature 392: 501-504. 9548255

Yu, X., Odera, S., Chuang, C. H., Lu, N. and Zhou, Z. (2006). C. elegans Dynamin mediates the signaling of phagocytic receptor CED-1 for the engulfment and degradation of apoptotic cells. Dev. Cell 10(6): 743-57. 16740477

Zhou, Z., Hartwieg, E. and Horvitz, H. R. (2001). CED-1 is a transmembrane receptor that mediates cell corpse engulfment in C. elegans. Cell 104: 43-56. 11163239

Ziegenfuss, J. S., et al. (2008). Draper-dependent glial phagocytic activity is mediated by Src and Syk family kinase signalling. Nature 453: 935-939. PubMed citation: 18432193

date revised: 10 August 2010

Home page: The Interactive Fly © 2006 Thomas Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.