Axon pruning is involved in establishment and maintenance of functional neural circuits. During metamorphosis of Drosophila, selective pruning of larval axons is developmentally regulated by ecdysone and caused by local axon degeneration. Previous studies have revealed intrinsic molecular and cellular mechanisms that trigger this pruning process, but how pruning is accomplished remains essentially unknown. Detailed analysis of morphological changes in the axon branches of Drosophila mushroom body (MB) neurons revealed that during early pupal stages, clusters of neighboring varicosities, each of which belongs to different axons, disappear simultaneously shortly before the onset of local axon degeneration. At this stage, bundles of axon branches are infiltrated by the processes of surrounding glia. These processes engulf clusters of varicosities and accumulate intracellular degradative compartments. Selective inhibition of cellular functions, including endocytosis, in glial cells via the temperature-sensitive allele of shibire both suppresses glial infiltration and varicosity elimination and induces a severe delay in axon pruning. Selective inhibition of ecdysone receptors in the MB neurons severely suppresses not only axon pruning but also the infiltration and engulfing action of the surrounding glia. These findings strongly suggest that glial cells are extrinsically activated by ecdysone-stimulated MB neurons. These glial cells infiltrate the mass of axon branches to eliminate varicosities and break down axon branches actively rather than just scavenging already-degraded debris. It is therefore proposed that neuron-glia interaction is essential for the precisely coordinated axon-pruning process during Drosophila metamorphosis (Awasaki, 2004).

The dorsal and medial axon branches of the larval γ neurons have many varicosities along its trajectory that are clearly visualized when a single γ neuron is labeled with cytoplasmic-GFP (cGFP) reporter combined with the mosaic analysis with a repressible cell marker (MARCM) system and the enhancer-trap strain GAL4-201Y. These axon branches are also labeled strongly with the anti-Synaptotagmin antibody, which detects synaptic vesicle protein accumulated in the presynaptic sites. Thus, the larval γ neurons seem to be extensively interconnected with other neurons through synapses. Because synaptic terminals are condensed in the varicosities, the fate of these varicosity structures were first examined during the time course of axon pruning (Awasaki, 2004).

The number of varicosities on a dorsal axon branch was counted at various developmental stages. A single branch contained 20.7 and 21.4 varicosities on average in the brains of late third-instar larvae (L3) and in pupae just after puparium formation (0 hr APF), respectively. The number of varicosities was reduced by approximately 35% at 6 hr APF (an average of 13.6 varicosities per axon branch). To compare the timing of local axon degeneration and reduction of varicosity numbers, the frequency of disconnected axons was simultaneously examined. Only a very few axon branches were disconnected at 6 hr APF (2 of 60 observed axons) (Awasaki, 2004).

At 12 hr APF, approximately 90% of varicosities disappeared (an average of 2.0 varicosities per branch) and 75% of axons were disconnected (30 of 40). At 18 hr APF, more than 95% of varicosities disappeared (an average of 0.9 varicosities per branch) and 90% of axons were disconnected (48 of 53). In many cases, distal portions of disconnected axons remained at 12 hr APF, and only distal tips of axons frequently remained at 18 hr APF, suggesting that the shaft region of the axon branch tends to degenerate first. At 12 hr and 18 hr APF, the number of varicosities was drastically reduced even on the axons that were not yet disconnected (an average 2.8 [n = 10] and 1.4 [n = 5] varicosities per branch, respectively). These data suggest that disappearance of varicosities precedes local axon degeneration (Awasaki, 2004).

Pruning of axons is triggered by ecdysone through the complex of ecdysone receptor type B1 (EcR-B1) and Ultraspiracle (USP). If the disappearance of varicosities is associated with programmed axon pruning, the disappearance should be suppressed by inhibition of ecdysone signaling. To address this, axon branches of the clones of usp mutant γ neurons were examined. Both disappearance of varicosities and axon disconnection were significantly suppressed. Thus, these phenomena were regulated cell autonomously by the ecdysone-induced stimulation of γ neurons (Awasaki, 2004).

There are more than 2000 neurons that form the larval MB, and these results have indicated that disappearance of the varicosities proceeds gradually during the first 18 hr of the pupal stage. How do the varicosities disappear in these masses of axon branches? Do they disappear randomly, one by one, or in a coordinated process? To examine this, the detailed structure of the dorsal MB lobe was observed by using the enhancer-trap strain GAL4-201Y and the anti-Fasciclin II (FasII) antibody, both of which label axons of γ neurons from the larva to adult stage (Awasaki, 2004).

In the dorsal lobe of late larvae (L3), a mass of γ neuron axons was labeled uniformly with cGFP driven by GAL4-201Y and with anti-FasII antibody. At 6 hr APF, however, many hole-like structures of unlabeled regions appeared in cross-sections. Considering that the number of varicosities was significantly reduced at this stage whereas most axons had not been disconnected, it is highly likely that the hole-like structures correspond to the vestiges of disappeared varicosities (Awasaki, 2004).

To confirm this, lobes labeled with the targeted expression of the presynapse marker n-Synaptobrevin::GFP (nSyb::GFP) and the membrane bound reporter mCD8::GFP (mGFP) were compared. Like cGFP, nSyb::GFP strongly labels varicosities along the axon branches. Images of the sections at 6 hr APF clearly reveal the hole-like structures. In contrast, mGFP labels voluminous varicosities much less intensively than cGFP, because these subcellular regions contain less membrane surface per volume. mGFP labels the hole-like structures much less clearly than does cGFP. This further supports the hypothesis that the unlabeled regions are associated with the disappearance of varicosities (Awasaki, 2004).

Each hole-like structure was considerably larger than the size of a single varicosity. The diameter of many holes was more than 5 μm, whereas the size of the largest varicosity was 1.8 μm. This indicates that each unlabeled hole corresponds to the vestige of multiple varicosities. Because the varicosities of a single neuron were scattered along its axon branch, it is unlikely that a single hole corresponds to the varicosities that belong to a single neuron. Rather, neighboring varicosities from different neurons must simultaneously disappear in a cluster (Awasaki, 2004).

The larval neuropil of the central nervous system of Drosophila is covered by sheaths of neuropil-associating glial cells. Three-dimensional reconstruction of the confocal serial sections revealed a smooth surface of the MB dorsal lobe in the larval brain. At 6 hr APF, however, the surface appeared eroded from the outside. This suggests that glial cells surrounding the MB lobe might erode this neuropil structure. The morphologic changes of glial cells were therefore traced by using the pan-glial driver GAL4-repo and the cGFP reporter (repo>cGFP). In larvae, glial cells surrounded the lobe, but the axon branches inside the lobe were devoid of glial processes. At 6 hr APF, glial processes had infiltrated into the lobes and formed many lumps. The positions of these lumps correspond to the hole-like structures that were unlabeled with anti-FasII antibody (Awasaki, 2004).

Close observation of the glial processes revealed various patterns of infiltration at 6 hr APF. In some sections, thin glial processes surround parts of axon branches labeled with anti-FasII antibody. Glial processes also spread into the surrounded mass, forming glial lumps. Vesicle-like structures in these glial lumps contained small dots that were FasII-positive. In the specimens with massive glial infiltration, glial processes formed large lumps that occupied whole regions unlabeled with anti-FasII antibody (Awasaki, 2004).

Considering that dorsal lobes are initially filled with FasII-positive axon mass and hole-like structures unlabeled with anti-FasII antibody are eventually filled with glial lumps labeled with repo>cGFP, images like the ones described are best interpreted as being intermediate steps. Hypothetically it would seem that at first, infiltrating glial processes surround clusters of varicosities, which are still labeled with anti-FasII antibody. Next, glial processes further penetrate between varicosities to form lumps. FasII-positive signals observed in the glial lumps might be parts of varicosities engulfed by glia. Intensity of the FasII signal within glia is remarkably reduced compared to the signals observed outside of glia. This might be because engulfed FasII protein is subject to intracellular degradation. Finally, the FasII signal totally disappears, and glial lumps replace the vestiges of varicosities (Awasaki, 2004).

If this hypothesis has validity, infiltrated parts of glia must exhibit evidence of high activity in the intracellular degradation process. To examine whether this is the case, vital staining was performed with LysoTracker, which labels acidic organelles such as late endosomes and lysosomes. In the larval lobes, no detectable signal was found with the LysoTracker staining. In the dorsal lobe at 6 hr APF, however, the LysoTracker-positive signals were constantly found. In most cases, the signals were localized in the lumps of the infiltrated glia. This finding strongly suggests that varicosities are engulfed by glia and degraded in the endosome-lysosome pathway (Awasaki, 2004).

Though most of the varicosities disappear by 12 hr APF, glial infiltration was still observed at 12 hr and 18 hr APF. Glial processes at these stages frequently form columnar structures lying along the axon branches. These columnar structures are often observed in the shaft region of the dorsal lobes. The disconnection of axons also occurs frequently in the shaft region rather than in the tip region at 12 hr APF. Thus, the formation of the columnar structures was spatially and temporally correlated with the occurrence of axon degeneration. This suggests that infiltrated glia are involved in the disconnection and degeneration of axon branches as well as the disappearance of varicosities (Awasaki, 2004).

These results so far suggested that varicosities and axon branches are engulfed and degraded in the glial lumps. There are two possibilities for the role of glia during this process: (1) axon branches of γ neurons might be degenerated by intrinsic mechanisms and infiltrating glia might just scavenge the already degraded debris; (2) degradation of the axon branches of the γ neurons may not be due to intrinsic mechanisms alone and may require infiltrating glia to actively break down the axon branches. To determine the most likely possibility, the pruning process was observed while glial cell function was conditionally inhibited by the targeted expression of the temperature-sensitive allele of shibire (shi^ts1). The shibire gene is the Drosophila homolog of dynamin, which is involved in cellular membrane functions, including endocytosis, phagocytosis, and membrane growth. Because Shi^ts1 protein acts as a dominant-negative at a restrictive temperature, targeted expression of Shi^ts1 in glial cells would inhibit endocytosis and other membrane-related functions (Awasaki, 2004).

Larvae expressing Shi^ts1 in glial cells by using the GAL4-repo driver (repo>shi^ts1) were raised at a permissive temperature (19°C) and then placed in the restrictive temperature (29°C) from 0 hr APF to either 6 hr or 18 hr APF. Raising repo>shi^ts1 pupae at the restrictive temperature results in an abnormal morphology of the glial cells. The most striking abnormality of these glia is that they did not extend their processes into the lobe at all. Whereas glial processes in a normal condition have infiltrated into lobes intensely already at 6 hr APF, dorsal lobes of repo>shi^ts1 in the restrictive temperature were devoid of any glial processes even at 18 hr APF. This demonstrates that Shi^ts1 almost completely suppresses the infiltration of glial processes (Awasaki, 2004).

Anti-FasII labeling of dorsal lobes revealed hole-like structures of unlabeled regions at 6 hr APF, and severe loss of axon branches at 18 hr APF. Neither holes of unlabeled regions nor loss of axon branches were observed in the lobes of the repo>shi^ts1 animal in the restrictive temperature at 6 hr and 18 hr APF. The phenotypes were qualitatively indistinguishable among different individuals. These indicate that axon pruning including both disappearance of clustered varicosities and local axon degeneration were severely suppressed in these pupae (Awasaki, 2004).

Because inhibition of glial infiltration caused severe suppression of axon pruning, it is highly likely that infiltrating glia actively break down and remove axon branches rather than just scavenging already degraded debris. This result, together with the engulfment and degradation of clustered varicosities by glial lumps, suggests that infiltrating glia actively eliminate clustered varicosities at the early phase of pruning (Awasaki, 2004).

If the pruning of γ neurons is suppressed by the targeted disruption of glial function, do the unpruned axon branches remain in the adult MB? The repo>shi^ts1 pupae incubated at the restrictive temperature did not survive until adult stage; therefore, glial function could not be disrupted throughout pupal development. When the temperature was shifted from the restrictive temperature to 25°C at 24 hr APF, however, a small number of survivors eclosed. In these animals (n = 10), abnormal axon branches expressing abnormal FasII-positive signals were observed shortly after hatching beside the α and β lobes, which are adult-specific branches labeled with anti-FasII antibody. Such abnormal FasII-positive branches were observed only in the distal tip of the lobes, because continuous fibers could not be traced from them to the shaft region of the lobe. Moreover, the abnormal branches were no longer visible in the brains more than 7 days after eclosion. It is therefore likely that these FasII-positive branches were the remnants of the γ neuron axons that were not pruned at 18 hr APF but gradually degenerated thereafter. Thus, transient inhibition of glial function in early pupae does not completely block axon pruning but severely delays the time course of axon pruning (Awasaki, 2004).

Axon pruning is triggered by a surge of ecdysone, which occurs just before puparium formation. The γ neurons express the ecdysone receptor EcR-B1 strongly in late third-instar larvae. The surrounding glial cells express EcR-B1 faintly at this stage. The expression of EcR-B1 in the glial cells, however, is unlikely to be associated with axon pruning. Although EcR-B mutation causes complete suppression of axon pruning, this suppression is rescued by expression of EcR-B1 only in the γ neurons. How, then, is the glial activation controlled? Because expression of EcR-B1 in γ neurons is sufficient for axon pruning, the glial cells must be activated via ecdysone stimulation of γ neurons. To test this hypothesis, expression of a dominant-negative form of the ecdysone receptor was induced in γ neurons by using the 201Y driver (201Y>EcR-DN), which drives expression in γ neurons, but not in the surrounding glia. If the activation of glia is dependent on the ecdysone-induced stimulation of γ neurons, glial infiltration would be suppressed in these pupae. Conversely, if the glial activation is independent of γ neurons, infiltration would not be disturbed. The results strongly suggest that ecdysone-induced stimulation of γ neurons is indispensable for the infiltrating and engulfing action of glial cells. Once γ neurons receive ecdysone by EcR-B1, these neurons would induce glial infiltration and engulfment. Thus, the neuron-glia interaction is essential for coordinated axon pruning (Awasaki, 2004).

The neuron-glia interactions for selective infiltration and engulfment resemble the interactions between phagocytes and apoptotic cells. It is not likely, however, that axon pruning and apoptosis use identical molecular mechanisms to activate glial cells and phagocytes. Mutation of Drosophila apoptosis activator genes (grim, hid, and rpr) and overexpression of apoptosis inhibitor genes (p35 and DIAP1) in γ neurons have no effect on axon pruning. However, Draper (Drpr), the homolog of the C. elegans cell corpse engulfment molecule CED-1, accumulates on the cell membrane of infiltrating glial processes. Engulfment of apoptotic neurons by glial cells is suppressed in the central nervous system (CNS) of drpr mutant embryos. This raises the possibility that sensing of apoptotic cells by phagocytes and sensing of unwanted axon branches by glial cells share a common molecular mechanism. Although the drpr mutant is isolated in Drosophila, it is embryonic lethal. Furthermore, clonal analysis with a flippase-mediated MARCM system does not work reliably in the larval and pupal glial cells. Analysis of the function of drpr in the engulfing glial cells with more sophisticated techniques to overcome the above problem, together with the analyses of the involvement of phosphatidylserine and lysophosphatidylcholine in the axon pruning of γ neurons, might reveal correlated mechanisms between removal of the cell corpse and unwanted axons (Awasaki, 2004).

Inhibition of glial function almost completely suppresses axon pruning until 18 hr APF, but the remaining axons still undergo degeneration in the adult. There is therefore a possibility that cell-autonomous mechanisms can prune axon branches without glial involvement. The time course of degeneration, however, is severely delayed (Awasaki, 2004).

Rapid degeneration of γ neuron axons seems vital for proper remodeling of the MB neural circuit. In the late larval brain, the α′/β′ neurons extend their axons into the core of the axon bundle of γ neurons. As γ neuron axons are pruned during the early pupal stage, α′/β′ neurons claim the region previously occupied by γ neurons. When pruning of γ neurons is suppressed by ectopic expression of yeast ubiquitin protease, UBP2, the unpruned axons remain in the vicinity of axon branches of α′/β′ neurons in the adult. The morphology of α′/β′ axon branches was distorted in this case. This suggests that proper elimination of larval axon branches might be important for the correct formation of adult axon branches. Rapid pruning with the help of the engulfing action of glia would be advantageous in such a situation (Awasaki, 2004).

Migration and proliferation have been mostly explored in culture systems or fixed preparations. A simple genetic model, the chains of glia moving along fly wing nerves, is presented to follow such dynamic processes by time-lapse in the whole animal. Glia undergo extensive cytoskeleton and mitotic apparatus rearrangements during division and migration. Single cell labelling identifies different glia: pioneers with high filopodial, exploratory, activity and, less active followers. In combination with time-lapse, altering this cellular environment by genetic means or cell ablation has allowed the role of specific cell-cell interactions to be defined -- (1) neuron-glia interactions are not necessary for glia motility but do affect the direction of migration; (2) repulsive interactions between glia control the extent of movement; (3) autonomous cues control proliferation (Aigouy, 2004).

The fly wing contains two purely sensory nerves that navigate proximally along the so-called L1 and L3 veins, and are lined by glial cells. L3 vein contains three gliogenic sensory organs, L3.3, L3.1 and L3.v, from distal to proximal. Each sensory organ precursor (SOP) divides asymmetrically several times and produces a glial precursor (GP), which repeatedly divides and is therefore called a first order glial precursor (GPI). The GPI, initially located adjacent to the neuronal soma, subsequently moves along the underlying axon. The final number of glial cells per lineage varies stochastically from four to eight (average=six). The whole glial lineage, from GPIs to post-mitotic glial cells, expresses the Repo glial marker (Aigouy, 2004).

On L3, proliferation starts several hours after GPI birth, at around 17 hours after puparium formation (APF), and by 20-22 hours APF all GPIs have divided once. The second round of division takes place by 24 hours APF, and the third by 30 hours APF, with little glial proliferation being observed afterwards. These data are in agreement with the GPI dividing symmetrically twice, to produce two second order and four third order precursors (GPII and GPIII, respectively. The presence of up to eight glial cells suggests that GPIIIs may or may not divide, and/or that cell death takes place throughout the lineage (Aigouy, 2004).

The observation that glia migrate and proliferate to variable extents suggests a role for extrinsic signals and makes it difficult to follow the development of glial lineages in fixed tissues. Therefore an in vivo glial marker was devised by crossing the repo-Gal4 driver with a UAS-GFP reporter (repo-GFP) (Aigouy, 2004).

When L1 glia are compared with fly border cells, functioning during oogenesis, and tracheal cells, differences and similarities become apparent. Border cells rely on an asymmetric structure, the actin based-LCE (long cellular extension), which triggers movement through a grapple and pull process. By contrast, wing pioneer glial cells send out numerous filopodia that dynamically assay the environment in all directions; this is also seen in tracheal cells. This may reflect the fact that border cells form a cluster and move simultaneously, whereas, in the chain of glia, distal cells do not contact pioneer cells and follow the movement of adjacent, more proximal, cells. Based on time-lapse and ablation data, the following model is proposed. Prior to migration, pioneer cells are free to explore the proximal axonal substrate whereas follower cells are submitted to bilateral repulsive glia-glia contacts. As pioneer cells move proximally, they free space at more distal positions. Follower cells rapidly occupy such space, thus freeing even more distal regions. This domino-type of migration proceeds until homogenous axon enwrapping is reached (Aigouy, 2004).

Most exploratory activity is spatially and temporally restricted as it specifically characterises cells that move on naked axons. Whether the presence of pioneer cells at the front of migration is lineage dependent or reliant on extracellular signals remains to be determined. In the future, performing ablations throughout the L1 will help to address the issue of plasticity. The presence of LCE versus pioneer filopodia may reflect different modalities of directional migration. It is known that border cells respond to chemoattractants, whereas glial directional migration may be driven by underlying axons. Interestingly, L3 migrating and proliferating cells all show filopodia, suggesting yet a different mode of migration compared with that of cell chains (L1 glia) and clusters (border cells). Finally, glia, but not tracheal or border cells, divide as they move, suggesting that the formation of a continuous chain along the axon bundle requires both migration and proliferation. In the future, it will be crucial to determine how the two events are coordinated (Aigouy, 2004).

The different features shown by border cells, trachea and glia suggest that cell specification controls motility strategies. The role of cell specification cues is further demonstrated by the fact that, even within glial cells, different lineages display distinct features. While wing GPs divide several times, glia arising from dorsal bipolar dendritic embryonic lineages and microchaete glia do not divide, the latter dying soon after birth. Understanding the molecular pathways specifying migratory and proliferative profiles represents one of the future challenges for developmental biologists (Aigouy, 2004).

Like oligodendrocyte precursors in the rat optic nerve that are controlled by an internal clock, GPIs divide a limited number of times. Different sets of data speak in favour of an internal clock that limits the absolute number of divisions to three. (1) The average number of glia derived by different sensory organs is constant, irrespective of the number of underlying axons (L3.3 and L3.1 glia line one and four axons, respectively). Thus, the number of axons does not control proliferation. (2) Time-lapse data show that divisions within and between lineages are rather synchronous. (3) Mitotic clones lacking the Gcm (Glial cell missing) glial promoting factor in one gliogenic sensory organ result in fewer wing glia, indicating that the remaining cells do not compensate for the missing ones. (4) Ablation data demonstrate a lack of compensatory divisions within and among glial lineages. Whether vertebrate and invertebrate clocks rely on the same signals will be a matter of future studies. In addition to the internal clock, extracellular signals may be at work and may control the fine-tuning of proliferation (four to eight cells per lineage). A role of cell-cell interactions in the control of glial proliferation has also been observed in the fly embryo (Aigouy, 2004).

Midline glia are a source of cues for neuronal navigation and differentiation in the Drosophila CNS. Despite their importance, how glia and neurons communicate during development is not fully understood. This study examined dynamic morphology of midline glia and assessed their direct cellular interactions with neurons within the embryonic CNS. Midline glia extend filopodia-like 'gliopodia' from the onset of axogenesis through the near completion of embryonic neural development. The most abundant and stable within the commissures, gliopodia frequently contact neurites extending from the neuropil on either side of the midline. Misexpression of Rac1^N17 in midline glia not only reduces the number of gliopodia but also shifts the position of neuropils towards the midline. Midline-secreted signaling protein Slit accumulates along the surface of gliopodia. Mutant analysis supports the idea that gliopodia contribute to its presentation on neuronal surfaces at both the commissures and neuropils. It is proposed that gliopodia extend the range of direct glia-neuron communication during CNS development (Vasenkova, 2005).

Axonal and dendritic growth cones are guided into commissures, longitudinal fascicles and neuropils within the embryonic CNS. Throughout this period, midline glia extend gliopodia not only into commissures near the midline but also at the medial region of longitudinal neuropils several cell diameters away from their edges. While navigating within the CNS, neuronal growth cones also extend filopodia that span several cell diameters. Put together, the distance of direct filopodial contact between midline glia and neurons expands dramatically. As a result, the embryonic CNS displays a 'filopodial web' in which a majority of neurons could contact midline glia directly. The main effect of gliopodial presence during CNS development could be to extend the reciprocal influence of specialized midline cells over, and/or sensitivity toward, outgrowing neurites that lie at a distance. Whether or not a similarly extensive 'filopodial web' exists in embryonic nervous systems of larger animals is yet to be determined. However, it is interesting to note that the floorplate cells at the midline in the vertebrates have been documented to extend 'side-arms' (Vasenkova, 2005).

Midline glia in Drosophila are generally thought to serve a role analogous to floorplate cells of the vertebrate CNS. Unlike other glia that migrate along the medial-lateral axis of the embryonic nervous system, midline glia remain at the midline of the CNS throughout development. Yet, the distance over which midline glia-derived factors exerts their influence reaches at least the neuropil. Although a subset of axons and dendrites cross the midline through commissures, the remaining majority stays within the neuropil. Axons from the peripheral sensory neurons terminate here as well, without crossing the midline. The positioning of these CNS neurites and PNS terminals is apparently under the influence of Slit, a midline glia-derived signal protein. This study shows that midline signaling protein Slit is enriched extracellularly along gliopodia. Thus, Slit, which has been previously thought to reach neuropil through diffusion, could be also 'hand-delivered' (or foot-delivered?) onto the neuropilar surface by gliopodia (Vasenkova, 2005).

Slit is perhaps the easiest endogenous midline-derived signaling molecule to be examined for its redistribution in vivo through immunocytochemistry. This is because in the CNS, unlike other molecules such as Netrins that are also expressed by surrounding neurons, Slit is exclusively supplied by midline glia. In theory, however, other midline glia-derived molecules could also be carried through gliopodia. Vesicular transport of synaptic vesicle proteins has been observed along neuronal filopodia. Similarly, vesicles containing midline-signaling molecules could travel along gliopodia to neuropils where simple diffusion of the cue might not be effective or sufficiently discrete to mediate cell-specific interactions. The data suggest that gliopodia could serve as a vehicle for cue delivery to axons and dendrites that develop within commissures and neuropil of the embryonic CNS (Vasenkova, 2005).

The development of nervous systems involves reciprocal interactions between neurons and glia. In the Drosophila olfactory system, peripheral glial cells arise from sensory lineages specified by the basic helix-loop-helix transcription factor, Atonal. These glia wrap around the developing olfactory axons early during development and pattern the three distinct fascicles as they exit the antenna. In the moth Manduca sexta, an additional set of central glia migrate to the base of the antennal nerve where axons sort to their glomerular targets. In this work, whether similar types of cells exist in the Drosophila antenna has been investigated. Different P(Gal4) lines were used to drive Green Fluorescent Protein (GFP) in distinct populations of cells within the Drosophila antenna. Mz317::GFP, a marker for cell body and perineural glia, labels the majority of peripheral glia. An additional ~30 glial cells detected by GH146::GFP do not derive from any of the sensory lineages and appear to migrate into the antenna from the brain. Their appearance in the third antennal segment is regulated by normal function of the Epidermal Growth Factor receptor and small GTPases. These distinct populations of cells have been denoted as Mz317-glia and GH146-glia respectively. In the adult, processes of GH146-glial cells ensheath the olfactory receptor neurons directly, while those of the Mz317-glia form a peripheral layer. Ablation of GH146-glia does not result in any significant effects on the patterning of the olfactory receptor axons. This study has demonstrated the presence of at least two distinct populations of glial cells within the Drosophila antenna. GH146-glial cells originate in the brain and migrate to the antenna along the newly formed olfactory axons. The number of cells populating the third segment of the antenna is regulated by signaling through the Epidermal Growth Factor receptor. These glia share several features of the sorting zone cells described in Manduca (Sen, 2005).

Studies in Manduca, have described three types of glia associated with the developing olfactory neuropil and nerve: (1) peripheral glia that arises from the antenna and ensheath olfactory axons; (2) lobe neuropil glia that surround and stabilize olfactory glomeruli; (3) sorting zone glia which are central in origin and migrate to the base of the antennal nerve where they serve to segregate axons that target independent glomeruli. In Drosophila, previous work has described the direct equivalents of peripheral antennal and the neuropil glia. This paper identifies an addition set of glia marked by GH146::GFP that share common features with sorting zone glia. A model suggests that these glia arise in the brain and migrate along the ORNs which have newly connected to the brain. Cells proliferate as they migrate along the nerve and enter the third segment of the antenna. Coincident with GH146-glia arrival at the periphery, the Ato-derived glia marked by Mz317::GFP (that are already present on the olfactory axons) move to wrap around the cell bodies of the sensory organs. The GH146-glial processes tightly ensheath the olfactory fascicles and form a layer below that of the Mz317-glial projections. In all 36 hrs APF preparations examined, ~30 GH146-glia were detected suggesting there must exist mechanisms to regulate the number of cells terminating in the antenna (Sen, 2005).

What is the function of GH146-glia during olfactory development? Although these cells share many of the properties of sorting zone glia described in the moth, it was not possible to implicate them in any guidance function. In Manduca, Fasciclin II (FasII), the insect ortholog of N-CAM, is expressed in a subset of ORNs scattered throughout the antennal nerve. Upon reaching the glial rich sorting zone, these projections segregate from the non-expressing axons and terminate in distinct glomeruli. In Drosophila, the presence of FasII protein on the ORNs during axon projection could not be detected using the available antibodies. When the GH146-glia were completely absent within the third antennal segment, no significant abnormalities were detected in ORN fasciculation within the antenna. The processes of the GH146-glia extend into the antennal nerve to ensheath axon bundles. This segregation is not based on Or gene type and its functional significance is obscure. A similar migration of glial cells from the centre into the periphery has been described during development of the Drosophila compound eye (Sen, 2005).

What are the factors that induce migration of glia from the brain towards the periphery? Cell migration in insect glia is well known to be mediated by signaling through the Fibroblast Growth Factor Receptors (FGFRs). Preliminary observations appear to rule out a requirement for FGFR signaling during GH146-glial migration. The downstream effector of FGF (Dof), is not expressed in these cells at any time during their development. Further, expression of dominant negative forms of Drosophila FGF receptors does not affect glial cell number in the antenna (Sen, 2005).

It is suggested that GH146-glial migration is mediated by EGF signaling. The ligand for Egfr, Vein, has been shown to be known to be expressed in the antennal epidermis at a time consistent with the arrival of cells from the center. The mechanisms involved in triggering this homing of cells into the antenna require extensive investigation (Sen, 2005).